Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Spherically symmetric model atmospheres for late-type giant stars Bennett, Philip Desmond 1991

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-UBC_1991_A1 B45.pdf [ 24.48MB ]
Metadata
JSON: 831-1.0085580.json
JSON-LD: 831-1.0085580-ld.json
RDF/XML (Pretty): 831-1.0085580-rdf.xml
RDF/JSON: 831-1.0085580-rdf.json
Turtle: 831-1.0085580-turtle.txt
N-Triples: 831-1.0085580-rdf-ntriples.txt
Original Record: 831-1.0085580-source.json
Full Text
831-1.0085580-fulltext.txt
Citation
831-1.0085580.ris

Full Text

SPHERICALLY SYMMETRIC MODEL ATMOSPHERES FOR LATE-TYPE GIANT STARS By PHILIP DESMOND BENNETT B.Sc. (Mathematics) Simon Fraser University M.Sc. (Astronomy) The University of British Columbia  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY  in THE FACULTY OF GRADUATE STUDIES DEPARTMENT OF GEOPHYSICS AND ASTRONOMY  We accept this thesis as conforming to the required standard  THE UNIVERSITY OF BRITISH COLUMBIA  June 1991 © PHILIP DESMOND BENNETT, 1991  In presenting this thesis in partial fulfilment  of the requirements for an advanced  degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department  or by his or her representatives.  It  is understood that  copying or  publication of this thesis for financial gain shall not be allowed without my written permission.  Department The University of British Columbia Vancouver, Canada  Date  DE-6 (2/88)  n  o*r<e&^  mi.  Abstract  The ATHENA computer code has been developed to model the extended atmospheres of late-type giant and supergiant stars. The atmospheres are assumed to be static, spherically symmetric and in radiative and hydrostatic equilibrium. Molecular line blanketing (for now) is handled using the simplifying assumption of mean opacities. The complete linearization method of Auer and Mihalas [7], adapted to spherical geometry, is used to solve the model system. The radiative transfer is solved by using variable Eddington factors to close the system of moment transfer equations, and the entire system of transfer equations plus constraints is solved efficiently by arrangement into the Rybicki [83] block matrix form. The variable Eddington factors are calculated from the full angledependent formal solution of the radiative transfer problem using the impact parameter method of Hummer, Kunasz and Kunasz [47]. We were guided by the work of Mihalas and Hummer [72] in their development of extended models of 0 stars, but our method differs in the choice of the independent variable. The radius depth scale used by Mihalas and Hummer was found to fail because of the strongly temperature-dependent opacities of late-type atmospheres. Instead, we were able to achieve an exact linearization of the radius. This permitted the use of the numerically well-behaved column mass or optical depth scales. The resulting formulation is analogous to the plane-parallel complete linearization method and reduces to this method in the compact atmosphere limit. Models of M giants were calculated for T  fJJ  = 3000K and 3500K with opacities of  the CN, TiO and H 0 molecules included, and the results were in general agreement 2  with other published spherical models. These models were calculated assuming radiative equilibrium. The importance of convective energy transport was estimated by calculating the convective flux that would result from the temperature structure of the models. The ii  standard local mixing length theory was used for this purpose. Convection was found to be important only at depths with r  RM  but significant out to T  RM  > 15 for the low gravity models with log g = 0,  ~ 1 at the most transparent frequencies for the higher gravity  models with log g — 2. Thus, the temperature structure of the surface layers and the emergent flux for the log g — 0 models should be accurately modelled but the emergent flux for the logg = 2 models may be in error by up to 5% at the most transparent frequencies.  The problem of fitting a series of beautiful internally consistent models to honest-to-goodness real stars that are up there, is horrible  A. B. Underhill  iii  Table of Contents  Abstract  ii  List of Tables  viii  List of Figures  ix  Acknowledgement  xi  1  2  The Stellar Atmosphere Problem  1  1.1  Introduction  1  1.2  Classical Model Assumptions  2  1.3  Red Giant Models .  4  1.4  The Physics of Late-Type Atmospheres  6  1.5  Plane-Parallel Models . .  8  1.6  Spherically Symmetric Static Models  14  1.7  Objectives of this Research  23  The Spherically Symmetric Model System  25  2.1  Introduction  25  2.2  The Model Equations  25  2.3  The Formal Solution .  33  2.4  The Boundary Conditions  38  2.4.1  The Formal Solution  38  2.4.2  The Moment Equations of Transfer  48  2.4.3  Other Boundary Conditions  . .  iv  . .. . . . . 0  53  3 The Discrete System and Solution Procedure 3.1  56  The Finite Difference System  56  3.1.1  The Formal Solution: Cubic Spline Difference Formulae  56  3.1.2  Quadrature Weights for the Moment Integrals  3.1.3  Moment Equations: Cubic Spline Difference Formulae  68  3.1.4  The Hydrostatic Equilibrium and Radius Equations  71  3.1.5  The Equation of Radiative Equilibrium  76  .  61  3.2  The Linearization of the Radius  78  3.3  The Linearized Model Equations  83  3.4  The Initial Solution  91  3.4.1  Introduction  91  3.4.2  The boundary temperature  92  3.4.3  The boundary pressure  93  3.4.4  The boundary radius  94  4 The Model Computations  95  4.1  Introduction  95  4.2  The Equation of State  95  4.3  The Opacity of the Stellar Gas  105  4.4  Convection  107  4.5  Convergence Criteria and Iterative Control  Ill  4.6  The Production of Converged Models  115  4.7  Model Results and Discussion  120  5 Conclusions 5.1  169  Discussion  169  5.1.1  169  The Complete Linearization Method: An Overview . .  v  5.1.2  The Effects of Convection  170  5.1.3  The Treatment of the H 0 Opacity  171  5.1.4  The Independent Variable Dilemma  171  5.1.5  Numerical Inconsistencies  173  5.2  Comparison with Other Models  175  5.3  Future Research  178  5.3.1  Short Term Improvements  178  5.3.2  Non-LTE Models  181  5.3.3  Self-Consistent Atmosphere and Wind Models  182  5.3.4  Time-Dependent Dynamical Models  183  2  References  192  Appendices  199  A The Linearized Model Equations  199  A . l The Model Equations  199.  A. 1.1 Transfer Equation .  199  A.1.2 Transfer Equation: Inner Boundary Condition  200  A.1.3 Transfer Equation: Outer Boundary Condition  200  A.1.4 Radiative Equilibrium . .  200  A.1.5 Hydrostatic Equilibrium  201  A.1.6 Hydrostatic Equilibrium: Outer Boundary Condition  201  A.1.7 Radius or Depth Equation  201  A.1.8 Radius Equation: Outer Boundary Condition .  201  A.2 The Linearized Equations  202  A.2.1 Transfer Equation  202  A.2.2 Transfer Equation: Outer Boundary Condition  206  vi  A.2.3 Transfer Equation: Inner Boundary Condition  209  A.2.4 Radiative Equilibrium  211  A.2.5 Hydrostatic Equilibrium  212  A.2.6 Hydrostatic Equilibrium: Outer Boundary Condition  214  A.2.7 Radius Equation  214  .  A. 2.8 Radius Equation: Outer Boundary Condition  216  B Model Output Tables B.l  217  Glossary of Variables  . .  218  B.2 Converged Models: Output Tables  219  B. 2.1 Model 01310191: SS 3000/0.0/-3.0, CN, TiO, H 0(sm)  220  B.2.2 Model 02310191: SS 3000/2.0/-3.0, CN, TiO, H 0(sm) . . . . . .  223  B.2.3 Model 03310191: SS 3000/0.0/-3.0, CN, TiO, H 0(hm)  226  B.2.4 Model 04310191: SS 3000/1.0/-3.0, CN, TiO, H 0(hm)  229  B.2.5 Model 05310191: SS 3000/2.0/-3.0, CN, TiO, H 0(hm)  232  B.2.6 Model 06310191: SS 3500/1.0/-3.0, CN, TiO, H 0(hm)  235  B.2.7 Model 07310191: SS 3500/2.0/-3.0, CN, TiO, H 0(hm)  238  2  2  2  2  2  2  2  vii  List of Tables  4.1  Species Considered in Model Equation of State  98  4.2  Model Opacities  4.3  Command Timings on the IBM 3081  122  4.4  Converged Models with Molecular Opacities .  123  4.5  Converged Models: Summary of Parameters . . .  124  . .  viii  106  List of Figures  2.1 Impact parameter coordinate system for formal solution  .  35  2.2  Domains of calculation of J„ and H : Model 01310191  39  2.3  Domains of calculation of J and E \ Model 02310191  4.1  Temperature profiles T(r )  135  4.2  Temperature profiles T ( r ) of the outer layers  136  4.3  Pressure structure log J ^ T ^ )  137  4.4  Electron pressure log ^ ( T R M )  138  4.5  Atmospheric extension r(r )/.R* — 1 . . ;  4.6  Rosseland mean opacity profiles log XRM{ RM)  4.7  Ratio offlux-meanopacity to Rosseland mean opacity log XH/XRM  4.8  Ratio of convective to total flux log H {r ) / H  4.9  Profiles of mean convective velocities log v  v  v  v  . . . . . . . . . .  RM  RM  139  RM  140  T  cmv  • • • • 141 •'. 142  RM  143  (T )  CONV  RM  4.10 Profiles of cooling functions log $ j(r ) coo  40  144  RM  4.11 Normalized radiative equilibrium residuals e  .  RB  145  4.12 Normalized hydrostatic equilibrium residuals e  .  HB  4.13 Normalized radius (depth) equation residuals e  146  DE  147  TBrm  148  4.14 RMS normalized transfer equation residuals e  4.15 Ratio of actual to prescribed model luminosity L/L  149  4.16 Emergent flux distributions log H  150  t  u  4.17 Ratio of photospheric to model radii  Rp,(v)/R*  4.18 Ratio of the mean intensity to the Planck function log «/„/#„ 4.19 Ratio of the monochromatic to the Rosseland mean opacity log XV/XRM  ix  151 152 • 153  4.20 The Eddington factor f (r )  154  4.21 The distribution of opacity sources in model 01310191  155  4.22 The distribution of opacity sources in model 02310191 .  156  4.23 The ionization structure of model 01310191  157  4.24 The ionization structure of model 02310191  158  4.25 The distribution of hydrogen in model 01310191  159  . 4.26 The distribution of hydrogen in model 02310191  160  4.27 The distribution of carbon in model 01310191  161  v  RM  4.28 The distribution of carbon in model 021310191  . 162  4.29 The distribution of oxygen in model 01310191  163  4.30 The distribution of oxygen in model 02310191  164  4.31 The distribution of nitrogen in model 01310191  165  4.32 The distribution of nitrogen in model 02310191  166  4.33 The distribution of silicon in model 01310191  167  4.34 The distribution of silicon in model 02310191  168  5.1  Temperature profiles of the surface layers: 3000/0.0 models.  186  5.2  Temperature profiles: 3000/0.0 models  187  5.3  Pressure profiles: 3000/0.0 models  188  5.4  Temperature profiles of the surface layers: 3500/2.0 models.  5.5  Temperature profiles: 3500/2^0 models  190  5.6  Pressure profiles: 3500/2.0 models  191  x  . . . . . . .  189  Acknowledgement  I would like to express my gratitude to my supervisor, Dr. Jason Auman, for introducing me to the study of stellar atmospheres, for his encouragement in undertaking the project described in this thesis, and for his continued support over the years in bringing this work to completion. I am also grateful to Al Irwin, Harvey Richer, Stephenson Yang, and Jeff Brown for advice and suggestions which proved helpful along the way. But I am most indebted to my partner and companion, Lynne Robinson, for her emotional support and sustenance over these past few years, without which this work would likely not have come to fruition. Finally, I wish to dedicate this work to my youngest daughter, Caitlin Agawa, and hope that she too may feel the sense of wonder about the universe that has inspired me over the years.  xi  Chapter 1  The Stellar Atmosphere Problem  1.1  Introduction  Much of stellar astronomy depends upon the observation and interpretation of the electromagnetic radiation radiated by stars. The region near the surface of the star from which photons can escape directly to space without undergoing further absorption or scattering is referred to as the atmosphere.  It represents the boundary between the  relatively simple physics of the stellar interior, where strict thermodynamic equilibrium accurately describes the state of the material, and the surrounding vacuum of space. The atmosphere thus represents a region of large gradients in the physical variables such as the temperature, gas pressure and the radiation field. It is also where the stellar material becomes optically thin with the result that the radiation field is no longer strongly coupled to the local environment. As such, stellar atmospheres are comparatively difficult to model. A meaningful interpretation of stellar spectra, however, is only possible with the assistance of consistent models. Consequently, much effort has been devoted to the construction of such models over the past fifty years. This chapter will review the standard assumptions encountered in modelling stellar atmospheres and, in particular, consider the reasons why these may be of limited applicability for red giant models. Also, the extensions of classical theory that are needed to realistically model the atmospheres of these stars are discussed.  1  2  Chapter 1. TJie Stellar Atmosphere Problem 1.2  Classical Model Assumptions  Traditionally, the stellar atmosphere problem has been to solve for the run of the physical variables through the outer boundary layers down to depths much larger than the photon mean-free-path. Simplifying assumptions were introduced to render the problem amenable to analytic approaches. Typically, these included the following. 1. Simplified opacities. Early analytic treatments often made the assumption that the opacity \ stellar material is independent of frequency, or \v = X>  a  n  v  of the  approximation referred  to as the gray atmosphere. 2. Plane-parallel geometry. The boundary layers of a star are approximated by a semi-infinite slab with planar stratification. 3. Static atmospheres. The atmosphere is assumed to be in hydrostatic equilibrium. Usually it is also assumed that the total pressure is simply the sum of the gas and radiation pressure. 4. Local thermodynamic equilibrium (LTE). The radiation field, which is in strict thermodynamic equilibrium (TE) at depth in the star, must become increasingly^ anisotropic toward the upper boundary of the atmosphere, since there the intensity over the inward directed hemisphere falls to zero. This approximation assumes that the gas can be treated as though it were still in T E , with well-defined (and equal) kinetic, excitation and ionization temperatures. This implies that the gas particle velocities are specified by the Maxwellian distribution and that their level populations and degree of ionization are given by the Saha-Boltzmann distribution. The radiation field, however, is  Chapter 1. The Stellar Atmosphere Problem  3  permitted to deviate from its equilibrium Planck distribution (since otherwise strict T E would still hold). Under the assumption of static LTE models, the variables describing the physical state of the atmosphere are the temperature T and the pressure p of the gas, and the specific intensity I of the radiation field. In general, for a system lacking any particular v  symmetry, T and p will depend upon position (r, 0,$) and the intensity I additionally u  upon the propagation direction described by the angles (0, <f)). In this case, the full transfer problem is five-dimensional. The assumption of either a plane-parallel or spherically symmetric geometry reduces the problem to a two-dimensional system. One coordinate is usually described by the optical depth r although this is often replaced by the radial distance r in the spherically symmetric case. The intensity retains an additional dependence upon the direction of propagation, described either by the angle 9 to the normal direction or by p = cos 9. These variables are coupled through the equation of radiative transfer, the equation of radiative equilibrium (which expresses conservation of energy in the radiation field), and the equation of hydrostatic equilibrium. These equations form a closed system and can be solved by a variety of numerical methods to yield a self-consistent model atmosphere. In many cases, detailed line transfer calculations are unnecessary when modelling atmospheric structure. The source function can usually be adequately represented as a function of just the mean intensity J , the zeroth moment of the radiation field. The v  use of the moment forms of the equation of transfer further reduces the system to a onedimensional problem since the remaining angle coordinate has been eliminated. From a computational viewpoint, though, quadratures over frequency remain and effectively constitute another dimension. Thus, the full angle dependent model problem is really three-dimensional, and the moment model, two-dimensional, in terms of timings of solutions of finite difference representations and the sizes of computationally feasible grids.  Chapter 1. The Stellar Atmosphere Problem  4  Adopting a spherically symmetric geometry, instead of plane-parallel, results in a similar system of model equations in the physical variables T, p, and I . The radiative v  transfer equation now contains an additional term due to the increased complexity of the representation of the divergence in spherical coordinates, but a transformation can be made to remove the offending term. Another major difference is that the spatial coordinate r (radius) enters into the transfer equation directly so that the optical depth and radius must both be iterated consistently. While this presents considerable complications and has hampered development of spherical models in the past, we will show that the radius r can be included in the consistent solution of the model equations with little increase in computational effort. With the development of both the opacity distribution function (ODF) and the opacity sampling (OS) techniques to handle line-blanketing opacities a decade ago, it then became possible to compute fully consistent static, plane-parallel, LTE model atmospheres with realistic opacities. Although molecular line lists are incomplete for some important species (e.g. H2O, HCN, and C2H2), grids of such models have now been calculated for K giants [42], for M giants [56] and for C stars [54]. Existing techniques for including realistic opacities using the ODF method can readily be incorporated into spherical models also. 1.3  Red Giant Models  Red giants are difficult to model since a variety of the classical assumptions break down. Some of the particular difficulties encountered arise from the following situations. 1.. The presence of molecules. Stars with T,  ff  < 4500K are sufficiently cool that molecular association becomes  appreciable. The formation of molecules complicates both the equation of state and the radiative transfer in the atmosphere.  Simple diatomic and polyatomic  Chapter 1. The Stellar Atmosphere Problem  5  molecules can give rise to an absorption spectrum consisting of hundreds of thousands to millions of lines, which may overlap and blend into a pseudo-continuous opacity. These atmospheres can not be accurately represented by models using gray opacities, or even by mean opacities. The ODF and OS techniques do allow models to be constructed, at least for static atmospheres, but they remain computationally demanding. 2. Atmospheric extension.  The atmospheres of red giant stars are weakly bound due to the low gravity and thus have a large scale height. The extremely nongray opacities characteristic of these stars aggravate this effect since the radius at which the atmosphere becomes optically thin is now wavelength dependent. Both effects combine to increase the effective thickness of the atmosphere. The wavelength dependence of the photospheric radius expected of extended atmospheres appears confirmed by the optical interferometry available [14]. When the atmosphere thickness is no longer small compared to the photospheric radius of the star, the curvature of the atmosphere becomes important in the treatment of the radiative transfer. At this point planeparallel models break down, and more realistic geometries such as spherical symmetry must be used. 3. Mass loss and stellar winds.  All M giants appear to be losing mass, typically on the order of 10 M©/yr, due -6  to an apparently steady wind of velocity about 10 - 20 km/s [82]. Dust grains condensing in the cooling outflowing gas both emit thermal radiation in the midinfrared (the 'infrared excess') and absorb photospheric radiation in the visible and ultraviolet. Modelling these extended stellar atmospheres necessitates giving up the assumption of static models in hydrostatic equilibrium. Instead, this constraint must be replaced by an equation describing momentum transfer to the outflowing  Chapter 1. The Stellar Atmosphere Problem  6  stellar gas. Although wind models lie beyond the scope of the present work, it is essential that fully consistent, robust, static models exist to serve as a base to further efforts in this direction. It is hoped that this work will provide a step toward that goal. 4. Variability.  Essentially all red giants are variable stars [37]. While some exhibit only small irregular variations in brightness and can likely be represented by static or steady state models, others, such as the Mira variables, undergo large amplitude, quasiperiodic variations which require fully time dependent, hydrodynamical models. It should be noted that all red giant stars with high mass loss rates appear to be Mira variables or OH/IR stars (all long period variables) [13, 106] so that pulsation seems to be a factor in the poorly understood mass loss mechanism. Variable stars lie beyond the scope of this work and will not be considered further. 1.4  The Physics of Late-Type Atmospheres  The solution of the stellar atmosphere problem requires an accurate knowledge of the equation of state. This is fairly straight forward for hotter stars, but the degree of molecular association present in the atmospheres of cool stars complicates the problem substantially. Most recent models base their equation of state on Tsuji's [97] compilation of equilibrium constants for several hundred molecules. This is an extension of an earlier work by Vardya [101] on molecular equilibria in K and M stars. More recently, Irwin [50] published a list of polynomial partition function approximations for a large number of atomic and molecular species of astrophysical interest. Sauval and Tatum [84] published a similar compilation of partition function approximations for diatomic molecules, and Irwin [51] has produced an updated list of partition functions of polyatomic molecules. The general subject of molecules in stars has been reviewed by Tsuji [100].  Chapter 1. The Stellar Atmosphere Problem  7  Bennett [17] developed a computer code to solve the equation of state problem for an arbitrary composition, using an iterative Newton-Raphson technique to solve the chemical equilibrium. This procedure is fast and is used in the ATHENA model atmosphere code described in this work. An accurate treatment of molecular bound-bound opacities is also needed. The opacity distribution function (ODF) method proposed by Strom and Kurucz [94] was the first to allow realistic molecular opacities to be included in model atmospheres. The idea is that since the structure of a stellar atmosphere depends on integrals of the radiation field over frequency, it may be expected that the detailed small scale frequency structure of the molecular band opacities, consisting of millions of lines arranged in a sequence of bands, is unimportant. It should then be possible to rearrange the actual opacity structure over a fairly large frequency interval (but small compared to variations in B {T)) to achieve V  a smoothly varying opacity over this interval. This results in an enormous reduction in the number of frequencies needed to accurately specify the opacity over the interval. Carbon [26] constructed model ODFs for the CN red system, and compared fluxes obtained with straight and harmonic mean models [27]. Kurucz [65] also provided a concise summary of the ODF technique. An alternative approach, developed and applied to atomic line blanketing by Peytremann [79], is the opacity sampling (OS) method. This relies on using a Monte Carlo technique to sample the opacities at randomly selected frequencies and has been described by Johnson and Krupp [57]. Sneden et al. [92] applied the OS method to molecular line blanketing by CN, CO, and C in carbon star models. 2  A review of the opacity modelling field was given by Carbon [28]. The treatment of molecular opacities was questioned by Gustafsson and Olander [43], who noted that computed TiO lines tend to be much stronger than the observed ones. They considered it likely that the true effects of molecular opacities upon atmospheric structure (e.g. backwarming and surface cooling or heating) were smaller than indicated  Chapter 1. The Stellar Atmosphere Problem  8  by existing models. This may indicate significant departures from LTE in the electronic transitions of certain molecules, such as TiO [46]. Similar non-LTE effects may also occur in the H 0 vibrational transitions. Gustafsson and Olander [43] claimed many molecular 2  lines may be more accurately considered to be formed in scattering, rather than absorption. A full treatment would require non-LTE modelling of the (many) molecular levels involved and, in particular, knowledge of the poorly known cross-sections for collisionally excited electronic transitions. Since detailed monochromatic opacities for H2O were not available, Alexander et al. [3] prepared a statistical representation of the water vapour spectrum suitable for use in opacity sampled (OS) models. Model atmospheres were calculated for T  eff  = 3000K and  3600K using the ATLAS6 model atmosphere code, as modified at Indiana University. An analysis of the differential effect of the treatment of the H 0 opacity (straight mean or 2  OS synthetic spectrum) for these M giant models was presented. J0rgensen [59] has calculated ODFs and line lists for HCN combination bands using newly available laboratory data. The polyatomic molecule HCN is abundant in the outer atmospheres of cool carbon stars, and represents a significant source of opacity. The focus of this study was to determine if HCN has potentially observable spectral features in the visible and near infrared (OAfi-lfi) region. The intent was to find more visible or near IR tracers of the outer atmospheric layers. However, it was clearly shown that HCN can not serve as such a tracer since no bands shortward of 1.5/i should be visible. 1.5  Plane-Parallel Models  An extensive literature exists on modelling LTE plane-parallel atmospheres. Much of the early analytic work along, with the first numerical solutions, was reviewed in the comprehensive summary by Pecker [76]. The technique that remains conceptually the simplest,  Chapter 1. The Stellar Atmosphere Problem  9  computationally the fastest, and most amenable to generalization is the complete linearization method of Auer and Mihalas [7], as adapted to the second order Feautrier [38] form of the equation of radiative transfer. The use of variable Eddington factors as described by Auer and Mihalas [8] permits an exact closure of the moment equations of transfer and so achieves the elimination of the angle coordinate from the transfer problem. Rybicki [83] showed that the many frequency LTE problem, linearized ultimately in terms of the fundamental variables T and n, can be rearranged into a tridiagonal block matrix that requires much less computational effort to solve as compared to the conventional matrix solution. The overall technique is summarized by Auer [5] in a paper which presents a very good review of the subject. More recent reviews of the technique (for both plane-parallel and spherical models) were presented by Auer [6] and Mihalas [71]. The literature, particularly with regard to modelling early-type stars, is very extensive. The review considered here is largely restricted to studies in the area of late-type stars. Plane-parallel model codes of late-type stars are now quite well developed, and several extensive model grids have been published in the last decade. None of the models published to date have used the complete linearization method as detailed by Auer and Mihalas [7], although Nordlund's [75] technique, used to compute the models of the Uppsala group, employs a partial linearization. Most are adaptations of earlier, less robust methods that have been coerced into the difficult task of converging models of red giants. Most recent models have used either the ODF or OS method to handle line blanketing opacities. All of the models described in this section assume molecular lines are formed in pure LTE absorption. Auman [10] calculated a grid of models for late-type giants and dwarfs over a range of 2000K< T < 4000K. The abundances used were approximately solar. The opacities cff  included absorption by H, H~,  , H e , H j and the metals, and scattering by H, H and -  2  Chapter 1. The Stellar Atmosphere Problem  10  free electrons. The molecular vibration-rotation lines of H2O, important in the later M stars, were included as a harmonic mean opacity. The model atmosphere code used the Avrett-Krook temperature correction method [12] iterated to constant flux. Convection was included in the model and treated using the mixing length theory of Bohm-Vitense [20]. Querci et al. [81] calculated model atmospheres for carbon stars. Opacity sources included absorption and scattering contributions by H, H ~ , H2, Hj , H e , C , C, e~ -  -  and metals as well as pressure induced absorption due to H - H and H -He collisions. 2  2  2  Molecular line blanketing by CN, C2 and CO was treated using ODFs to represent the opacities. The temperature-pressure structures of these C star models were very different from those of the M star models of Auman [10]. An expanded grid extending to cooler carbon stars was subsequently produced by Querci and Querci [80] using the same model code. Gustafsson et al. [42] calculated a grid of models for G and K Population II giants, the details of which have been published by Bell et al. [15]. Abundances ranged from solar to [Fe/H] = —3. Bound-free and free-free continuum absorption from H ~ , H , Hj , H , and +  C, Mg, A l , Si, and Rayleigh scattering by neutral H, along with electron scattering, were included. Approximately 50000 atomic lines were considered, along with molecular lines from CH, OH, NH, CN, CO and MgH. The line blanketing was handled using ODFs, and the radiative transfer solved by linearization using, the method of Nordlund [75]. This linearization was only partial in that the implicit dependence of the opacity upon temperature and density was not included in either of the radiative transfer or radiative equilibrium equations. As a result, the convergence of these models was poor, and it was found necessary to adopt a modified Aitken extrapolation to obtain solutions. Auman and Woodrow [11] calculated two sequences of model atmospheres for latetype giants. In one set of models the ionization equlibria were calculated assuming LTE.  Chapter 1. The Stellar Atmosphere Problem  11  The other set were calculated using the statistical equhbrium (SE) equations to solve for the level populations and ionization equilibria of the metals that were significant sources of free electrons. The approximation that the bound-bound transitions of these metals were in detailed radiative balance was made. The opacities used were similar to those used by Auman [10] with the addition of the bound-free opacities of K, Na, Ca, A l , Mg, and Si. The H2O line opacity was also included using the harmonic mean approximation. They found a significant shift in the ionization equilibrium in the surface layers of the coolest (T = 2000K and 2500K) models, resulting in a increase in the electron pressure eff  of 1-2 orders of magnitude. Kurucz [65] published an extensive grid of models of early type stars, extending in spectral type through the G stars (T = 5500K). Solar abundances were adopted, and a eft  modified version of Kurucz's ATLAS5 code [64] was used. The usual continuum opacities for hot stars, along with 10 atomic lines from the list of Kurucz and Peytremann [66], 6  but with no molecules, were included. Kurucz [65] presents a good overview of the ODF method used to handle the atomic line blanketing in these models. Tsuji [99] calculated the first models of M giants to include a reasonably accurate treatment of molecular opacities. Models both with solar abundances and with C/O = 1.05 were considered. The continuum sources included bound-free and free-free absorption by H , H , Si, Mg and Ca, free-free absorption by Hg and H e , Rayleigh scattering -  -  by H, H arid He, and electron scattering. Molecular opacities of CO, OH, CN, CH, TiO, 2  MgH, SiH, CaH, H 0 and H (collisionally induced) were handled by the Voigt-analog2  2  Elsasser band model (VAEBM) which, although considerably less accurate than ODF or OS treatments, was an improvement upon previous work using mean opacities. Tsuji's models iterated the temperature until flux constancy was attained. Krupp et al. [62] investigated the role of TiO on the atmospheric structure of late-type giants of solar abundance. A significant warming of up to several hundred degrees was  Chapter 1. The Stellar Atmosphere Problem  12  found in the outer layers when the TiO opacity was included. The models were adapted from the ATLAS5 code of Kurucz [64], modified to include atomic and molecular line blanketing using the OS method. Johnson et al. [56] also produced a grid of M giant models. The code used was also based on Kurucz's [64] ATLAS5 model atmosphere program. Solar abundances were assumed, and bound-free, free-free and scattering opacities of H, H~, Hj , H , He, 2  H e , C, Mg, A l and Si were included as described by Kurucz [64]. A list of atomic lines -  provided by Bell (from the list of Bell and Rodgers [16]) were included, along with the molecular lines of CN, CO, C , TiO, CH, NH, OH and MgH, handled using the OS 2  method. Also, H 0 was represented by means of straight mean opacities. A subsequent 2  series of models, based on the above code and with similar opacities were subsequently calculated by Johnson [54] for carbon enriched stars. More recently, Saxner and Gustafsson [85] devised a technique for forming composite ODFs (necessary when more than one opacity source is important in the same spectral region) from single species ODFs. This removed one of the previous objections to the ODF method, that it was necessary to calculate new sets of distribution functions when the composition of the atmosphere was changed. Instead, it is now feasible to calculate ODFs for individual opacity sources and then merge these to construct composite ODFs as needed. Implicit in this is the assumption that the individual opacities are uncorrelated over the frequency interval used to construct the ODFs. In addition, a significant penalty in computing time is required, although much less than needed to recompute a full set of ODFs. Eriksson et al. [36] calculated cool carbon star models that included absorption by HCN. The atmosphere code used was a modified version of the program described by Gustafsson et al. [42], with pressure induced opacities from collisions between H - H 2  2  and H -He added and revisions to the atomic and molecular opacity data. The HCN 2  Chapter 1. The Stellar Atmosphere Problem  13  band strengths were essentially unknown and were modelled by analogy to the more well studied molecules CO2 and N2O. The line blanketing obtained from this was handled using ODFs. The effect of inclusion of this HCN opacity in the models was crucial to the atmospheric structure. Models of cool carbon stars (T < 2900K) without HCN are eff  compact, whereas those with HCN are more extended by up to two orders of magnitude (as indicated by the gas pressure at a given temperature). It is quite possible that the inclusion of a realistic H2O opacity would have the same effect in cool M giant stars. A comparison of the structure of model atmospheres obtained using the ODF and OS methods was studied by Ekberg et al. [35]. Significant temperature differences of up to 150K were found for cool carbon star models. This appears to be due to the lack of correlation in the fine structure of the opacity over the range of the atmosphere, in violation of one of the basic assumptions of the ODF method. The problem is worst when the surface layers are dominated by polyatomics, since diatomic absorbers will always dominate deeper in the atmosphere. The status of red giant models was reviewed by Johnson [55] with particular emphasis on the opacity problem. The consistency of current models with other models and observational constraints was discussed. Brett [21] calculated opacities for the VO A-X, B-X, and TiO e band systems using the Just Overlapping Line Approximation (JOLA) technique [96]. No laboratory analyses exist for any of these band systems, which are potentially important opacity sources for the cooler M stars. The plane-parallel code of Wehrse [103] was used to compute a series of models for 2250K< T < 3500K, both with and without these new opacities. eff  Brown et al. [23] computed plane-parallel model atmospheres over the range 3000K< T  eff  < 4000K, and 0 < log<7 < 2, which included energy transport by both radiation  and convection. These models were calculated using the Indiana ATLAS6 code. The opacities included the standard bound-free and free-free continuous opacities of H, He  14  Chapter 1. The Stellar Atmosphere Problem  and various metals, and included the molecules CO, H 0 , TiO, SiO, CN, C , CH, NH, 2  2  OH, MgH, and SH as well as atomic lines from the list of Kurucz and Peytremann [66]. The molecular and atomic lines were treated using the opacity sampling method. Water vapour was handled using the statistical spectrum generated by Alexander et al. [3]. 1.6  Spherically Symmetric Static Models  When the photon mean free path in the atmosphere becomes a significant fraction of the stellar radius, the curvature of the atmospheric layers becomes important and planeparallel models become inadequate. One must then resort to spherically symmetric models. The complexity of the transfer equations in spherical coordinates made strictly analytic treatments of the problem impossible except under very special conditions, and so no solutions of a general nature were obtained until after the advent of electronic computers. Kosirev [61], assuming a gray, Kramers opacity of the form np = r , was able to find _ n  an approximate temperature gradient and, further assuming a Planckian source function, was able to integrate the formal integral of the transfer equation. Chandrasekhar [31] obtained the formal integral of the transfer equation for an infinite, spherical atmosphere assuming the Eddington approximation and demonstrated the characteristic flattening of the continuum that results from extended, as compared with planar, atmospheres. Chapman [32] examined the formal solution of the spherical transfer equation assuming a Kramers opacity law with n = 3. He demonstrated that the Eddington approximation assumed in almost all of the early work on spherical transfer was inaccurate (and totally invalid at small optical depths) due to the strong peaking of the radiation field in the outward radial direction. The modern era of spherical models was ushered in by Cassinelli's [30] numerical models of the central stars of planetary nebula. Bound-free and free-free opacities of  Chapter 1. The Stellar Atmosphere Problem  15  H, He, H e , ionized metals and scattering by electrons were included. The radiative +  transfer problem was attacked by discretizing the two dimensional transfer equation in both radius r and angular variable p using the discrete Sjv method of Carlson [29]. The method is stable only for step sizes of small optical depth. Large variations of the opacity with frequency then force the adoption of a grid spacing set by the most opaque frequency. This was avoided by smoothly fitting diffusion approximation results to the discrete ordinate calculation at large optical depths. The models were converged to radiative equilibrium using Lucy's [67] temperature correction procedure. These models yielded the first accurate continuum fluxes for extended stellar atmospheres. Hummer and Rybicki [48] derived approximate solutions of the spherically symmetric transfer problem with power law opacities and introduced the impact parameter method to obtain an accurate numerical solution of the formal transfer problem with arbitrary opacities. The impact parameter method is described in detail, with computer coding and benchmark tests, by Hummer, Kunasz and Kunasz [47]. Auer's [5] introduction of the sphericality factor transform allowed the reduction of the coupled system of moment equations, closed by the use of the variable Eddington factor method of Auer and Mihalas [8] adapted to spherical geometry, to a single second order differential equation of the Feautrier form involving only the mean intensity J . v  This is then exactly analogous  to the plane-parallel problem and can be solved by the Rybicki [83] formulation of the complete linearization method of Auer and Mihalas [7]. The method just described is presented in considerable detail by Mihalas and Hummer [72] and forms the basis of this current work. In their paper, Mihalas and Hummer combined this solution of the radiative transfer problem with a simultaneous solution of the equations of radiative and hydrostatic equilibrium to derive the first self-consistent approach to the solution of the spherical atmosphere problem. The opacities considered were those appropriate for early-type stars. Models (LTE and non-LTE) were calculated for a star with a spectral  Chapter 1. The Stellar Atmosphere Problem  16  type near 06. Schmid-Burgk [86] solved the spherical transfer problem, under the assumption that the opacities are arbitrary but known functions of radius, by expressing the source function in terms of a cubic spline (with undetermined coefficients) in the optical depth variable. An impact parameter scheme was then considered with the inward and outward directed intensities (I~ and I respectively) expressed by means of the formal solution of +  the transfer equation along each impact parameter ray. The intensities I  +  and I~ were  then written in terms of the unknown spline coefficients. Quadrature over each spherical shell then permitted the moments J and H to also be expressed in terms of these spline coefficients. This allowed the definition of the discrete integral operators A and analogous to the standard plane-parallel operators that yield J and H respectively when operating on the source function S. The constancy of the total flux was used to close this matrix system, which was then iteratively solved for the temperature profile. Another study that considered the solution of the full spherical model atmosphere problem, as opposed to just that of the equation of transfer alone, was done by Hundt et al. [49]. The Schmid-Burgk method was adapted to solve the spherical radiative transfer, and this solution iterated until the equation of hydrostatic equilibrium integrated consistently. Continuous opacities were computed following the methods of the ATLAS code [63] and included contributions from H, H , H , He, H e , H e , A l , Si, S i , the +  _  +  -  +  metals C, N , O, Ne up to the triply ionized species, Rayleigh scattering by H , He and electron scattering. Atomic line blanketing was not included, but molecular absorption by H2O and CN was approximated using mean opacities as suggested by Tsuji [95] and Johnson et al. [58]. Solar abundances were assumed. Apparently no actual cool star models were converged at the time this paper was written since the authors describe only test gray models in their summary. A subsequent paper by Watanabe and Kodaira [102] adapted the method of Hundt  Chapter 1. The Stellar Atmosphere Problem  17  et al. [49] for solving the spherical transfer problem to converge realistic nongray atmospheres of late-type stars. The intent of this study was to demonstrate the differential effects of atmospheric extension through comparison of spherical models with their planeparallel counterparts. The iterative procedure used to converge to constant flux was found to oscillate, and it was necessary to impose a damping term to actually converge models. Even then, the flux at the innermost layers (r  RM  > 5) did not always converge, apparently  due to the strong temperature dependence of the opacities. In these cases, the converged T-logT  RM  relation was extrapolated inward. Bound-free and free-free opacities of H and  H~, free-free opacities of H and He", Rayleigh scattering by H and H , and Thomson 2  2  scattering by electrons were all included in the absorption coefficient. Other continuous metal opacities included were adopted from the ATLAS code of Kurucz [63]. No atomic lines were included. The molecular line opacities of H 0 , CO, CN, OH and TiO were 2  included. The H 0 opacity used was the mean opacity given by Auman [9], the CO 2  opacity was from Tsuji's [95] JOLA opacity of the fundamental and first overtone of the vibration-rotation bands, the CN opacity was from Johnson et al. [58], and the TiO and OH opacities were taken from Tsuji's [98] V A E B M band model results. Solar composition was assumed. Wehrse [104] published spherically symmetric models of red giant and supergiant stars, solving the equation of radiative transfer using the discrete space theory of Peraiah and Grant [78]. This technique discretizes the full equation of transfer on the cell [^d>^d+i] x [/^i-i/25^+1/2] and integrates this over the cell to obtain a tridiagonal block matrix system relating the inward and outward specific intensities (I~ and J ) to the +  source function for each spherical shell. One significant advantage of this method over the complete linearization approach of Mihalas and Hummer [72] is that the angle points fii are fixed for all radii rj, and so the calculation of the total flux (which involves a quadrature over //) is riot influenced by arbitrary changes in the /^-discretization with  Chapter 1. The Stellar Atmosphere Problem  18  varying radius. In effect, this method achieves a solution of the transfer problem along logarithmic spirals (curves of constant p), rather than along the impact parameter rays of the Mihalas and Hummer approach. Wehrse [104] then used a Newton-Raphson temperature correction technique to iterate to constant flux. A recent review of the discrete space theory is presented by Peraiah [77]. This method does have its drawbacks. The aspect ratio of the spherical shells must be kept small for stable convergence, and the maximum stable thickness shrinks with decreasing albedo of the atmosphere. In the limit of pure absorption (zero albedo), the method fails altogether. Finally, the formalism of this method and its associated algebraic complexity obscure the underlying physics making intuitive interpretation of the system that much more difficult. Wehrse [104] adopted mean opacities for the absorption due to the molecules OH, CH, CN, MgH, SiH, TiO, CO and H 0 in his models. The molecular bands were assumed 2  to be formed in pure absorption. While it was recognized that scattering processes may play an important role in molecular line formation, it was considered infeasible to handle this in the model calculations (which would require a non-LTE treatment of a molecular absorber). The usual sources of bound-free and free-free absorption were included, along with Rayleigh scattering by H, H and He. 2  Wehrse calculated a series of extended models (and one plane-parallel benchmark), all with T  eff  = 3000K and logg = 0 for varying He, CNO and metal abundances, and  found widely ranging amounts of geometric extension of the atmosphere. Values of the extension d, defined by ,  r(n.  2ti  = 1)  ranged from d = 0.05, which is nearly compact, to d — 0.30. Included in these models, however, were two with considerably enhanced He abundances (up to [He/H] = +2) which essentially had compact atmospheres. Omitting these He-rich models, the extensions were  Chapter 1. The Stellar Atmosphere Problem  19  all in the range d = 0.19 — 0.30. Flux distributions and profiles of selected Fe I lines were also calculated. Overall, the direct effect of sphericity on both the temperature structure of the atmosphere and the line profiles considered was found to be quite small. Indirect effects, such as opacity changes induced by temperature changes, were often substantial but unpredictable without detailed modelling. The treatment of radiative transfer in the spectral lines (whether pure absorption or absorption and scattering assumed) also had a major effect on the emergent spectrum. Schmid-Burgk et al. [87] computed nongray model atmospheres of late M giants and supergiants, for both the plane-parallel and spherically symmetric cases. The radiative transfer problem was solved by both the Schmid-Burgk [86] and Peraiah and Grant [78] methods. Both schemes yielded identical results within the 1% error limit adopted for convergence of the model iterations. The equation of state included 52 particle species plus free electrons, and the chemical equilibrium was solved by a Newton-Raphson iteration procedure. The continuous absorbers H , H , H , -  2  , Hj", H e , C, C , Mg, Al, Si and Fe were included along with -  -  Rayleigh scattering by H, H and He, and Thomson scattering by free electrons. Also, 2  the species CH, CN, CO, OH, H 0 , TiO, MgH, SiH and CaH were included as molec2  ular band absorbers, and treated in the JOLA approximation. Since this mean opacity representation 'smears out' existing flux windows, a 'picket fence' test model was also run in which the opacities were set at double the JOLA values for half of the frequency points, and to zero for the remaining frequencies. Models were also computed in which the molecular opacities (except H 0 ) were treated as pure scattering sources instead of 2  as pure absorbers. The resulting models show temperature differences of about 200K in the outer atmosphere (logr < — 2) between corresponding spherically symmetric and plane-parallel models. Similar differences exist between models run with the varying test opacities just  Chapter 1. The Stellar Atmosphere Problem  20  described. The authors also demonstrate that significant extension is common, with the geometric extension d as denned by equation (1.1) > 0.1 for practically all M stars with masses near 1 M and luminosities > 10 Z©. 3  0  Scholz and Wehrse [90] applied the models of Schmid-Burgk et al. [87] in an attempt to devise an observationally verifiable three-dimensional spectral classification scheme for late M giants and supergiants. While the standard two-dimensional MK system [73] essentially measures T and g, the significant extension expected for some late M giants cff  warrants the addition of a third parameter. Scholz and Wehrse proposed the use of the atmospheric extension d, as denned by equation (1.1). If the extension becomes significant, the photospheric radius R is no longer well defined. Scholz and Wehrse then proposed to arbitrarily define the radius by R = r(ri.2 = 1), and to use the standard M  relations for T and g to define these quantites in terms of this R. eff  Scholz and Wehrse calculated a series of models of solar composition with T  e / /  of  2750K and 3000K, for a variety of gravities and extensions. Synthetic spectra were calculated, and corresponding narrow band colours derived both on Wing's [107] original eight-colour system, and its extension into the infrared [108]. On the basis of these models, the authors suggest it should be possible to deduce the geometric extension d of the atmosphere from a knowledge of the TiO colour indices, using continuum colour indices and line strengths of ionized metals to infer T  eff  and g respectively. This is because  the TiO band strengths are a sensitive indicator of the extent of the cooler, outer layers of the atmosphere. While in principle this seems feasible, models used for this purpose should include VO as an opacity source (Scholz and Wehrse do not), since VO bands are prominent in the infrared near lfi for giants of spectral type later than M5, and may contaminate supposed 'continuum' windows. Scholz and Tsuji [89] computed a series of spherically extended model atmospheres for T  eff  = 2500K, 3000K and 3500K, for both M and C star compositions using the method  Chapter 1. The Stellar Atmosphere Problem  21  of Schmid-Burgk [86] to solve the spherical transfer problem. A revised version of Tsuji's [99] program was used to solve the equation of state, which included 34 elements and 61 molecular species, and to calculate the absorption and scattering coefficients. Continuous opacities considered included H, H~, H J , H e , Mg, Si and Ca as absorption sources, -  and H and H Rayleigh scattering and Thomson scattering by electrons as scattering 2  sources. Molecular absorbers included were H (collision-induced dipole), CO, OH, CN, 2  CH, C , TiO, MgH, SiH, CaH and the polyatomic species H 0 , HCN and C H , although 2  2  2  2  contributions from the latter two molecules were incompletely included due to the lack of laboratory data.  All molecular line blanketing was represented by straight mean  opacities. The most obvious result found was the pronounced lowering of temperatures and pressures in the outer layers of the spherically symmetric models as compared with their plane-parallel counterparts. Some problems are evident with the opacities used in these models. Scholz and Tsuji's [89] opacities for C star models at 3000K, even omitting polyatomics, are more than two orders of magnitude larger than those of Johnson [54], for depths near logrR = —2. M  The inclusion of HCN and C H at 3000K [89] results in a further increase in log K 2  2  R M  of  about 0.5 at this depth. The models of Eriksson et al. [36], which are plane-parallel C star models incorporating a more detailed analysis of the HCN opacity, show an increase in log KRM « 2 when HCN is included at 2500K at l o g r ^ = —2. Unfortunately, neither Scholz and Tsuji [89] nor Eriksson et al. [36] present comparable opacities used in their respective models, and thus a direct comparison of the models remains difficult. Still, it is evident that the cooler temperatures found in the outer layers of extended models greatly exacerbate the problem of calculating molecular opacities. Scholz [88] used the atmosphere code of Scholz and Tsuji [89] to calculate spherically symmetric models of M giants over the range 2500K< T  tff  < 3800K for various  values of the gravity and atmospheric extension. The purpose of this study was again  Chapter 1. The Stellar Atmosphere Problem  22  to examine the construction of a three-dimensional spectral classification scheme based on (T ,\ogg, d) and its sensitivity to changes in temperature and composition. Various eff  narrow band colour indices were also considered as indicators of (T ,log g, d). While e//  extension effects were shown to be significant and sometimes important enough to dominate T and g effects, the implementation of this classification scheme will be severely tff  hampered by the complex behaviour of the molecular band strengths. More accurate and consistent opacities are needed. Preliminary work on modelling extended atmospheres of late M supergiants using the opacity sampling (OS) technique of Johnson et al. [56] has been reported by Brown et ah [22]. The code used was based on Wehrse's [104] implementation of the Peraiah and Grant [78] solution of the spherical transfer problem. The bound-bound opacities of atomic lines, CO, CN, TiO, C , CH, NH, OH and MgH have been included using the 2  OS method, while H 0 was represented by a straight mean opacity. Difficulties were 2  encountered since these models did not attain the diffusion limit (as is necessary) at depth, due to limitations of the radiative transfer method [24]. These models remain unpublished, and the current status is unclear. Recently, Bessell et al. [19] have published an extensive grid of spherically symmetric models of M giants. The models were computed using the method described previously by Scholz [88]. This study continued progress toward the development of a quantitative, three-dimensional system of stellar classification of M giants and supergiants, based on the parameters T , g, and extension d. These models incorporated the additional eff  straight mean opacities calculated by Brett [21] due to the VO A-X, B-X, and TiO e molecular band systems. Models were calculated for a variety of luminosities and metallicities over the range 3000K< T  tft  < 3800K with extensions down to T  tff  = 2500K in  certain cases. An analysis of narrow band colours, mainly on Wing's [107, 105] 8 colour system, and broad band colours was presented. Particular consideration was given to  Chapter 1. The Stellar Atmosphere Problem  23  the determination of gravity-sensitive and extension-sensitive indicators. The authors concluded that a photometric separation of the effects of T , gravity and extension is eff  in principle possible. Spherically symmetric models of A stars, with line blanketing, were calculated recently by Fieldus, Lester and Rogers [39]. The model code used was a modified version of ATLAS [64] in which the line blanketing was treated using opacity distribution functions. The radiative transfer in spherical geometry was solved using the half-range moment method of Martin, Rogers and Rybicki [68]. The model was converged using a modification of the flux correction scheme of Simonneau and Crivellari [91] for spherical geometry. This procedure iterates an integral form of the first moment transfer equation to improve the estimate of the mean intensity J , and therefore the temperature via the equation v  of radiative equilibrium. Convergence was slowed in the spherical case, however, since some assumed 'invariants' are no longer invariant. No molecules were included in the equation of state or as sources of opacity since the code was intended to model stars of intermediate spectral type. An analysis of the effect of extension upon the temperature structure, emergent flux, and broad band colours was presented for a grid of stars with Tc = 10000K for varying gravities and extensions. ff  1.7  Objectives of this Research  This research has two main goals: • To develop a method to solve the spherically symmetric atmosphere problem based on the complete linearization method [7]. • To implement a working model atmosphere code for spherical geometry, with the aim of studying the extended atmospheres of late-type giant and supergiant stars.  Chapter 1. The Stellar Atmosphere Problem  24  The method used is broadly based upon that of Mihalas and Hummer [72] as employed in their pioneering study of atmospheric extension in 0 stars. The molecular line opacities are represented by mean opacities to reduce the computational effort involved. We would like to develop a model code which minimizes the additional solution difficulties of the spherical atmosphere as compared with the plane-parallel problem. Ideally, we also seek a spherical code that would reduce directly to the corresponding plane-parallel code in the limiting case of compact atmospheres (in order to aid in differential studies of sphericity). We have achieved success in both endeavours. Our hope is that the ATHENA model code developed here will permit a definitive study of the importance of extension in the atmospheres of late-type giant stars, and serve as a base for further work in the areas of steady-state dynamics (i.e. mass loss), pulsation, and non-LTE.  Chapter 2  The Spherically Symmetric Model System  2.1  Introduction  This chapter describes the theoretical method used to implement fully self-consistent models of late-type giant stars with static, but extended atmospheres. We have broadly followed the complete linearization method of Mihalas and Hummer (MH) [72] in solving the spherically symmetric model atmosphere problem. In this chapter we develop the moment equation formalism and discuss the numerical difficulties encountered in pursuing the MH method. The formal solution procedure of Hummer, Kunasz, and Kunasz (HKK) [47] is reviewed and the problem of numerical inconsistency between the moment and formal solutions is addressed. We present boundary conditions that account for the thermal emission of gas exterior to the outer boundary of the model. 2.2  The Model Equations  The complete linearization technique of Auer and Mihalas [7] using variable Eddington factors [8] to close the system of radiative transfer equations and adapted to spherical geometry as described by Mihalas and Hummer (MH) [72] will be implemented. The equation of transfer, assuming spherical symmetry, for a ray with direction fi = cos 0 crossing a radius r is  where I ,S ,p,Xv v  v  a r e  the radiation intensity, the source function, the gas density, and  the opacity per unit mass respectively. The source function is denned as the ratio of 25  Chapter 2. The Spherically Symmetric Model System  26  emissivity r\ to opacity Xv The opacity used here is the total extinction coefficient. v  We shall assume that this opacity can be expressed as the sum of a pure absorption component K and a pure scattering component o~ so that Xv — K -\- o~ . We shall V  v  v  v  further assume that all of these coefficients are isotropic. The emissivity per unit mass can then be decomposed into a thermal emission coefficient rf^ and a scattering emission coefficient  77*°.  For an LTE atmosphere, rf^ —  K B (T) U  U  and rj" — o- J . The source v  u  function is then  M^T^)  <2  -  2)  which depends only upon the mean intensity J„ of the radiation field and formally, at least, this dependence is linear. Under these conditions, the transfer of radiation in the atmosphere depends only upon «7„. Knowledge of the full angle dependent intensity I (p) is unnecessary. This allows the solution of the radiative transfer to be solved using v  the moment equation analogues of equation (2.1) and so reduces the dimensionality of the problem from 3 to 2. The linearity of S with respect to J„ permits the resulting v  finite difference system to be efficiently solved by arrangement into the Rybicki [83] block matrix form. Multiplication of equation (2.1) by powers of the angle coordinate p followed by integration over p yields the corresponding moment equations, any set of which always contains one more unknown than equations so that an additional relation among the moments must be supplied to close the system. Since only the zero order moment J  v  is actually needed to solve the transfer problem, the number of additional moments evaluated should be kept minimal. At depth in the atmosphere strict thermodynamic equilibrium (TE) is approached, the radiation field must approach the Planck field, and the diffusion approximation becomes valid. Here, J ~ B , and it can also be shown [70] that the second moment K & |i?„ v  u  u  so that the ratio of moments K jJ v  v  = f ~ | . This ratio /„ is called the Eddington v  Chapter 2. The Spherically Symmetric Model System  27  factor, and its knowledge can provide the additional relation needed to close the moment equations. The spherically symmetric radiative transfer problem can then be reduced to the determination of the Eddington factors and the solution of the system consisting of the first two moments of the transfer equation (2.1). The zero and first order moment equations are, respectively, ±£(r H )=p (S -J )  (2.3)  2  u  Xv  v  v  ^ + -{3K -J„) ar r  = -p vH  v  X  (2.4)  v  where knowledge of the Eddington factor implies K = fJ v  v  (2.5)  v  and where H is the first moment of the intensity, often referred to the as the Eddington u  flux. H is related to the standard flux by T = kirH . v  v  Unfortunately, while f  v  v  = | is valid at depth, its value in the outer layers of the  atmosphere is not known. To obtain f throughout the atmosphere, it is necessary to v  solve the full angle-dependent transfer problem for the specific intensity /„(/*) at each radius, and then perform the appropriate quadratures over the angle coordinate p, to obtain the moments J and K . The Eddington factor /„ = K /J v  v  v  v  is then immediately  obtained. In this study the impact parameter method of Hummer, Kunasz, and Kunasz (HKK) [47] was used to obtain the formal solution for the specific intensity  given  the temperature T and the total pressure p. While reverting to the full angle dependent problem just to find /„ and close the moment equations may seem to be a regressive step, it requires much less computational effort to solve the full formal problem (where T and p are assumed known) and to iterate the two-dimensional moment equation system to full consistency, than to directly iterate the full three-dimensional transfer problem.  Chapter 2. The Spherically Symmetric Model System  28  The second term on the left hand side of equation (2.4) numerically destabilizes the system [70]. An integrating factor for the left hand side can be found, namely, the sphericality transform,  '3/„-l\<fr'' introduced by Auer [5]. This permits the analytic removal of the offending term involving 1/r, yielding spherical moment equations exactly analogous to the plane-parallel form. Using the (radial) monochromatic optical depth dr — —pXudr as the independent variv  able, the moment equations (2.3, 2.4) become ±(r H )=r\J -S )  (2.7)  2  v  v  v  -J-(f»q»r J ) q dr  = rH  2  u  (2.8)  2  v  v  v  Eliminating H and introducing the generalized optical depth coordinate dX — q dr = v  v  v  v  —PlvXvdr, we obtain d , {Uq r J ) dX£ 2  ,  2  v2  v  v  2  r -S ). = -(J q v  (2.9)  v  v  which is the desired equation of radiative transfer for our spherically symmetric, static models. The method just described is essentially that employed by M H in their study of spherically extended O stars. They assumed a fixed radial grid and expressed the optical depth coordinate dr = —pXvdr in terms of the radius scale. The full system of equav  tions consisting of the moment transfer equations, radiative equilibrium and hydrostatic equilibrium were then linearized in terms of the fundamental physical variables, the temperature T, the number density n, and the mean intensity of the radiation field J . The v  various constitutive relations describing the density and opacity of the stellar gas were also linearized in'terms of T and n, assuming P = p(T,n)  (2.10)  Xu = Xu(T,n).  (2.11)  Chapter 2. The Spherically Symmetric Model System  29  The Eddington factor /„ was obtained from the formal solution of the full angle dependent transfer problem, where T(r) and n(r) were given. The Eddington factor was held constant during the subsequent linearization step. The linear corrections ST, Sn, and SJ found were used to update their respective quantities. The entire cycle was then V  repeated and (hopefully) iterated to convergence. In practice the MH method fails completely for late-type stellar atmospheres. While it proved possible to converge test models with constant or only slightly variable opacities, we were unable to converge any model with opacities typical of real late-type stars. These stars have continuous opacities dominated by the bound-free and free-free absorption of the H ion in addition to significant absorption from a variety of molecular sources in their -  outer layers. The opacity of the stellar gas under these conditions is very temperature sensitive, increasing by 5-6 orders of magnitude from the outer to inner boundary of the atmosphere. The local power law dependence of opacity on temperature,  K(T) OC T V  (2.12)  typically yields values of 7 ~ 10 in the vicinity of the hydrogen ionization zone (T ~ 10 K). 4  The temperature structure of the atmosphere is essentially determined by the constraint of radiative equilibrium. When K(T) exhibits a strong temperature dependence, as above, a typical temperature correction of ~ 10% needed to recover radiative equilibrium would result in a corresponding change in K by a factor of l . l  1 0  ~ 2.6! Effectively,  the optical depth scale slides back and forth with respect to the fixed radius grid as the temperature is perturbed. However, the response of the linearized system obtained using the MH method is generally not to follow the shifting optical depths at each radius (which would require all of the dependent variables T, n, and J to shift in unison) but v  to attempt to compensate for the change in K(T) by an offsetting change in the density of the gas. In the above example, a change in n by a factor of 2.6 will restore the original  Chapter 2. The Spherically Symmetric Model System  30  A r but at the expense of destroying hydrostatic equilibrium. These changes far exceed the typical radius of convergence of a nonlinear system of equations (typically ~ 10% ) so that all hope of convergence is lost. The fundamental problem here is that we are not only solving the transfer equation for J but for T as well. This greatly complicates the V  V  coupling of an already highly nonlinear system and destroys the convergence properties of the linearized system. The standard solution procedure in plane-parallel model atmosphere work is to use a 'natural' variable tied to the intrinsic 'depth' in the atmosphere, such as column mass or optical depth. The use of column mass makes the equation of hydrostatic equlibrium nearly an exact integral and numerically well-behaved. By contrast, the spherical form of the equation of hydrostatic equilibrium expressed in terms of the radius r yields solutions which grow exponentially inward; a difficult situation to handle numerically. The choice of radius for the independent variable appears to be dictated by the explicit appearance of the term involving 1/r in the original first order moment equation (2.4), or equivalently, by the appearance of the sphericality function q(r) in the transformed moment equation analog (2.8). As noted by MH, if r is not chosen as the independent variable and held fixed during the linearization step, it must also be linearized in terms of the alternate depth variable employed. Their conclusion was that linearization of the radius would lead to more complicated equations and that the effect upon the sphericality function q(r) would be to disastrously destabilize the transfer equations. In fact, this is not the case. We shall show in section 3.2 that it is possible to generalize the standard plane-parallel solution to the spherical case with relatively little increase in complexity over the MH formulation. Either column mass or optical depth may be used. Both were implemented in this work. The column mass scale models converged well (even better than did the optical depth based models) if the opacity was well-behaved and possessed no strong dependencies on the temperature or pressure. However, these  Chapter 2. The Spherically Symmetric Model System  31  models faired badly when realistic opacities (which are very temperature-dependent) were added. The most suitable depth scale was found to be the Rosseland mean optical depth. The column mass still suffers from the drawback that the differential optical depth dr = v  Xvdrn remains quite sensitive to temperature perturbations in regions where x(T) is a strongly varying function of temperature T. The advantage using a mass scale is that the hydrostatic equation remains unaffected by such temperature perturbations, provided the gas pressure p rather than the number density n is used as the fundamental variable. Basically, the gas pressure is insensitive to perturbations in the temperature structure of the atmosphere. We have, therefore, retained p as the fundamental variable throughout this work, in contrast to MH. However, the use of a Rosseland mean optical depth scale automatically provides T(T ) RM  and  J (T ) V  RM  = B (T) V  at depth, given simply by  T =^[r 4  f i M  +  ?  f  i  (2.13)  M ] ,  where 9H(T) is the Hopf function appearing in the solution of the plane-parallel grey atmosphere problem. Strictly speaking, this result is true only if the atmospheric scale height h <C R* at depth. In practice, this result remains fairly accurate since extension effects do not generally penetrate too deeply into the atmosphere. Furthermore, we now have dr = Xvdm = (xulXRM)dT v  where  Xv — XV/XRM-  RM  =  Xvdr , RM  (2.14)  The overall temperature dependencies of the opacity, such as the  rapid increase in x„(T), are largely cancelled out in the ratio x.v(T). The drawback is that the hydrostatic equation now suffers from column mass variations as a result of temperature perturbations. Since the hydrostatic equation seems better able to cope with such perturbations than do the transfer equations, the decision was made to use the Rosseland mean optical depth as the independent variable throughout this work.  Chapter 2. The Spherically Symmetric Model System  The final formulation of the model uses r  32  as the independent variable, and T, p, r  RM  and J as the dependent variables in the linearization. The system is closed by the use v  of the constraint equations. One constraint is provided by the equation of radiative equilibrium which ensures the conservation of energy in the radiation field by requiring that the energy radiated is balanced by the energy absorbed in each layer. Therefore,  f  (v,-xMdu  =0  (2.15)  lo Jo  which for the source function specified by equation (2.2) reduces to  f  K (B v  - J ) dv = 0.  v  v  (2.16)  Jo  The form of this equation is important. It proved necessary to make the following rearrangement  I I  n J dv v  v  -1 = 0  OO  KB U  (2.17)  dv  v  0  in order to obtain a numerical solution. This is due to the fact that the left-hand side (LHS) of equation (2.16) is a complex, strongly temperature-dependent function for a cool molecular gas where molecular opacities are important. Most of this variation is cancelled out in the latter equation (2.17), however, leaving the LHS a well-behaved, slowly varying, monotonic function of temperature. As a result, the range of convergence of the corresponding linearized equation is greatly increased. A second constraint is provided by the assumption of hydrostatic equilibrium in the stellar atmosphere. For a spherically symmetric atmosphere, this constraint can be expressed using the expression from Mihalas [70] converted to optical depth variables, P dr d  RM  *  _ l£L  GM =  XRMT  2  ( 2  47rcr ' 2  l  g  )  Chapter 2. The Spherically Symmetric Model System  33  where the gas pressure p is assumed to obey the ideal gas law p = nkT. The second term on the right-hand side of this equation arises from the radiation pressure gradient (V • P ) , and the parameter 7 is defined, following Mihalas [70] as r  7 =— ^/ X-RM-"  (2.19)  XuH„dv,  Jo  which describes the efficiency of momentum coupling between the radiation field and the stellar gas. The final constraint, the radius (or depth) equation,  dr  RM  PXRM  provides the relation between the radius and Rosseland mean optical depth scales. 2.3  The Formal Solution  The formal solution of the transfer problem refers to the direct solution of the equation of transfer for J for given values of the fundamental physical variables T(r ), p(r ) v  and  T^HM).  RM  RM  This solution by itself is of little physical significance since the quantities T  and p in turn depend upon J in a real stellar atmosphere, coupling via the equations v  of radiative and hydrostatic equilibrium. However, it is of value as a step in an iterative solution procedure. The solution of the model atmosphere described so far relies on the collapse of the inherently two-dimensional transfer problem (depth, angle) to a one-dimensional one (depth) through the use of the moment equation system. This system is closed through the use of variable Eddington factors and is then iterated to a self-consistent solution. The angular distribution of the radiation field is then effectively described by a single value (the Eddington factor) which is not linearized in the moment equation solution. Therefore, although iteration of the linearized moment equations will yield a self-consistent  Chapter 2. The Spherically Symmetric Model System  34  solution, this procedure ignores any changes in the angular distribution of the radiation field arising during these iterations. Therefore, it is necessary from time to time to re-evaluate the Eddington factors, which is done by a formal solution of the full twodimensional transfer problem. The method of solution of the formal problem used in this study is the impact parameter scheme of H K K which is also summarized by Mihalas [70]. The procedure involves solving the one dimensional transfer problem along a series of parallel rays (the 'impact parameters') intersecting concentric spherical shells, as illustrated in Figure 2.1. Each ray pi intersects the radial shell r at an angle with cosine d  p = fi(r ,pi) = — Jrl-p} r * di  d  = z /r di  (2.21)  d  d  and so a solution along each of these rays solves the full two dimensional spherical symmetric transfer problem, since /„(/*<£) is obtained at all grid radii r . 4 d  There are two classes of impact ray in this scheme distinguished by differing inner boundary conditions: (1) those rays tangent or exterior to the inner radius of the grid; and (2) the remaining rays which penetrate the 'core', interior to the inner radius. The equation of transfer along each ray has the standard form  ±  ^  d  P  i  )  = PMr)-  X,(r)£(*,ft)]  (2.22)  where the + and — signs refer to outward and inward directed radiation respectively. Also, it is assumed r = y/p + z . Denning the optical depth along the ray, dt = 2  2  u  —pxvdz, and with S = v  i] /xv, u  ^  dI±( ,t ) I„{Pi,tu) ~ S {r) dt V It is now convenient to define the variables Pi  v  (2.23)  v  u {pi,U) = \[l+{piX) + K{pi,t )\  (2.24)  v {pi,U) = \[l+(p ,t )-l;{p ,t )]  (2.25)  v  v  v  i  v  i  v  Figure 2.1: Impact parameter coordinate system for formal solution  Chapter 2. The Spherically Symmetric Model System  36  which behave as 'mean-intensity-like' and 'flux-like' quantities respectively. These obey the relations du„( ,t ) _ dt„ dv (pi,t ) = u ( ,t ) - S (r) dt Pi  v  =  v  (2  Pi  u  26)  (2.27)  v  u  _  v  v  from which v may be eliminated to get the final form of the transfer equation along the v  ray ^  U  v  ^  t  v  )  = u„( ,t ) Pi  v  -S„(r)  (2.28)  The solution of this second order differential equation yields u (pi,t ), and v (pi,t ) can v  u  v  u  then be found by equation (2.26). The moments of the radiation field are obtained by quadrature over the angle coordinate p as follows, J (r)  =  ( u (r,p)dp Jo  (2.29)  H (r) u  =  I pv (r,p)dp  (2.30)  K (r)  =  j p u (r, p) dp. Jo  (2.31)  v  u  v  u  2  v  The updated Eddington factor is then given by ^ ( r ) = K (r)/J (r). u  u  There are two difficulties with using this formal solution to determine /„. One is the potentially large number of impact parameter rays needed for problems with opacities displaying a strong dependence upon frequency. This is since the impact parameters chosen must provide good coverage of p over the radial range where the atmosphere becomes optically thin. For the situation mentioned here the required optical depth range may extend over a considerable radial extent. The second drawback is more fundamental. While the calculation of moments by quadrature of the formal solution intensities is mathematically equivalent to the direct solution of the moment equations, assuming /„ is known exactly, the corresponding discrete representations of these problems are not  Chapter 2. The Spherically Symmetric Model System  37  identical. Thus, the impact parameter solution can not be made exactly consistent with the direct solution of the moment equations of transfer [71]. This problem is exacerbated since the moments found by the formal solution method are obtained by quadrature over a varying number of intensities at varying values of the angle coordinate y, for each radius. The lack of a consistent quadrature grid results in the introduction of unavoidable numerical 'noise' in the radial behaviour of the moments so obtained. For example, in the outer layers of the atmosphere, r J approaches a constant value. 2  v  For sufficiently large r, numericalfluctuationsin the value of the quadrature integral will eventually dominate over the actual physical variations. Any attempt to use these values of J„ in the moment transfer equation (2.9), which requires the accurate evaluation of the second radial derivatives of J„ will result in severe problems with the convergence of the linearized system. It appears that this difficulty can be circumvented in the outer layers where r J is nearly constant by retaining H from the formal solution instead 2  v  v  and integrating equation (2.8) to obtain J . Elsewhere in the atmosphere J is returned v  u  from the formal solution, and H is then found by evaluating the derivative on the left v  hand side of equation (2.8). In detail, the criterion used was to evaluate J directly by quadrature and H by v  v  differentiation using the moment equation (2.8), unless A(f q r J )/(f q r J ) 2  v  u  2  u  v  u  u  for the  depth interval [£.,•, i,- i] was < 1%. In this case, H was instead evaluated by quadrature +  u  and J calculated by the integration formula v  AJ = -—— UW V  2  f J  3+1  q r H dt 2  v  v  v  (2.32)  tj  derived from equation (2.8). The relative accuracy of the quadrature integrals is of order 1 0 - 1 0 . The expression AJ /J -5  -6  V  V  provides an estimate of the loss of significance  incurred in the numerical differentiation of J . When this loss of significance exceeds two u  decimal places, we choose instead to evaluate H directly by quadrature. The domains v  of these modes of evaluation of J„ are shown in Figures 2.2 and 2.3 for two converged  Chapter 2. The Spherically Symmetric Model System  38  models. This procedure ensures that the J used in the subsequent iteration of the v  moment equations differentiates to give the correct flux according to equation (2.8) at the boundaries. This is extremely important since both boundary conditions involve flux constraints, and if these are not accurately imposed, systematic errors will be present throughout the entire atmosphere. 2.4 2.4.1  The Boundary Conditions The Formal Solution  First, we shall consider the boundary conditions needed for the formal solution of the equation of transfer. This involves the solution of a series of one dimensional transfer problems along each of the impact rays. The usual boundary conditions for this second order differential equation involve the specification of the derivative du /dt of the v  v  intensity at both the inner and outer boundaries. From equations (2.24) - (2.25) we have v = u -I~ v  =1+ - tt„.  v  (2.33)  Since du  v  =  < - > 2  34  we obtain the following relations which supply the necessary boundary conditions, namely, du - = u -Idt V v  (2.35)  at the outer boundary, and du  v  (  2  -  3  6  )  which proves to be the more convenient form for the inner boundary, since here v can v  be directly evaluated in the diffusion limit.  Chapter 2. The Spherically Symmetric Model System  •L  60|  Figure 2.2: Domains of calculation of J and H : Model 01310191 u  u  39  Chapter 2. The Spherically Symmetric Model System  IAIH  -L 60|  Figure 2.3: Domains of calculation of J and H„: Model 02310191 v  40  Chapter 2. The Spherically Symmetric Model System  41  Formal Solution: Inner Boundary Condition To express the actual inner boundary condition, we must distinguish the two classes of impact rays mentioned earlier: those rays intersecting the core interior to the innermost radius of the grid, and those rays tangent to, or exterior to, the core. At the inner boundary, it is assumed that the optical depth is sufficiently great that the diffusion approximation is valid. Then, dB /„(/*) « B + fi-r^  (2.37)  vu = Wt - K) « ^  (2-38)  v  'v  and  The gradient dB /dT can be directly evaluated, as follows. The flux H is given by v  v  v  ^ldB _  1 dB _  v  IdT  v  (ldB \ v  assuming the diffusion approximation. Then, since the total flux,  is known, we must have t  rr ,  00  1  d  T  r  1 dB  v  '  ,  •  Therefore, dT dr  ZpH S7xZ {dB ldT)du\ X  (2.42)  v  and from equation (2.39) dB dr  IdT ( 1 dB p dr \x dT  v  s  v  v  (2.43)  u  or, after substitution of equation (2.42), d  -^ = ZH  dr  v  X^jdBjdT) rx^idB./d^du  (2.44)  Chapter 2. The Spherically Symmetric Model System  42  An alternate form can be derived using the value of H from equation (2.40) and the Rosseland mean opacity XRM defined by _ J ~(dB„/dT)du _ 4.T3 ~ r X^(dB /dT) du 7r / " x^(dB,/dT) du  .  o  X R M  ^  v  Therefore,  r±^ Jo  Xu  i  v  ^Il,  =  dT  .46)  (2  TTXRM  and so an alternate form of dB /dT is obtained by substitution of the above expressions v  v  for H and / xZ dB /dT dv into equation (2.44), yielding 1  v  dB dr  v  v  where Xu =  /dB \ 64ir<rr T Xv \ dT J ' 3i,  v  2  ^  3  XV/XRM-  For impact rays intersecting the core, then dB Vu = P~rr, v  (2.48)  while for rays exterior to the core the interior boundary condition is applied, not at the inner radius, but rather at z — 0 along the axis of symmetry. Here p = 0, and therefore, v = 0,  (2.49)  v  as also follows directly from symmetry considerations. The general inner boundary condition for an impact ray is thus du  v  dt  u  3J&,  ^647r<7r r x 2  3  Formal Solution: Outer Boundary Condition Traditionally, the outer boundary for radiative transfer has been derived under the assumption of no incident radiation, or I~ = 0. Then, the boundary condition is just du  dt  u  v  u  v  (2.51)  Chapter 2. The Spherically Symmetric Model System  43  as follows from equations (2.33)-(2.34). However, stellar material is present exterior to the outermost radius, as is explicitly assumed by the nonzero values of the gas pressure and optical depth adopted on the outer boundary. In this study, thermal emission from gas exterior to the outer boundary is also included, yielding a nonzero boundary value of I~ also. For a spherically symmetric, static isothermal atmosphere, the mass obtained integrating equation (2.18) to infinity is itself infinite, as is the optical depth, and the pressure and density remain finite at infinity [30]. For full consistency, the equation of hydrostatic equilibrium should be replaced by its hydrodynamic generalization, which permits a flow. Lacking this, an arbitrary cutoff procedure must be adopted. M H assumed an exponential, isothermal atmosphere existed exterior to the outer optical depth grid point at T\ and also assumed the Rosseland mean opacity to be independent of pressure in this region. We will generalize this latter assumption and permit a power law dependence of the opacity upon density of the form K(P) OC p , where the index A is then given by x  x  p &K  ,  = ir -  <- > 2  P  52  We still ignore any spectral variations in A and evaluate this quantity assuming K = K  R M  ,  the Rosseland mean absorption coefficient. Then, we have the following asymptotic behaviour  i T(r)  = Ti  (2.53)  p(r)  =  (2.54)  p(r)  =  P l P  l  exp exp  ( - ^ )  Xu(r) = x u > e x p ( -  A(r  "•»>  ~ ) ri)  (2.56)  where h is the scale height of the atmosphere at the outer boundary, and radiation pressure is ignored.  Chapter 2. The Spherically Symmetric Model System  44  An incident radiation boundary condition for radiative transfer can now be derived. The formal solution procedure of H K K requires the specification of a boundary value of the incoming radiation along each impact ray making an angle, p, = cos 6, with the outer boundary radius r = T\. The transfer equation for radiation incoming along each ray exterior to the atmosphere is then dl~  ~K-S  v  dt  (2.57)  V  v  or dl7, d  z  (2.58)  PXv(Iv-S ), V  where the source function is as defined by equation (2.2). Assuming in this region that I~ <C S for r > r i , equation (2.58) reduces to  •'  V  ^ •  (2.59)  = -PXUS . V  For late-type stars we can assume negligible back scattering of J into I~, V  S  V  =  K B / V, V  V X  and so we obtain  -^ - -{-^r)^- ^=  px  pK  -  (2 60)  For early-type stars it will, in general, probably be necessary to include scattering contributions in a calculation of this type. Integrating inward along the ray, assuming I~(oo) = 0, yields J "(zi) = 5 1  l v  A  p(z) (z)dz, Ku  (2.61)  Jzi  where Z\ — z(ri). We define the extension parameter, d = h/ri, and also £ = z/ri and recall that r = <\J'z + p , where z is the distance along the impact ray and p is the 2  2  transverse coordinate. The reader is again referred to Figure 2.1 for details of the impact parameter coordinate system. Note that £i = Z\jr\ = p,. Substitution of the asymptotic behaviour of p{r) and Xv{ ) given by equations (2.54) and (2.56) into the integrand of r  Chapter 2. The Spherically Symmetric Model System  45  equation (2.61) yields •KM  = Pi*i B\ V  V  i J UL  (l + A ) ( r - n )  exp  (2.62)  rid*  or hub ) = PiKivB\ r\ I f 1  v  di.  exp -(1 + A)  (2.63)  Further manipulation then gives +  ?-P?-I)  di  (2.64)  which can be written in the form T  -/x  PlhK Bi lv  1+ A  v  (  d  (2.65)  V1+  where the function G(p, d') is defined by (2.66) where  =  (1 + A) above.  There are two simple special cases. First, note that for the radial ray p = 1, the expression for G reduces to G(l,d) = 1,  (2.67)  while another reduction is realized in the case of a compact atmosphere (d — 0), where G(p,0) = -. The Rosseland mean optical depth T  R M  (2.68)  may be integrated radially inward from infinity  to the outer boundary to obtain PlhXl.RM  Tl.RM —  (2.69)  1 +A  and the corresponding monochromatic (radial) optical depth is  A  1,RM ( X " \Xl,RM/ 1  Til/ —  T  RiaXlv —  1+A "  (2.70)  Chapter 2.The Spherically Symmetric Model System  where  X  L  /  =  XV/XRM-  Letting  =  46  then  K /XV, U  and so equation (2.65) simplifies to IrAf ) = i,RM^iuXiuBlvG(fi,d') 1  T  = t (fi)B ,  (2.72)  lu  abStV  where t bs,u{p) = l,RM^luXluG(n,d')  (2.73)  T  a  is the absorption optical depth at the outer boundary along the ray with direction fi. One further generalization is useful. Ideally, the boundary optical depth T i  | R M  is  chosen sufficiently small that T\ <C 1 for all frequencies. However, during the production V  of the models it was found that some frequencies in the far ultraviolet were optically thick due to strong bound-free absorption from abundant light metals and that the optical depth in the strongest molecular bands approached optical depth unity, even with T 10~ . 6  A considerable improvement in accuracy occurs if the above t b a  3iV  R M  =  is replaced in  equation (2.72) by 1 — exp(—£„&,,„). Then, in the optically thin limit of t b, <C 1, the a  iV  previous result, equation (2.72), is recovered for J^,(/i). In the optically thick limit, t bs,v ^ 1 the correct result a  5  IM  = Biv  (2-74)  is obtained as well. The final expression for incident radiation at the outer boundary of the atmosphere thus becomes ^ ( ^ ) = {l-exp[-T , ^ Xl.G(/i,c^ )]}^• ,  ^  nM  I/  (2-75)  Given this value of J~(/i), the outer boundary condition of the radiative transfer equation along the impact ray with direction fi is ^  = «*--tt/0,  (2-76)  Chapter 2. The Spherically Symmetric Model System  or,  du -u -{ldt„  47  exp[-T i XuG(p,d')]}B  v  v  liRM v  (2.77)  u  where the depth subscripts have been suppressed for clarity. Parametrization of the G function  The function G(p, d') is defined by the integral d^  (2.78)  The change of variable £' = £ — p results in the equivalent form 1 f°° G(p,d') = j,J  e x  di'  P  (2.79)  which is more convenient for numerical evaluation, since the constant limits of integration permit the use of a fixed (p,d') grid. This representation was used in this study to evaluate G(p,d') over the two dimensional grid {(pi,d' )}, with k  Pi e  {0,0.001,0.002,0.005,0.01,0.02,0.03,... ,0.99,1}  d' <E {0,10- ,10~ ,0.01,0.03,0.1,0.3,1,3,10,30,100} 4  3  k  (2.80) (2.81)  except for the case d' = 0, where the exact value G(p,0) = 1/p was used. In order to permit the rapid evaluation of this integral in the ATHENA atmosphere code, an analytic parametrization was sought. The limiting values in the previously mentioned special cases suggested a function of the form G(p,d') =  a .(p-iy d  with a > = d  0,&d< =  1 + b .(p-l) + V  (2.82)  d  1 for the hmiting case of d' = 0. Then the results G(l,d') = 1 and  G([i,0) = 1/p are recovered directly.  Chapter 2. The Spherically Symmetric Model System  48  In practice, a reasonable fit to G(p, d') was achieved using the following modification of equation (2i82), C(p,d') a ,(p - l ) + b ,{p - 1) + 1'  •G(M')  (2.83)  2  d  d  where ((p,d') is a slowly varying correction of order unity given by, M  ~ 0-17  A>\ -  W> >d  M  (2.84)  (C„.-1) + 0.83'  and the coefficients, which are functions of the extension parameter d', are:  a  logd  d> = p{& — log d') exp  p = 0.072697 Q = 0.866726  with  r = 1.355135  (2.85)  s = 0.419602 logd' + q  1 - erf  Cd' =  = 0.166208  with  r = 1.083183  with  l + t exp  t  = 0.13289  V  -  2.14884  (2.86)  (2.87)  w = 0.83685 This parametrization, obtained using the OPDATA analysis package [18] yields the value of G(p, d') to within ~ ±2% over the ranges p = [0,1] and d' = [0, oo]. This parametrization of G should be more than adequate for our purposes since the assumption of an exponential atmosphere is itself only approximate in spherical geometry, 2.4.2  The Moment Equations of Transfer  This system involves the two low order moments of the transfer equation, averaged over angle and closed by the use of the variable Eddington factor /„. Typically, boundary conditions are imposed assuming knowledge of the flux H at the boundaries, leading v  naturally to the use of the first moment equation d {f q r J ) = r H dX 2  v  v  v  2  v  v  (2.88)  Chapter 2. The Spherically Symmetric Model System  49  for this purpose. Moment Equations: Inner Boundary Condition  The diffusion approximation is assumed to be valid at the inner boundary, giving 1 dBv Z~oX  TJ„ =  647r<rr r Xp 3  2  where equation (2.47) has been used to evaluate dB /dr . The inner boundary condition u  v  is then 'i<'""^> =  (#)  •  < - °) 2  9  Moment Equations: Outer Boundary Condition  The approach of MH was to supply an outer boundary condition for the moment transfer equation by the introduction of a second Eddington factor g = H /J . v  u  v  The boundary  flux is H = g J , and the resulting boundary condition becomes v  v  v  •j^-(fuq r J ) = r'g J  (2.91)  a  v  v  v  v  which involves only the fundamental variables T, p, r and J . v  For the case of no incident radiation, MH assume g to be invariant, thus allowing v  the boundary condition to be linearized. However, in the presence of nonzero incident radiation, /^"(/i) > 0, this assumption of the constancy of the ratio of H~ /J is false. u  u  In this case, both J and H will now depend on the inward directed intensity, which is u  v  a function of the boundary temperature, and g will possess a temperature dependence v  and can not be assumed invariant. An appropriate generalization of the approach of MH is to define the outgoing radiation moments J+ = \ [ I {p)dp Jo  (2.92)  H+ = \ ( til (u-)dp,  (2.93)  v  v  Chapter 2. The Spherically Symmetric Model System  50  and their ratio gi = H*/J+, and to assume that gi may be held constant during subsequent linearization steps. The boundary condition, however, involves the full angleaveraged moments J and v  JET„, SO  that «/+ and  must be related to J and H . v  v  For completeness, we define the complementary incoming radiation moments, J-  =  \ J I (n)dfi v  (2.94)  H~ =  \ J fil„(n)dn,  (2.95)  and note that Ju = . J+ + JH  v  ,  (2.96)  = H+ + H^.  (2.97)  Then, H  Hj + H~ _ Hj(l + HZ I Hj)  J~  J+ + J," ~ J+(l + J-/J+) •  v  9V  ~  v  ( 2 , 9 8 )  If we assume that I~ <C /J", J,7 <C «7j~, and JET~ <C H*', we obtain, to first order,  (  1+  § - | ) ^ (  1  f - f )  +  -"»  (2  since iZi, « if+ and J « J+ to this level of approximation. v  The quantities J~ and ITj" are just the moments of the inward directed radiation and, therefore, K = \ I[ Iu{ti)dti = \ I I-(n)dn = ^T ( XuB [ G(fi,d')dfi J-i Jo Jo ltRM  v  v  (2.100)  and H~ =  lf J-i  nl {u)dn = v  -\ f  Jo  where, by definition, Iy(fi) = I^—fi).  ftl-(ii)dp =  -\TI Z UB IRM  Jo  VX  V  f  /iG(fi,d')dfi, (2.  Chapter 2. The Spherically Symmetric Model System  51  Denning the moment functions, Gj(d) = \ f G(»., d) dp  (2.102)  G (d)  (2.103)  1  Jo  H  = \ f pG(p,d)dp Jo  the incident radiation moments can now be written as J~ = +r ^X,B Gj(d')  (2.104)  H~ = -T ( uB GH{d')  (2.105)  hRM  ltRM  v  vX  v  where d' = d/(l + A), as before. Also, as in the formal solution, the monochromatic absorption optical depth on the outer boundary Tab,,u  =  (2.106)  TRM^XU  has been replaced by 1 — exp(—T & ). This approximation is less valid than in the 0  4II/  formal solution case since the moment functions Gj and GH were evaluated under the assumption of optically thin outer layers. These functions will therefore be inaccurate for moderately optically thick boundary layers. This approximation, which we used for our model calculations, gives Ju.=  +[1-eM-n,RM£vXv)]BvGj(d')  (2.107)  H~ = -[1 - exp(-T i Xv))BvG (d') hRM v  H  (2.108)  In retrospect, a better estimate would probably be K  = H[l-eM-2n,RMivXvGj(d'))}B  H~ =  - exp(-AT i x G (d'))}B liRM u  v  H  u  u  (2.109) (2.110)  which yields not only the correct optically thin results above but also the correct optically thick results J~ — B /2 and H~ = —B /4. This estimate will be tried in future models. v  v  Chapter 2. The Spherically Symmetric Model System  52  The outer boundary condition can now be written as -^riUqvT J„) = r g J = r gt (j 2  2  2  v  v  v  + —H~ - J~) ,  (2.111)  where g has been expanded using equation (2.99). Since to first order v  ^H~ = H-,  (2.112)  9v  we obtain the result d dX  (f,qyJu) = r [gUu - {9tJ; ~ H-))  (2.113)  2  v  With this, our goal of finding a boundary condition involving only the fundamental variables J and r, and the invariant ratio g+, has been realized. This expression may v  now be expanded in terms of the moment functions Gj and Gjj as •^•{UqyJu) = r {gtJ - [1 - exp(-T ^ u)][G (d') + gtGj{d')}B } 2  v  RM  X  H  v  (2.114)  and yields the desired outer boundary condition. Parametrization of the Gj and GH Functions  The moment functions Gj(d') = \ f G(fi,d')dfi Jo  (2.115)  G (d') = \ f pG(n,d')dfi Jo  (2.116)  H  can also be parametrized in a form suitable for repeated, rapid evaluation. Numerical quadrature of the previously obtained values of G(fi,d') over the angle grid {^} yielded values of Gj and GH for values of the extension d' G {d'k}- Again, the OPDATA software package was used to obtain analytic parametrizations to Gj and GHThe function Gj was parametrized by the form  j( ) = -j-Tj—r~TZV~h~  G  d>  7d'(l — tid') + gd'hd'  <- ) 2  117  Chapter 2. The Spherically Symmetric Model System  53  where lb  = ad 9d>  (2.118)  'IT 1 + tanh ^log d' + g 1 +. tanh  (2.119) (2.120)  and, a = 1.13238 b = 0.213648 (2.121)  q = 0.131120 r = 1.13353 s = 1.56082 t =  1.42257.  The function GH was represented by the expression 1  G (d') =  (2.122)  H  f  1+  l  1 + tanh  +q  with coefficient values q = 0.020576  (2.123)  r = 0.900053 These parametrizations of Gj and GH are also accurate to ± 1 % .  2.4.3  Other Boundary Conditions  The remaining equations in the model system consist of the equations of radiative equilibrium, hydrostatic equilibrium, and the equation defining the radius depth scale. Of these,  Chapter 2. The Spherically Symmetric Model System  54  the equation of radiative equilibrium, as formulated here, is not a differential equation and requires no specification of boundary conditions. The other equations are both first order differential equations and require the specification of a single boundary condition for the model to be well-posed. It is convenient and numerically preferable to specify these conditions on the outer boundary. Hydrostatic Equilibrium: Outer Boundary Condition  The assumption of an isothermal, exponential atmosphere exterior to the outer boundary with an opacity dependence given by XRM(P) C< p allows the equation of hydrostatic x  equilibrium to be integrated analytically in this boundary region. Assuming the pressure scale height fe<ri gives P l  - (1 +  A)—  L*~/lXl,RM  -j-  Xl.RM  "/»  .>.  10  4*cGM,(l + A)J '  ^1  as the boundary pressure, where =  ^ Xl.RM  )  V °P  •  (2.125)  / 1  The (1 + A) term in the denominator of the radiation pressure term in equation (2.124) was accidentally omitted from the calculation of i in the current series of models. The P  effect on the models should be very small. Radius Equation: Outer Boundary Condition  We have simply set the outer boundary radius ri = R = £R.,  (2.126)  where R is a reference outer boundary radius, determined by the scaling the stellar radius R (a fundamental model parameter) by the factor £. The factor £, and consequently, t  the outer reference radius R, remain unchanged by the linearization step.  Chapter 2. The Spherically Symmetric Model System  55  The scale factor £ in our models is determined initially such that r(ro) = R*, where the reference optical depth r = 1. This value of £ could be updated during the formal 0  solution step. This has not been done in this study, however, since this would further slow the overall convergence of the already computationally intensive models. The only consequence of this neglect is that while the converged models produced remain fully self-consistent, the parameter il» specifying the stellar radius will have drifted slightly from its originally prescribed value. The values of the originally specified Af* and Z* remained unchanged. In the domain of T , g and M , the values of T eff  iol  eff  and g will  drift slightly from their original specifications. In practice, these effects were found to be small with AT  tff  < 60K and A log 5 < 0.03.  Chapter 3  The Discrete System and Solution Procedure  3.1  The Finite Difference System  3.1.1  The Formal Solution: Cubic Spline Difference Formulae  The model equations must be discretized in order to obtain a numerical solution. The second order Feautrier form into which both the transfer equation along an impact ray (for the formal solution) and the moment form of the transfer equation have been cast lends itself naturally to a discrete representation by cubic splines. We shall first consider the formal solution using the (one dimensional) transfer equation along impact rays, following the approach of H K K . Since the formal solution proceeds independently for each frequency of interest, all frequency subscripts will be suppressed in this section for clarity. Consider a function u(t), which is desired to approximate the solution of a second order differential equation. It follows that u(t) must be at least twice differentiable; denote this second derivative by  £  = "<i).  (3.1)  Following the approach of Alhberg [1], a simple candidate for a suitable approximating function is obtained by letting M(t) be the piecewise-linear interpolant of the grid {Afj}, or = XT  - *) + *+i (* " *<*)] M  f o r  [**  (3-2)  where Atd = tj+x — t . Direct integration of Mj(t), constraining the result to interpolate d  56  57  Chapter 3. The Discrete System and Solution Procedure  the grid values {ud}, then yields the cubic spline formula u (t)  = -^-[Md(td i-t)  d  + M (t-tdY  3  +  (3.3)  d+1  + (6u - At M )(td+i - t) + (6«  - At M )(t - t )\  2  d  d  d  2  d+1  d  d+1  d  for t e [t , t ] d  d+1  It is seen that the continuity of u(t), and of M(t) = cPu/dt , is ensured by the interpo2  lation of the respective grid values. However, the first derivative '  U  {^)  " ^ [ZMd ,(t-tdY-ZMd{td+,-t)  d{t)=  d  (3.4)  2  +  + (6u i - At M ) - (6u - At Md)] 2  d+  2  d  d+x  d  d  for t € [t , t ] d  d+1  is not automatically continuous across the nodes (grid points), and this requirement must be additionally imposed. By direct evalation for t — td, we have u' (t ) = d  Ud+1  d  ~  - \\M i  Ud  A  d+  + \M )At d  (3.5)  d  and  «i_i(*a) =  Ud  Ud  ~ + {\M + iM _ )At _ 1  d  <J  1  d  1  (3.6)  L±t -\ d  which expresses the value of u'(t ) as obtained by interpolation on the adjacent intervals d  [t , t i] and [£d_i,i<j] respectively. Continuity of u'(t) then requires d  d+  u' (t ) = u' _ (t ). d  d  d  1  (3.7)  d  Rearrangement of this constraint yields the desired spline difference equation, Ud+l  Ud  At At _ d  d  1/2  (  Atd-1/2  1 —  ,  \t \At  dd  + \M  d+1  where  Atd-1/2  1  1  ^  \  At -J Atd-iJ  +  d  Atd^Atd-i/2  (-  = \{Atd-i + A ^ ) - Then, if the second order differential equation is of the  form du 2  ^  .  =^ , 4  ,  (3-9)  Chapter 3. The Discrete System and Solution Procedure  58  substitution of the {M } values directly yields the cubic spline discrete representation of d  the differential equation (3.9). The Feautrier form of the transfer equation along a ray M(t) = ^ = u-S(t)  (3.10)  d  conforms to this form, and we immediately obtain Ud+l  At At _ d  d  Ud  At _  1/2  d  1  ( 1/2  + ixr^-(«*Hi -  \At  ,  1  \  «<f-l  At ^J  d  At ^At _  d  S ) + d+1  d  d  1/2  l(u - S ) + I - ^ z L ^ d  d  (3.11)  5 _ ), <l  1  Finally, the above derivation constrains u'(t) at all interior nodes but leaves this derivative unconstrained on the boundaries. Thus, if d = 1 and d = D refer to the inner and outer boundary nodes of the grid respectively (where D is the number of depths), the corresponding boundary values u'(t\) and u'(tr)) must also be specified for uniqueness. This is very convenient in our case of the transfer equation, since the usual boundary conditions involve the specification of du/dt = u'(t), and the discretized versions of the differential equation boundary conditions immediately supply the necessary spline boundary values. Then, from equations (2.35)-(2.36) we have 1  =  ^AJT " ^ (  «i>-i(*z>) = ""T J  ~  UD  A  M2  *  +  ML)A<1  + {\M  1  D  =  UL  ~ ~  + §Afi>_ )Ai _i = v 1  '  IK  D  (3 12)  (3.13)  D  as the corresponding discrete boundary conditions for impact ray k. The solution of this tridiagonal difference system requires knowledge of M = u -S , d  d  (3.14)  d  with the source function S assumed to be given by _  KB' +  crJ  S=  KB  + crJ  = K +  (T  ,„  = iB + (l-i)J X  »  '  3.15  Chapter 3. The Discrete System and Solution Procedure  59  where £ = K/XThe immediate difficulty is that J is not known a priori. However, since dp  Jo  (3.16)  this can be written in the discrete case as (3.17)  Wkd^kd,  fe=i where the coefficients Wkd are integration weights (to be determined) and the summation is taken over the number of impact rays I that intersect the shell of radius r . Then, for d  each frequency the discrete system of transfer equations (3.11) and boundary conditions (3.12)—(3.13), after substitution of the values of S and J given by equations (3.15) and (3.17), can be arranged into the following Rybicki block matrix form, / Ti  0  0  T  0  0  2  u  K  2  2  (3.18)  where  0  0  \ Wi  W  ••• T  7  • • • Wj  2  U E  uj  7  /  \ J  /  \0  /  = (tffei)«fe2j • • • Wfer») is the vector of length D representing the depth variation r  5  of the intensity Ukd along the impact ray k. Similarly, J = ( J , J , . . . , JD) describes T  x  2  the behaviour with depth of the mean intensity. The other quantities are coefficient terms derived from rearrangement of the spline equation. Here, the terms T*. and are tridiagonal matrices of dimension D x D,  is a D x D diagonal matrix containing  the quadrature weights Wkd, E = -I is a D x D diagonal matrix equal to the negative of the identity matrix, and K& are vectors of length D. The first I rows of this matrix system represent the transfer equations along each of the / impact rays while the last row describes the quadrature of the mean intensity J in terms of the individual intensities Uk along each ray.  Chapter 3. The Discrete System and Solution Procedure  60  In summary, the following nonzero elements are obtained. The subscript k (labelling each impact ray) has been omitted for clarity from the notation for the optical depth intervals, properly written as At d. The frequency subscripts have also been suppressed. k  = ^ ( A + Afcr) + i.' =  - A ^ - A -  (Tk) ,d-i  -  -At^Ut^x +1 (^Si)  (Tk)n  = 1 +^ +  (Tk)i2  =  d  (T )D D k  k  -  =  (CM**  = -|(i -  k  h  (^k,-i  k  h  (^fc)ll  (^)l2  = -|At!(l-6)  Dk  de[2,D -1]  d e %D -1]  =. -I (7^) (1 - 6-i), = -§A*i(l"6)  k  '  '•= - - j f a x ^ U - • & « ) .  d+1  {U ) D  fc  ^ + JA*I  (Tk)D D -l (u )d,  <*€[2,D -l]-  =  k  k  k  i(^)'  (Tk)wi  +  de[2,D -i]  fe  i € [2,D - 1] k  -|At _i(l  •=  k  <*e[2,D -i]  B|k  {Uk)D D -l = --f AiDj-^l - ^£» -l) k  (tf )i  k  fc  = Atxd^Bx + ^ O + Jfc  fc  (#fc)£>  = A i u ^ i d ^ ^ + l^-iBjn^-i) + v  (W )dd  = kdi  d e [r, Dfc]  ^  = -1, .  de [l,D ]  fc  k  Dk  w  k  (3.19) In the above, D refers to the index of the innermost radial shell intersected by the k  fc-th impact ray. If the ray intersects the core, then D = D, the total number of radius k  points. It should be noted that, although the matrices in the Rybicki construction are of dimension D x D, all elements with an index d > D are zero. k  The solution now proceeds by forward elimination, using thefc-throw to eliminate the corresponding element w d in the final row. After a total of I such steps, all the k  Chapter 3. The Discrete System and Solution Procedure  61  elements Wkd are gone, and the remaining single entry in the lower right hand corner is decoupled from the rest of the system. This leaves the matrix equation E'J = P',  (3.20)  where i  E'  =  -I-^WkT^Uk  (3.21)  fe=i I P  '  -.^WkT^Kk  =  (3.22)  which can be solved to get the mean intensity J. Back substitution into row K of the Rybicki block matrix then yields u = T ^ K - T^U J, k  k  k  (3.23)  and so all the specific intensities Uk along each impact ray have been found. 3.1.2  Quadrature Weights for the Moment Integrals  The remaining problem in the completion of the formal solution is the evaluation of the moment quadratures J , H and K. In general, these integrals are of the form Q =  [  n F(n)dfi  (3.24)  e\ . u _ if —n is even ., . v if n is oc )dd  (3-25)  n  n  Jo  where Qo = J , Qi = H, and Q2 = K, and  {  *V)H  They can be represented by the corresponding discrete formula 1  Q" =  S wF w  kd  1  =  knF ,  w  k  (3.26)  Chapter 3. The Discrete System and Solution Procedure  62  where I is the number of impact rays and the depth subscripts d have been suppressed for clarity. Initially, cubic splines were used to evaluate these integrals, and quadrature formulae were derived following the approach of HKK. However, splines proved not to be the best choice in this case for two reasons. First, the immediate advantage gained in interpolating a function which can be represented by a second order differential equation is not applicable. Second, and more importantly, a disadvantage of cubic splines became evident in use. Cubic splines, it will be recalled, have first derivatives determined globally and not locally. In practice, this often results in unphysical oscillations of the interpolating function, or 'ringing', in regions of large second derivative, due to the poorly constrained derivatives. The radiation field in an extended, spherically symmetric atmosphere becomes strongly peaked in the radially outward (ft = 1) direction at small optical depths. This led to instances of severe ringing in the spline interpolation of the intensities u(fi) for certain test examples. Therefore, the decision was made to abandon splines in favour of a more well-behaved interpolating function for the evaluation of the moment quadratures. Another cubic polynomial approximation is provided by cubic Hermite interpolation [60].' For a point.a; £  [ajj,JCi+i]  the cubic Hermite interpolating polynomial  Hz  is, for  a given set of function values {Fi} and the corresponding values of the first derivatives {Si} on the grid {a^}, defined as H (x) 3  = Fi<t>i(x) + Fi <t> (x) +1  i+1  + Si1>i(x) +  S tl>i (x) i+1  +1  (3.27)  where the basis functions <f>,ip are 1 - 3£ + 2£ 2  </>i+i( ) 3£ - 2£ x  ipi(x)  2  3  3  (3.28) (3.29) (3.30)  63  Chapter 3. The Discrete System and Solution Procedure  = &Xi(t* ~e) and where A  and £ = (x — Xi)/Axi. In particular, note that  4i(xi) = 1 , t  derivatives  H (x) 3  Si,S{ i +  V;(*i+i)  = 0  i+1  i+1  i+1  =0,  i>i{xi)  =0, tf&Zi) =0, <%{x )  <f>i(x )  = 0, <j> (x )  <f>i+i(x )  and therefore,  (3-31)  =  i+1  1, <^- (xi) +1  -=0,  = 0,  <t>' (x ) i+1  = 1 , 'il>i(x ) i+1  =  i+1  0  =0  correctly interpolates the specified function values  and  Fi,Fi i +  on the interval [xi,a:i i]. +  In the case of the moment quadrature problem, however, the first derivatives are not known a priori and must be estimated numerically. These derivatives can be conveniently estimated to second order by finite differences which, for an non-uniformly spaced grid {xi} of data points give Si =  + qiFi + n F  (3.32)  i + u  where AXJ  •  Vi = — i TT x Ax,-_ (Aa!^ + Axi) A  1  ,  (3.33) v  ;  1  Axi — Axj_i ri —  Ax^Axi-x  *  1  + Axi)  (3.35)  which reduces to the familiar 2-point centered difference formula for a constant grid spacing. The use of this method to supply first derivatives for interpolation using cubic Hermite polynomials is sometimes referred to as cubic Bessel interpolation. It provides a robust, (once) differentiate interpolation that, because of its strong local constraint on first derivatives, resists ringing and provides an interpolation that usually looks quite natural to the human eye. This interpolation, unlike splines, does not possess a continuous second  Chapter 3. The Discrete System and Solution Procedure  64  derivative, but that lack is of no consequence for our problem. Consequently, cubic Bessel interpolation was chosen to evaluate the moment quadratures for this study. The boundary derivatives Si and Si present a particular problem since they can not be evaluated by the above three point difference formula. For the moment quadrature problem we have used symmetry considerations where possible to provide these derivatives; otherwise, we have simply adopted the 'free-end' approximation of zero second derivative. At the p = 0 boundary, symmetry considerations imply F(0) = 0  for n even (J,K)  (3.36)  F"(0) = 0  for n odd (H)  (3.37)  At the fi = 1 boundary, no such constraints apply, and the 'free-end' boundary condition was chosen arbitrarily as being the most natural. This implies F"(l) = 0  for all  ra.  (3.38)  Then, in discrete form, we obtain directly S\• = 5  *  =  0  for n even (J,K)  f ( ^ T  1  ) - ^  Sr = I l^j£f)  for  » odd (H),  ~  (3.39)  ^ all n.  With this, all of the first derivatives {Sk},k = 1,...,/ can now be directly estimated from the given {F }. k  The moment quadrature Q , over the interval [pk,pk+i], can now be evaluated using k  the cubic Bessel interpolating polynomial to yield an expression of the form Q = (a F +b F )Ap k  k  k  k  k+1  k  + (c S + d Sk )Api k  k  k  +1  (3.40)  The values of the coefficients a , b , c and d depend upon the particular moment k  k  k  evaluated, and are listed below. J(n = 0) :  a .= +| k  k  Chapter 3. The Discrete System and Solution Procedure  h  H(n = 1) :  a  = + } •  k = +\l*k +  ^A/ife  h  = +\pk + ^A/x  c  =  fe  fe  -rjiPk + gTj^fe  d = -jtPk - ^Ap k  K(n = 2) :  k  Ofc = b  k  + -fiVkApk + Ts^l  = +|4 + ^fcA/Xfe + ^A/z*  Cfc = +^4 + YiPkApk + ^ A / i |  4 = -n/ * 4  The total quadrature  _  idVk^k - jo-Apl  1-1 fc=l  can therefore be written, after  e algebraic manipulation, in the form I  Chapter 3. The Discrete System and Solution Procedure  66  where the weights w are given by k  a A^ 2  + 6'iA/ii  2  a' Apz  +  3  ™fc = •{ a£. A / i +1  b' Afi2  +  2  c[Api  for k = 2  + 4_ Afik-i  fe^A/ife  +  f e + 1  for k = 1  x  +  6j_ A/t/_i 1  CJ_ A/ZJ_ 2  2  Cj_ A/f/_i  A/i _  +  d' _2  +  d'l-zApi-z  fc  fc  + d'i- Afii-  1  (  2  2  2  for 3 < A: < J - 2 forfc= I — 1 for k = I  (3.46) and fe  a  =  CkPkAfik  for 2 <  =  (cj_iP7_i + <Z/_itj)A///_i  < 7— 2  ?4  = a + (cfcg + 4pfe+i)A/i  forl<A;<J-l  fc  = fe + ( * * + d q i)Afi  forl<Aj<J-l  d'i  = (ci<i + c?ir )A/ti  d'fc  =  fc  fc  fe  c  c  fc  r  k  k+  k  (3.47)  2  dkrk+iAfik-  for 2 < fc < J — 2  The derivatives are estimated by S =p F .x h  k  + qF + rF  k  h  k  h  for2<fc<7-l,  k+1  (3.48)  where the coefficients p^, 5*. and r are given by the previously mentioned 3-point differk  ence formula, Pk = -n 9fe =  7 7 ^ 7  x—r  7  (3-49) (3.50)  ApkApk-i r  . =  ***-i  A  f 3 5  n  A/x^A/x^ + A ^ ) ' for 2 < k < I — 1. The boundary derivatives Si (for H) and Sx (for J , H, and if) can fe  ;  be put into an analogous form, Si =  qiFi + nF + tiF  (3.52)  Si =  tjF^+PiFj-i  (3.53)  2  3  + qjFj  Chapter 3. The Discrete System and Solution Procedure  67  with coefficients given by (3.54)  *  =  h  =  +  5.(^T-*)  ( 3  -  5 5 )  (3.56)  -|r,.  and ti •=•  »  (3.57)  =  -2te *-v  =  A  -  (3 58)  +  (-7^—  -ri-i).  (3.59)  Again, Fk = u for the evaluation of J and i f , and F = v = (du/dt) for the evaluation k  k  k  k  of H. With the moments J , If and i f in hand, the Eddington factors / = K/J can be directly evaluated for all radius and frequency points. Finally, from the definitions of u and v , k  k  1+ = u + v k  and the equations for J  k  (3.60)  k  and H , where k  J+  =  H+ =  \fl{p)dp Jo  (3.61)  \ f pl{p)dp J°  (3.62)  can also now be evaluated on the outer boundary, for all frequencies, by taking F = l£ k  with the n = 0 and n = 1 quadrature weights tu*. respectively. This yields the value of the second Eddington factor 4-  H+  s = 7+on the outer boundary, and the formal solution is complete.  ''  (- ) 3  63  Chapter 3. The Discrete System and Solution Procedure  3.1.3  68  Moment Equations: Cubic Spline Difference Formulae  The moment form of the equation of transfer in spherical geometry d  r  2  2  -j^(fuqur J ) dX* 2  v  = —(Ju ~ S ) q  (3.64)  M(t)  (3.65)  v  v  is also of the second order form dH dt 2  that can be naturally represented by a cubic spline. Therefore, making the following substitutions «d  At  d  M  d  =  (3.66)  fdqdTdJd  = AX*  (3.67)  =  (3.68)  Qd  -±(Jd-Sd)  into the spline difference equation (3.8), we immediately obtain the cubic spline finite difference representation of the moment transfer equation. Writing S = UB + (1 - U)Jd  (3.69)  Jd~ Sd = £d{Jd — B )  (3.70)  d  d  then, d  and M  d  qd  (Jd  - B ).  (3.71)  d  Upon rearrangement, this yields fd+iqd+ir  d+1  1r  d+1  £  d+1  a  AA  d  o  /d-i?d-i^d_i AXj_i  A  q  A  d  Jd+i — fdqdfd  d+1  1 d-ltd-l AXj_x 6 q -i r  d  Jd-l —  AX  d  + AXd.J  +  o  3 q  ^<J-l/2 d  Jd  Chapter 3. The Discrete System and Solution Procedure  1 r U+i 2  —7  d+1  o  69  ^ A Y n d-xJd-x ^^-d-tfd+i — - —i\Ad-i/2-t>d — <) ? d o q -i 2  1r  r  _  ZXAd-iXJd-i.  d  (3.72) Let XRM  a n  d  THAT  represent the Rosseland mean opacity and optical depth respectively.  Then the generalized monochromatic optical depth variable dX can be expressed in v  terms of the independent variable (the Rosseland mean optical depth) since dX = q dr = q ( v  v  v  v  J dr  RM  \XRMJ  = q \ dr v  v  (3.73)  RM  where the notation Xv = XV/XRM has been introduced to represent the normalized monochomatic opacity. Then, in discrete form, AX  = \(q d + qd+iXd+i)Ar  d  where A r j = Td+i  —  dX  (3.74)  d  r shall be taken to refer to the Rosseland mean optical depth. For d  the remainder of this section, we shall adopt the convention that x d a n  refer to the  r  respective Rosseland mean quantities unless explicitly specified otherwise. The discrete transfer equation then becomes fd+iqd+i (idXd + q +iXd+i) Ar d  - fdqd  1 d  &f+i  (qdX.d + qd+ix.d+i)AT 24 q  d+l  d  d+1  1  1  + . (q -xXd-\ + q X.d)AT . .(qdXd + qd+\Xd+\)AT d  d  d  d  1 Cd  (I'd \  2  J^ — [{qdXd + qd+iXd+i)Ar + {q -iXd-\ + qdX.d)AT -i] ^—J J fd-iqd-i 1 £d-i + qdXd)ATd-i^ ( ^ n ) Jd+ .(qd-iXd-i + qdXd)Ar -i 24 q 'Hqd-iXd-i . d  d  d  d  d  d  1 id+l - \A D -77 {qdXd + qd+iXd+i)AT [—-I B i f  d  - 12,K ^qd Xd +  9 d + i X « t + i ) ^ + (Gf-iXd-i +  i id  d+  qdXd)AT _ ] d  -(qd-iXd-i + qdXd^r^J^)  -i 2* qd  \ it y  x  [—  B_ d  u  )B \RJ  d  (3.75)  Chapter 3. The Discrete System and Solution Procedure  70  where R = r\ is the outer grid radius, and has been introduced as a reference radius for scaling purposes. In an exactly analogous manner, substitution into the spline boundary equations  (3.12) - (3.13) for du/dr  and M, and expanding the terms  and vp appearing there,  yields the cubic spline representation of the transfer equation boundary conditions. At the inner boundary, we obtain after some manipulation, 7  :  .{qD-iXD-i  -  1 63-1, [QD-IXD-I 24g _i  /r>-i£D-i  7 — .{qD-iXD-i  ~ rr  ;  + qDXDj&TD-!  1 £D,  ,  -  : VA + 77;—{VD-iXD-i + 9r>Xu)Ar _i 12 q  A  r  ,  - \*  - ~^{qj>-iXD-i  z  + gr>Xu)Ar ,_ £  I—— )  J V R J  1  1 ( D\  2  J_ D  X  2  ( — J JD = J V i? / T  B  D  1  + qDXDjAro-x  -{QD-IXD-I  1 /f\D-i\ 1  T  + gz>xz>)Ar ,_  D  D  q.  - x  + ?£)XD)Ar ,_  D  fDlD  24  .  777  J  B  D-I  D  (3.76)  1  128ir<rR T XD  \dT)  2  D  D  Similarly, at the outer boundary, we have  (qiXi + 92X2)Ar  x  +  .(9lXl + ?2X2)AT!  - 2i| '  1^|  ( 9 i  ^  i +  ( 9 1  ^  ^ ^ +  2 ) A r i  2 ) A r i +  ©  2 j B 2  \ \ G) = g t  -^| ^  -^[i- ^(- ^ )][G (d') e  T  lXl  2 Ji  (  i + ? 2  + gtGj(d')}  H  ^  (i)  A t i  Q) ^i, 2  2 j B i  (3.77)  where the extension parameter is ,1  h  d  IT*  =  rT(TTA)'  , ( 3 7 8 )  the scale height h on the outer boundary (neglecting radiation pressure) is given by  kT r\ x  h  ~  miGM.-L.xi/^cy  -  (3 79)  Chapter 3. The Discrete System and Solution Procedure  71  and pi refers to the mean mass per particle on the outer boundary. Finally, as before,  <"•> 3.1.4  The Hydrostatic Equilibrium and Radius Equations  Discrete Formulae for First Order Differential Equations Both the equation of hydrostatic equilibrium and the equation defining the radius depth scale are first order differential equations of the form  and can be represented to first order by the trapezoidal rule, namely  (^ )  = d  l  {fd  + f d + l )  ( 3 - 8 2 )  or, in an alternate form  ttrf+i = u + ^(/ + f )Ar d  d  d+1  (3.83)  d  where Ar = r ^ + i — r . Thus, the function / ( r ) is represented by a straight line on the d  d  interval'[Td,Td i]. +  A more accurate discretization of the above differential equation may be obtained by approximating  /(r)  by a cubic Hermite polynomial on [ r d , T d  +  i].  This representation  yields the discrete expression of third order, u i =u + i(/ d+  d  d  + f )Ar d+1  d  + ^  -( d r  J  Arl  ,  (3.84)  \ J +ii dT  d  d  where the derivatives df/dr may be approximated by the 3-point difference formula  fdf\ Pdfd-i+qdfd  =  + r f +i, d  d  (3.85)  Chapter 3. The Discrete System and Solution Procedure  72  with the coefficients pd, qd and r<f defined by  ^ A ^ T  = P  «* =  A /A^"  <-> 38 7  A  1  (3-88)  The Equation of Hydrostatic Equilibrium  The equation of hydrostatic equilibrium is written in differential form as dp =  ^dT-—^-dr  Xr* = where, again, both x  a n (  i  r  3.89  'VKcr  1  ^ { G M „ dm - ^  dAj ,  (3.90)  refer to the Rosseland mean quantities. Note that both terms  are exact differentials, except for the term 7 ( T ) and the factor 1/r , both of which vary 2  slowly with depth. The right hand side is exact for compact stars, and remains nearly so even for reasonably extended stars. Therefore, the simple trapezoidal rule, equation (3.83), should be sufficiently accurate. Thus, we obtain the following discrete form  (£)+(£) K-^-). >«) 2  / r>\  2'  where the reference radius R — r has been introduced for scaling purposes, and the x  variables, GM,  9 = dpn _ dr ~  -JJT £. (r) 47TC.R '  (- ) 3  92  ^  6 )  7  2  { 6  representing the gravity at the outer boundary and the radiation pressure term, respectively, have been defined. These variables have been introduced in this form since all three  Chapter 3. The Discrete System and Solution Procedure  73  (g, R, and dp /dr) will remain invariant during the linearization step. The remaining R  problem now is to express A m in terms of the independent variable A r . Since dm  1  (3.94)  we can write, to third order, from equation(3.84) A m j  _  1  =  i"(_L 1) _, _ 1 f J _ (f) 2 \xd-i  _ i (fs)  Arj  +  12 Lx_! V ^ / d - i  XdJ  d  x3 V*v  d-i-  r  (3.95)  (  The higher order terms have been included here since the dependence of %(T) can become very large, e.g. in the hydrogen ionization zone, and the right hand side, 1/x, was found to exhibit significant variation over the grid interval [r^-i,^] in some instances. The first order expression was originally used, following M H , but it proved to be inaccurate. Letting which will also be considered invariant during the linearization step, we then have the final form of A m ,  J_\  Yd-i  Yd  1 / 1 + (3.97) 12 \Xd-i Xd, 2 \Xd-i Xd) Substitution of this result in equation (3.91) yields the final discrete form of the equation Am _ d  of hydrostatic equilibrium. The Hydrostatic Equation: Outer Boundary Condition  The discrete form of the hydrostatic equilibrium outer boundary condition is immediately obtained from equation (2.124), Pi =  ( l + A)  GM, X\ r\  or, equivalently, ri + ^ A \ \) — — ( Pi = (1 Xi V i g T l  where g = GM /R  2  t  L*KXi 47rcGM»(l + A)  TI  R  and R = r\ as before.  2  (3.98)  r  4TTCGM,(1 +A)  (3.99)  74  Chapter 3. The Discrete System and Solution Procedure  The Radius Equation The other first order differential equation in the model system is the radius equation, dr dr  (3.100)  -PX,  which defines the radius grid, where both % and r refer to their respective Rosseland mean quantities. To third order, from equation (3.84), the discrete representation is Ar _i = Upa-iXd-i + pdXd){rd-i - r ) d  d  12  dr Jd-i  \  dr )  {r _ - r f. d  x  t  d  (3.101)  Let H  (3.102)  = — ( 7T~ ) = reciprocal density scale height Pd \OrJ  d  d  Yd =  Xd Xd  (3.103)  \dTj/ / d  These will be assumed to remain invariant during the linearization step. Then, d dr (PX)  = pdXdH d  J  (3.104)  p x\ d Y  d  d  Furthermore, for an ideal gas, P-P p — fin —kT  (3.105)  where p is the mean mass per particle. With the above substitutions for d(px)/dr and p, the radius equation in discrete form becomes 1 /Pd-iPd-iXd-i  2  V  , PdPdXd\ (r -i - r )  +  d  d  Ti.  x  1 fPd-iPd-iXd-iH^x T .x 12  fi pdXdH T d  d  / ^ L i P L i X L i ^ - i +, H P X"Yd | (r _i - r f ' "\"^ 2  d  d  d  vd  d  d  d  kAr -x. d  (3.106)  The values of Hd and Y are obtained from second order differences using the 3d  point formula of equation (3.85) for interior grid points d = 2,..., D — 1. At the inner  Chapter 3. The Discrete System and Solution Procedure  75  boundary, the values of (dp/dr)r> and (dx/dr)r) are estimated by extrapolation of the available derivatives at D — 1 and D — 1/2, assuming constant second derivatives. This procedure yields the boundary derivative  and (dx/dr)i) is represented by an analogous formula. Then, Hp and YD follow immediately from their definitions. The corresponding values at the outer boundary are obtained by application of the isothermal, exponential atmosphere assumed to exist exterior to the outermost radius. The Rosseland mean optical depth at the outer boundary was previously shown to be  Mxi '  1  (3.108)  1 + A'  _  where  =-(¥)  (3.109) •  Xi \OpJi  and  H  (dp/dr)i  x  is the density scale height on the outer boundary. Then,  is immediately obtained. Also, by writing y  =  i  ^ Xdr  =  i / W  n  x \Or/drJ  and substituting the expressions for (dx/dr)i and (dr/dr)i derived assuming the isothermal, exponential boundary atmosphere approximation,  Chapter 3. The Discrete System and Solution Procedure  76  is obtained. The integration of dt/dz = — px along impact rays for the formal solution is performed using the same discrete equation (3.101), except that the radial coordinate r is replaced by the distance z along the impact ray. The inner boundary value of [d(px)/dz]i) is obtained in an analogous manner to the radial case, and the outer boundary value is similarly evaluated using the isothermal, exponential boundary approximation to yield (3.114) where, for this one case only, Xi refers to the monochromatic opacity and p\ to the impact ray angle coordinate. The Radius Equation: Outer Boundary Condition  Finally, the radius equation boundary condition is supplied at the outer boundary by r i = R = CR,  where £ is a scaling factor defined such that T(TQ) m  (3.115) We have chosen the reference  optical depth To = 1 for this models. 3.1.5  The Equation of Radiative Equilibrium  The remaining equations to be cast into discrete form are the various quadratures over frequency, needed both for the equation of radiative equilibrium and the transfer equation inner boundary condition as well as for various auxiliary quantities such as the Rosseland mean opacity. These quadratures were evaluated using the trapezoidal rule, which approximates the integrand by a piecewise-linear function. The opacities of stellar material, particularly at low temperatures, are quite discontinuous in frequency as will be the stellar radiation field present in such a medium. Higher order integration formulae should not be used in this case since they require continuity and differentiability  Chapter 3. The Discrete System and Solution Procedure  77  properties that the integrand may not possess. The trapezoidal rule yields the following frequency quadrature weights, |(a:i —  x)  for i — 1  2  Wi = < \{xi^ - x )  for 1< i < N  i+1  \(XN-I  — XN)  for  i —  (3.116)  N  where the integration is performed over the frequency grid {xi},i = 1,..., N and N is the number of frequency points. The equation of radiative equilibrium in discrete form is thus N  i=l  = 1,  N  (3.117)  where the depth subscript d has been suppressed for clarity. The Rosseland mean opacity XRM is defined as  ^  J~(dBJdT)dv XRM — »oo / xZ\dB /dT)dv Jo v  3  — Too 5 Tr / :\dB /dT)du Jo X  (3.118)  u  and this is evaluated in discrete form by N  XR  M  i=l = ^-  \ /t ——• / dB\  Wi  (3-119)  Note that the numerator in the discrete form is evaluated by numerical quadrature and not directly using the exact analytic expression. This is because the analytic result represents the quadrature over the full semi-infinite range from zero to infinity while the analogous numerical result corresponds, by necessity, to a finite range of integration extending from some lower cutoff frequency u i to an upper limit u . Using the ratio m  n  max  Chapter 3. The Discrete System and Solution Procedure  78  of expressions with differing limits of integration would introduce systematic errors in the Rosseland mean opacity calculated. It is more accurate to evaluate both the numerator and denominator over the same finite range of integration. The truncation errors then largely cancel out in the calculation of the ratio. A similar argument holds for the calculation of the flux-weighted mean opacity XH  x  -H~L  "" H  ( - °)  du  3  12  which is evaluated using the discrete expression  '  XH = ^~  (3.121)  where the total flux H has been calculated using the discrete sum H  = ]Ci=i iHi w  a n (  l  not the analytic expression H — L*/(167r r ). 2  3.2  2  The Linearization of the Radius  We seek to solve the discrete model system, consisting of the moment form of the radiative transfer equations, the equations of hydrostatic and radiative equilibrium, and the radius equation, for T, p, r and J as functions of r . v  RM  This system of equations is implicitly  nonlinear since the constitutive properties of the stellar material, the density p, the mean mass per particle p, the opacities «„ and Xv>  a n (  l thus the ratio £ = «/%, depend (often  strongly) upon the local temperature and pressure of the gas. However, the temperature T is constrained to depend upon the mean intensity J by the constraint of radiative v  equilibrium while J is in turn determined by p and x through the equation of transfer. v  To solve such a coupled nonlinear problem, we resort to the complete linearization method (CLM) of Auer and Mihalas [7]. All quantities are expressed in terms of the fundamental physical variables describing the state of the atmosphere in our model, T,  Chapter 3. The Discrete System and Solution Procedure  79  p, r, and J„, and the resulting system of equations is linearized in terms of the differential changes ST, Sp, Sr and SJ in these variables. For the variables that can not be directly V  expressed in this manner, the linearization proceeds by expansion of the differential changes in these variables in terms of first-order differences in the fundamental variables. Hence, we can express  * = (l)*(l)* r+  ?  '>  3 (2 3 i  and so on. All that matters is that the differential form of the equations be entirely expressed in terms of the differential changes ST, Sp, Sr and SJ . Since all of the discrete V  model equations involve relations between (at most) the immediately adjacent depth points, the linearized equations can be represented by (at most) tridiagonal matrices of dimension D x D, where D is the number of depth points. The entire linearized system can then be organized into a block matrix form, such as that proposed by Rybicki [83], and solved to yield the first order corrections ST, Sp, Sr and SJ . This will be described U  in more detail below. However, a serious difficulty is encountered in attempting to apply the C L M to a spherically symmetric geometry. This adaptation requires the linearization of the radius variable r, which does not eiiter into the plane-parallel formulation.  This is readily  accomplished for the explicit occurrences of the variable r in the model equations, but the presence of the sphericality function q (r) presents more difficulty. As will be recalled, v  the function q (r) allows the collapse of the original, cumbersome first moment equation v  (2.4).to the more convenient form of equation(2.8), analogous to that of the plane-parallel case. This function is an integral transform over the grid points interior to the present radius r and is defined by q v { r )  = (ZJrp) exp  /  '3/„ - 1\ dr ' 1  fu  (3.124)  Chapter 3. The Discrete System and Solution Procedure  where r i is an arbitrary reference radius, here taken to be r i m  n  m  80  n  = rr>, the inner-  most grid radius, and f is the Eddington factor. Unlike the rest of the model system, v  <lv( d) r  depends not just on the adjacent depth points  Td-\  and r^+i but on all the points  {r<i'}, d' < d interior to the one in question. It would appear that linearization of the transfer equation, which involves q (r), would therefore destroy the tridiagonality of the v  complete linearization method. This would render the approach computationally infeasible, due to the large number of full (as opposed to tridiagonal in the plane-parallel case) matrix solutions required for the solution of the entire system. This appears to be the reason behind the choice of a radius scale by MH in their pioneering work on the application of linearization methods to the solution of models with extended atmospheres. The choice of radius as the independent variable obviates the need for linearization of (^(r) since the radius scale is now held fixed. The price to be paid in choosing a radius scale is the destabilization of the equation of hydrostatic equilibrium, as described in section 2.2. In our experience, it proved to be impossible to converge any model resembling that of an actual cool giant star using the M H method with the radius as the independent variable. However, we were able to show that a direct linearization of q {r) was possible. The discovery of a direct method v  of linearizing q (r^) is of central importance in this study. Without this method, it would u  have been impossible to put models on an optical depth scale and it is unlikely that the convergence of such models would have been achieved. The key to the linearization of q(r) lies in the realization that terms involving q in the discrete form of both the transfer equations and their boundary conditions always occur in the form of a rational expression, with both the numerator and denominator a homogenous linear polynomial in q. Then, defining the ratios, a  d  =  *L = 3~ \ d  (3.125)  g.d-i  fa =  — = «d+i  (3-126)  Chapter 3. The Discrete System and Solution Procedure  81  it is clear upon inspection that the transfer equation and boundary conditions can be rewritten entirely in terms of the ratios a and B. When this is carried out, equation (3.75) can be rearranged to obtain the following form of the discrete transfer equation, v TTjtd+iiPdXd + Xd+i)Ar ffl' (PdXd + Xd+i)&T 24 f a  A  d  D  1  +  ( j\ j r  d  1 (d \ [7J&[(X<* + a<i+iX<i+i)Ar + (B _ d-i + X f)Ar _ ] ^—J J 12 r  d  +  d-i  lX  1  (Xd-l + OL Xd)A.T . D  d  d  -  -  <  '  d  '  2  1  1 /^d-lN  d  2  (^-g-J Jd-i  ^7&-i(Xd-i + adXd)^^!  - 77jtd+i(PdXd + Xd+i)Ar ( — - d+l B, \ R ) 24: d  /7"d\  1  fiUKXd  + a x +i)AT d+1  d  2  + (B _ d-i + Xd)Ar<i_i] ^ — J £  d  d  lX  1  fTd—l\  -^£d-i(Xd-i+«<iXd)AT _ D  1  d  2  #<f-i,  (^-^-J  (3.127) with the inner boundary condition now written in the form fo-i (XD-I  1  + O:DXD)AT ^X  24  D  fu (PD-IXD-I  +  £ D - I ( X D - I  +  + -T;:(D(8D-\XD-I XD)AT£,_I  12  1 - — Mfo-iXfl-i  <XDXD)AT -I D  +  XD)AT _X D  /^D \  ^—J  +XD)AT _ D  1  - ^6>-i(xz?-i + a c X i > ) £ - i 1287r^il r Xi3 A r  2  3  and the outer boundary condition as 1 TTjWiXi + X2)Ari  24  5ZJ  /I'D—1\  1  LCAxi + ^ A n  2  (^)z>'  (^^"J  2  (3.128)  Chapter 3. The Discrete System and Solution Procedure  82  A .(Xi+«2X2)Ari  - ^ 6 ( A x i + X )Ar 2  g)  1  2  £ - ^ i ( X i + «2X2)Ar 2  1  g)  2  -^l-exp(-r £ xO][GH(^0 + 5 i ^ ' ) ] ( ^ ) 5 . +  1  2  1  1  Bi  (3.129)  Note that q(r) no longer appears exphcitly in any of these transfer equations. Direct evaluation of the ratio B yields d  f*n / 3 / - l \ dr'~  <ld _ ( d+i\' qd+i \ r J r  f  d  (3.130)  The above integrand varies only slowly with r, and so a discrete representation using the trapezoidal rule was judged to be sufficiently accurate. This procedure yields the discrete analogue r +i r  2  d  d  and similarly a  d  ^)-H[(^)(^)](—)}• +  Both of these expressions are bidiagonal, and thus only three radius points are involved in the transfer equation. Therefore, the model system retains its tridiagonal nature. We thus have the explicit functional dependences <x = otd(r -urd)  (3.133)  (3d = 0d(rd,rd+i),  (3.134)  d  d  and thus can linearize the ratios a and B as follows 8a = d  ®  d  m  =  5r _!  ad  dr -  d  x  [^i)sr dr B  + fdrT ~ I 8r  d  (3.135)  d  d +  (^)Sr dr, d+i  J + 1  .  (3.136)  Chapter 3. The Discrete System and Solution Procedure  83  The values of these partial derivatives are listed below, dct dr -i da dr  d  +<*d  2  r -i Td  d  d  2  d  Old  r ( 3/<i 1 3/ _! 1 fdrl 2 /d-i^Li Zfa-x-l Td-l (Zfd- 1  rd  d  2  V  d  2  d  (3.137) (3.138)  fd-irl-x  V fdrj  rd+i (3/<i+i — 1 _^ 3/d — 1  r dr d/3d = +3d dr i Jd+l  fd+ir d+l  r_d (Zfd+i - 1 2 V fd^r f r\ 2  1  +  +1  d+  (3.139)  fdrl  2  d  d  +  (3.140)  d  This completes the linearization of the sphericality function q(r).  3.3  The Linearized Model Equations  The transfer equation (3.75) and boundary conditions (3.76)-(3.77), the equation of hydrostatic equilibrium (3.91) and boundary condition (3.99), the equation of radiative equilibrium (3.117), and the radius equation (3.106) and boundary condition (3.115) form a closed system of equations in the fundamental variables «7i,..., J/y, T, p, and r (where N is the number of frequencies). We chose to linearize j = (r/R) J instead of J since 2  n  n  n  ] is nearly constant for r <C 1, which reduces the coupling between variables. n  v  We now proceed with the solution using the complete linearization method (CLM) as follows. First, define the 'block' vector x = (ji,...,j ,T,p,r) , T  N  whose components j i , .  ,JN,T,p,r  (3.141)  are themselves vectors of dimension D (the number  of depths), and Jn  —  (jnl, • • • , jnD )  (3.142)  T = (Ti,.. .,T )  (3.143)  P  (3.144)  T  D  =  (PI,V,PD)~  r = (ri,...,r ) . T  B  (3.145)  Chapter 3. The Discrete System and Solution Procedure  84  Here the bold font is used to denote quantities which are vectors of dimension D, or matrices of dimension D x D. The system of model equations can then be compactly written as f(x) = 0  (3.146)  where each component of the vector function f, f =  (f  (3.147)  l, . . . ^XE.NjfREjfHEjfDE) ^ 2  T E i  represents a constituent equation of the system. In general, the exact solution x is not known. Instead, we have a trial estimate x(') obtained after i iterations, hopefully with f(x^) « 0.  (3.148)  An improved estimate x( ) can be obtained assuming i+1  f( (i+i)) i x  na  — x <C x and expanding  Taylor series about x ^ . Let A x  W  =  x  (*+i)  _ W  (3.149)  X  Then to first order •  f( ( >) = f ( x « + A x « ) = f ( x « )  / fit \ (*)  i+1  x  Ax«  (3.150)  which must vanish if x( ) = x, the exact solution. Therefore, the desired first order t+1  correction Ax'*' is found by solving / f)f\ AxW = 0.  f(x«)+f^j  (3.151)  Denoting the Jacobian matrix (df/dx) of linearized coefficients by R , and f(xW) by —b, the linearized model system after i iterations is R(0 A x  w  =b . (i)  (3.152)  Chapter 3. The Discrete System and Solution Procedure  85  The solution of this linear system yields the desired correction vector AxW. Iteration of this method will produce estimates that converge to the exact solution of the model equations, provided the initial estimate x^) is chosen sufficiently close to the exact solu1  tion. The linearized coefficients Rk„ are given by the partial derivative of the k-th equation (row) with respect to the component x of the vector x, n  R-kn = (^)  (3.153)  evaluated at the current value of x = x^'^. Each of these components are vectors of length D, and thus the coefficients Rk are matrices of dimension D x D, where D is the n  number of grid depths.  1  The block matrix R is not a full matrix. The transfer equation (3.75) for the n-th frequency, written more concisely as  (3.154)  . frE,„(x) = 0  depends only upon j , and T , p, and r. Therefore, n  fTE,n  = fTE,n(jn,T,p,r)  (3.155)  and the only nonvanishing contributions to the differential  «~ =- ( « • ( " r ) :  +  s t +  n  n  n  are  ("r) +* $ p  • =; T 6 j + V ST + V 6p + S Sr. n  6TTE,II  (3.156)  n  Furthermore, since ^frE.n (at depth d) depends only upon variables at depths d — 1, <f, and d+ 1, the coefficients T  n  U  n  - R, nn = R , n  N + 1  =(^  )  (3.157)  Chapter 3. The Discrete System and Solution Procedure  86  ( C^TE n  s  K„,N+3 -  -  n  y—fc-  are themselves tridiagonal matrices of order D. The equation of radiative equilibrium (3.117) has the functional dependence fR = f (ji,...,JN,T,p,r)  (3.158)  RE  E  and, therefore,  *--S("?)«- ("?)" (^)* ("f) +  +  +  f c  N  =  ~T w £ j + C6T + T>8p + ESr. n  (3.159)  n  n=l  In this case, the coefficients w„  =  C  =  RN+I,N+I  D  =  RN+1,N+2  E  =  R-N+l.N+3  RN+l,n  =  (If) C^RE  (3.160)  dT  are all diagonal matrices of order D. The equation of hydrostatic equilibrium has the dependence f  HE  = f (T,p,r)  (3.161)  HE  and thus  (w)"+(*)* (~?) +  =  AST + BSp + FSr  f c  (3.162)  Chapter 3. The Discrete System and Solution Procedure  87  where the coefficients A  =  RN + 2 . N + 1  —  B  =  R  =  F  =  RN+2,N+3  N  +  2  L  N  +  2  =  dT  (3.163)  )  0f„ dr  are bidiagonal matrices of order D. Similarly, for the radius, or depth, equation we have  (3.164)  f (T,p,r) DE  and 6fnF.  6T +  -  = Q£T + HSp + GSr  (3.165)  with coefficients  Q  —  RN+S,N+I  H  =  RN+3,N+2  G  =  RN+3,N+3  \dT J  (3.166)  m dp  that are bidiagonal matrices of order D. Therefore, the coefficient matrix R has the block matrix form  ... 0 '  0  R  =  0  T  0  0  • ••  2  Wx W  0  • T • •• W  N  2  N  2  Si \ s  N  SN  Vi  u  2  u  N  c  0  0  ••  0  A  0  0  ..  0  Q  v v D B H  2  E F G /  (3.167)  Chapter 3. The Discrete System and Solution Procedure  88  The differentials of the constitutive variables are expanded in terms of the fundamental variables; thus  ««  =  «.  -  ( 3  ( f ) ^ + ( f ) *  P  -  1 6 9 )  (3.170)  where /x refers to the mean particle mass of the stellar gas. The gas density p is not expanded in this manner but rather expressed in terms of p, and the other fundamental variables using the relation p = pp/kT. Since p varies through the atmosphere by only a factor of three while p ranges over several orders of magnitude, this further reduces the coupling between the equations and improves the condition of the numerical problem. The sphericality function q is expressed in terms of the ratios a and 8, as described v  previously, and the differentials of these quantities are expanded as  *<= (£r)*" (£K ...«*= (©M^r)*+  <> <> 1 3 7 1  2 3 7 1  The right-hand side vector b in equation (3.152) has the components  b = (K ,...,K ,M,L,P) 1  N  r  (3.173)  where K  n  = • -f B,n(x«) T  (3.174)  M = -f (x«)  (3.175)  L = -f (xW)  (3.176)  P. = -f (x ).  (3.177)  M  HE  (i)  DE  Chapter 3. The Discrete System and Solution Procedure  89  The linearized system (3.152) can thus be expanded into the block matrix form  IT  x  0  0  T  0  0  2  w, w 0  0  yo0  o0  0  Ui  V  x  0  U  V  2  T W  2  ...  N  0  S  / Ajx Ajx \ Aj  2  C  D  E  A  B  F  Ap  Q  H  G  N  V  S  N  N w  .  / K K  2  A jj A AT  U  N  2  Sx \  NN  r j • JA \  = =  K  \  x  3  N  (3.178)  M L  which is of the Rybicki type. The first N rows represent the radiative transfer equations at the N frequencies considered, and the last three rows represent the equation of radiative equilibrium, the equation of hydrostatic equilibrium, and the depth equation defining the radius scale, respectively. The actual solution procedure used normalized variables so that differences were relative rather than absolute. Define the scaled vectors ~ ojn K  -  (Sjm I- —  Sj \ J nD  JnX  \  (3.179)  JnD J  (3.180) Sp Si  SpiA  ( Ji 8  '' PD  (Srx  J  •M r  D  (3.181) (3.182)  )  and the corresponding scaled coefficients R-kn = ~Lk„X = n  5x  (3.183)  E  Then, the form of the block matrix equation (3.178) remains unchanged under this scaling transformation. For convenience of notation we will hereafter drop all tildes on the variables and coefficients, but it should be assumed that the above scaling has been applied. The complete definitions of the coefficients of the linearized system appearing in the matrix equation (3.178) are provided for reference in Appendix A .  Chapter 3. The Discrete System and Solution Procedure  90  This matrix system is solved using thera-throw, corresponding to the transfer equation for frequency n, to eliminate the W term in the row representing radiative equin  librium. After repeated application, all JV quantities W , n = 1,. ..,JV, are eHminated n  from the radiative equilibrium row. The lower right hand 3x3 block is now decoupled from the rest of the system and can be solved directly. This reduced block matrix system has the form  A  where  u  / M' D' E' \ / A T \ Ap = L B F H G / V ) / A  (3.184)  r  JV  c  =  (3.185)  C-^WnT" ^ 1  n=l JV  (3.186)  D' = D - ^ W . T ^ V o n=l JV  E' =  (3.187)  E - ^ W ^ S , , n=l JV  M' =  M - ^ W J "  1  (3.188)  ^ .  n=l  Matrix equation (3.184), of order 3D x 3D, can now be solved to obtain the corrections A T , Ap and A r for each depth. Finally, the values of Aj can be obtained by back n  substitution into row n of the full matrix equation (3.178) and solution of the resulting tridiagonal system. The updated values of T, p, r and J are now used to evaluate the necessary secondary v  quantities, such as p, fi,  and  as well as the revised sphericality function q . With v  this, the block matrix coefficients in equation (3.178) can be re-evaluated and the updated linear system solved for new corrections A T , Ap, Ar and A«/„. This procedure is iterated until convergence is attained, at which point a self-consistent solution T, p, r and J„ of the nonlinear model system has been found for the assumed value of the Eddington  Chapter 3. The Discrete System and Solution Procedure  91  factor /„. Then the impact parameter method of H K K is used again to yield the formal solution of the intensity -T„(/i), now on the updated radius grid {r }, d = 1,..., D, and the d  moments J„, H and K calculated by quadrature over the angle coordinate fi. Finally, v  v  a revised Eddington factor f — K j J is calculated, and another solution of the finite v  v  v  model system consistent with this value of f is again found by complete linearization. v  After several such iterations, the whole procedure should converge to yield values of T, p, r and J fully consistent with the Eddington factor f . This completes the solution of v  v  the spherically symmetric model atmosphere. 3.4  The Initial Solution  3.4.1  Introduction  The iterative procedure outlined in the previous section requires an initial solution before it can be applied. This is not trivial, since such an estimate must not be too far from the true solution if the iteration is to converge. To start the linearization, initial estimates of T(r), p(r) and r(r) are needed. Then, a formal solution of the equation of transfer will yield values of J and /„ for this initial v  atmospheric structure. The linearization step can then be started. After some experimentation, the following prescription was chosen to obtain an initial temperature profile T(r). First, the optical depth scale {r } is defined. Next, a trial outer d  radius r\ is denned, usually by scaling the nominal stellar radius i?» (which is a specified model parameter). Then, assuming a uniform disk model, the angle subtended by the stellar disk in the atmosphere is given by 6 = cos ft*, where -1  m  r > R* r < R. t  (3.189)  Chapter 3. The Discrete System and Solution Procedure  3.4.2  92  The boundary temperature  We assume a boundary temperature of the form Ti •= iT?„(l-P*)C(p.)  (3.190) i  where the factor 1 — //* represents the dilution of the radiation field due to extension. Two limiting cases are: 1. the compact (plane-parallel) limit.  Here /z* « 0, and the classical plane-parallel atmosphere has a boundary temperature of ^ = 1^9(0)  (3.191)  where q(r) is the Hopf function. Therefore, we can identify C(0) = q(0) = 1/V~  (3.192)  2. the point-source limit.  Now p*  1, and the temperature can be well represented by T? = \T*„(1-p.)  (3.193)  and so C(l) = f  (3.194)  A reasonable estimate of a general outer boundary temperature might then be obtained by linear interpolation between these two limiting cases. This yields C(p.) = lp + (l-p.)q(0), t  (3.195)  and a boundary temperature of  = 121,(1 - /».)[§**. + «(0)(l" - p.)}.  (3.196)  Chapter 3. The Discrete System and Solution Procedure  93  At depth, we must recover the grey atmosphere result T{TY=\T*„[T  (3.197)  + {T)}, Q  provided the optical depth scale used is the Rosseland mean. An expression that yields the above temperatures for both limits of r <C 1 and r  1 is then given by  T(r) = f l j , ( l - p.)[r + lp. + q(r)(l - p.)], 4  (3.198)  which has been adopted as the initial temperature profile in this study.  3.4.3 The boundary pressure The boundary pressure is obtained by integration inward from infinity of an exponential, isothermal atmosphere. This yields Pi = (1 +  T.GM. A)-—-j-  xi n  1 -  4TTCGM*(1 + A)  (3.199)  where  and both x  a n  d T refer to the Rosseland mean quantities. Since the opacity, Xi = x(~i,pi),  (3.201)  will generally also be a function of pressure, it is usually necessary to iterate the boundary pressure p\ until a value is found consistent with both constraints. A Newton-Raphson iteration loop is used in the current implementation. Convergence is rapid, normally occurring within three iterations. The initial pressure profile p(r) may then be found by integration of the equation of hydrostatic equilibrium inward through the atmosphere.  Chapter 3. The Discrete System and Solution Procedure  3.4.4  94  The boundary radius  The remaining integration to be done is the radius equation, %  = -PK  (- ) 3  202  which defines the radius scale {rj}, d — 1,..., D, where D is the number of depth points. We start with the outer boundary radius of r = r =$R<  (3.203)  1  where £ is a scale factor depending upon the extension of the atmosphere. Initially, we arbitrarily set £ = 1.01. The stellar radius i?« is not well defined for stars with extended atmospheres. Therefore, some (arbitrary) criterion is required to link the stellar radius (a fundamental model parameter) to the optical depth scale. Most workers have invoked a relation of the type, r(r ) = R„  (3.204)  0  where ro is a reference optical depth, to relate these two scales. This is the approach taken by MH, who use r = 2/3. It is the approach that we follow also, although we have 0  adopted ro = 1 instead. The coupled integration of r(r) and p(r) now proceeds inward until r(r ) is reached. 0  This is compared to the specified radius 2?*, and another iteration of the entire starting sequence is performed if the disagreement exceeds the specified tolerance. If another iteration is taken, the scaling ratio | is updated using a Newton-Raphson technique, a better estimate of r i , T\ and pi is made, and the integration performed again. Eventually, convergence is attained, and then the integration continues inward through the entire atmosphere. This yields the desired initial estimates of T(r), p(r) and r(r). A formal solution is now calculated to obtain J (r) and / I / ( T ) , and the linearization step can begin. v  Chapter 4  The Model Computations  4.1  Introduction  The previous chapters developed the spherically symmetric model atmosphere equations and the corresponding discrete system. Before these equations can be solved, the equation of state for a given atmospheric composition P = P(T,P)  (4.1)  and the opacity of the stellar material must be available. For our purpose, it suffices to determine the total opacity per gram of stellar material Xu = Xv(T,p)  and the ratio £„ =  K /XV V  (4.2)  of absorption to total opacity,  Z = UT,P) V  (4.3)  The following, sections summarize the treatment of the equation of state and of the stellar opacity. Model implementation details are then described and the results of calculations are presented. 4.2  The Equation of State  The equation of state of the stellar gas has been determined in this study using the method of Bennett [17] to solve the chemical equilibrium of the gas. Grain condensation was not considered.  95  Chapter 4. The Model Computations  96  In general, the determination of the equation of state of a cool, molecular gas requires the solution of a nonlinear system containing an equation for each element that contributes significantly to either the total gas pressure, the electron pressure, or the opacity. The accurate determination of the equilibrium partial pressure of a species essentially requires the determination of all species of greater or comparable abundance. Since some important sources of opacity are of only trace abundance, an accurate solution of the equation of state for cool atmospheres requires the inclusion of a large number of elements. Our treatment of the problem considers 25 elements and 105 chemical species (neutral atoms, ions, and molecules) formed from these elements. A complete list of these species is given in Table 4.1. A direct determination of the equation of state thus requires the iterative solution of a matrix of order 25, which would be quite costly for use in a model atmosphere code. The method of Bennett [17] introduces considerable economy into the solution through the division of the elements into the following three groups: 1. M a i n elements. These comprise the significant sources of pressure which are also involved in molecule formation. We have included the elements H, C, N, 0, S, and Si in this group. The most serious omission from this group is probably Ti, which depletes 0 by up to 2% through the formation of TiO. 2. Metal elements. These elements are not involved signficantly in the formation of molecules, but do contribute to the total pressure or the electron pressure. The metals are assumed to be present only in the neutral or singly ionized state and not associated into molecules. We included He, Ne, Na, Mg, Al, K , Ca, Fe, and Ni in this group. Doubly ionized C a  + +  strongly violates this assumption at the  very bottom of the atmosphere, but this neglect is of no consequence since the resulting effect on the total electron pressure is negligible. Probably the most significant violation is due to the depletion of Mg (by up to 2%) to form MgH at  Chapter 4. The Model Computations  97  low temperatures. 3. Minor elements. These include all the remaining elements which do not contribute significantly to either the total pressure or the electron pressure but may be responsible for significant sources of opacity. In this group we included the (somewhat arbitrary) list of elements CI, Sc, Ti, V, Cr, Mn, Co, Sr, Y , and Zr. Two reductions of the solution are now possible. The third group of minor elements are of such low abundance that they exert negligible influence on the abundances of elements in the first two groups, and on molecules formed among elements of these two groups. Therefore, the solution of the equation of state for the first two groups (the main elements and the metals) decouples from the solution of the third group (minor) elements, and thus may be solved independently of this third group. A further reduction of the problem can be achieved by lumping together all of the group (ii) metal elements into a single fictitious element, which we shall denote by the symbol Z. We will show that the equation of state problem can be solved using a reduced matrix system involving only the main group (i) elements, along with the fictitious metal Z, and the electron pressure. We shall first briefly describe the solution procedure for the full matrix of group (i) and (ii) elements, and then shall proceed to demonstrate the reduced solution method mentioned above. We use the following nomenclature, specific to this section. Let the index k refer to one of the elemental species included in the linearization and the index m refer to an arbitrary metal element in the neutral atomic state. Also, let K denote the number of elements included in the full linearization (the full matrix of group (i) and (ii) elements), K' denote the number of elements included in the reduced linearization (the matrix of the group (i) elements and the fictitious metal Z), M denote the number of metal elements, JV denote the total number of chemical species formed entirely from elements of the full set, and JV' denote the total number of chemical species formed  Chapter 4. The Model Computations  98  Table 4.1: Species Considered in Model Equation of State Element H  He C  Z  Log Abundance (Number) 1 12.00  2 6  10.84 8.72  N  7  8.98  0  8  8.96  Species containing this element H  He C N NO 0 C0 Sc0 Zr0 Ne Na Mg Al Si S TiS CI K Ca Sc Ti V Cr Mn Fe Co Ni Sr Y Zr 2  Ne Na Mg Al Si S  10 11 12 13 14 16  7.92 6.25 7.42 6.39 7.52 7.20  CI K Ca Sc Ti V Cr Mn Fe Co Ni Sr Y Zr  17 19 20 21 22 23 24 25 26 27 28 38 39 40  5.60 4.95 6.30 3.22 5.13 4.40 5.85 5.40 . 7.60 5.10 6.30 2.85 1.80 2.50  2 2  H+ NH SiH HCO He+ C+ CO HCO N+ HCN 0+ MgO TiO HCO Ne+ Na+ Mg+ A1+ Si+ S+ ciK+ Ca+ Sc+ Ti+ V+ Cr+ Mn+ Fe+ Co+ Ni+ Sr+ Y+ Zr+  H" NH HS MgOH  H N HS AlOH  H+ H 0 HCl CaOH  CH MgH CaH  C H A1H HCN  c  C . CS  CH SiC  C H SiC 2  CN HCN  NH  NH  3  CN  o A10 VO MgOH  OH SiO V0 AlOH  H 0 SO YO CaOH  NO CaO Y0  NaCl Mg++ A1H SiH HS  MgH AlO SiO HS  MgO A1C1 SiS SO  MgCl AlOH . SiC CS  HCl  NaCl  MgCl  A1C1  CaCl  Ca++ ScO TiO VO  CaH Sc0 TiS V0  CaO  CaCl  CaOH  YO ZrO  Y0 Zr0  2  2  C0 N  2  3  2  3  2  2  2  2  2  2  2  2  2  2  2  2  2  NH  2  2  2  2  CO ScO ZrO  MgOH SiC SiS  2  Chapter 4. The Model Computations  99  entirely from elements of the reduced set. We shall define thefictitiouspressure of an element k (denoted pi) to be the partial pressure that would result if all ions of the element and all atoms of the element bound in molecules were free to contribute to the partial pressure of the elemental species p . k  This is sometimes referred to as the pressure of the nuclei since it is basically arrived at by counting the nuclei of element k present in the gas. Similarly, the total fictitious pressure is the gas pressure that would be obtained if all molecules were fully dissociated into their constituent atoms, and the electron pressure was ignored. The derivation, of the abundance equations in the following section make use of the fact that the fractional abundance ct of element k is just (by definition) the ratio p /p*. We make the following k  k  further definitions: Pe  = electron pressure  Pk  = partial pressure of element in neutral atomic form  Pi  =.  partial pressure of chemical species i  Pi = fictitious partial pressure of element k P = total gas pressure *  V ru  - fictitious total gas pressure = total number of atoms in species i  ik  = number of atoms of element k in species i  li  = electronic charge of chemical species i  n  ct  k  fractional abundance of element k.  Note that  K  5>*  =  (4-4)  L  fc=i  We choose the pressures p ,k = k  and p as the K + 1 independent variables e  describing the state of the gas. Then the pressure pi of any other chemical species can  Chapter 4. The Model Computations  100  be expressed in terms of these variables via the equations of ionization and dissociation equilibrium as K l l p e  h=l  Here Ii is the ionization equilibrium constant relating the partial pressure i to that P  of the corresponding neutral species p°, and Ki is the dissociation equilibrium constant relating the partial pressure p° of the neutral (molecular) species to the partial pressures P  k, k = 1,..., K of its neutral atomic constituents. For neutral species Ij = 1 while for  single atoms Ki = 1. The important point for our purposes is that i has the functional P  dependence Pi = Pi(Pe,Pl,---,PK)-  (4.6)  For each of the K elements in groups (i) and (ii) we can write an abundance equation of the form N  nikPi  ^UiPi *=1  by counting the atoms distributed among the various species. Upon rearrangement we have the abundance equation, N  =0  (a ni - n ) i k  ik  P  for  k  = 1,..., K,  (4.8)  i=l  of which only K — 1 are linearly independent, due to equation (4.4) above. The charge neutrality of the gas yields another relation, N  Y J P ^ - P ^  0  -  (-) 4  9  t=l  The sum of all the partial pressures must give the total gas pressure and this constraint yields the final equation of the system, N  "^^+Pe=P i=l  (4.10)  Chapter 4. The Model Computations  101  This set of K + 1 equations in the K + 1 variables p ,pi, • • • ,PK forms a closed e  system which can be solved for the elemental pressures. We proceed by linearizing the above system of equations, solving the resulting linear system to obtain the first-order corrections 5p ,6px,... ,Spx, and iterating until convergence is attained. The solution e  timing is fairly slow, however, since the coefficient matrix is of order K X K; in our case K = 15. To economize this solution, we consider only the main elements of group (i) above, and we lump the metal elements of group (ii) together and treat them as the single fictitious metal element Z. This yields the following reduced system of equations. Let the fractional abundance of Z be denoted ct . Then, the abundance of Z is just z  the sum of the abundances of the individual metals, or M  m=l  But we also have a = p  (4.12)  z  where p* = p + pf and thus z  M  M  M  ?;=p*]C ™ E " ^ = E ^ a  m=l  =  a  m=l  (- ) 4 13  m=l  The reduced abundance equation for element k becomes ^ nikPi  = $ =-sr  <« > 4  »=i or JV'  ^(a ni k  i=l  - n )pi + a p* = ik  k  z  0.  (4.15)  Chapter 4. The Model Computations  102  Since i = Pi(p ,P\i • • • ,PK') this equation can be linearized in terms of the independent P  e  variables p ,Pi, • • • ,PK',P* - We adopt p*, instead of p , as our additional independent e  Z  z  variable since this results in simpler equations.  1  The abundance equation for the fictitious metal Z is a  (4.16)  A"  z  ~T UiPi + p*  z  i=l  or  JV'  a*~~>W-(l-a.)p;=0  (4.17)  The total pressure constraint for the reduced system is JV'  ~~>+P* + -« = °>  (4-18)  i=i  and the equation of charge neutrality becomes N'  X > « f c + P ^ - P e = 0.  ( - ) 4  19  1=1  We now need to express p+ in terms of our independent variable p*. Let 1+ be the equilibrium constant describing the ionization equilibrium m r=* m  + e~  +  (4.20)  where 4  = 4^  (4-21)  and therefore ft. = ^  .  .  (4-22)  m  By the definition of p^, we have P„  (l + ~)pi  (4-23)  Chapter 4. The Model Computations  103  and thus  * = nfe-ifnS;-  <-> 42 4  Therefore, the total fictitious metal ion pressure is M  M  *  T  M  _ L  rf = E r i = E ^ - ^ E F ^ m=l  m = l  J  m  +  P  m=l  e  (4  -  25)  m + Pe  and since  . _* = P. a = PI -^^p* ^ z  P*  (4.26)  «z  we finally obtain the desired expression *  T+  M  a. ~ ,ft+ V, Substitution of this result into equation (4.19) yields the reduced equation of charge neutrality p* s-^  a IZ  z  a  z  m=l  m  m  ^  in a form dependent only upon the independent variables pi = Pi(Pe,P\, • p ,P e  z  • • )PK>)  and  We now have a system of i f ' + 1 abundance equations, only K' of which are  independent (since YJ a + a = 1), and the two constraint equations (total pressure fc  k  z  and charge neutrality) for a total of K' + 2 equations in K' + 2 variables. This reduced system can now be solved by linearization and the coefficient matrix is only of order K' 4- 2, which in our case is 8. The individual metal partial pressures p can be recovered from pi, once the solution m  has been found, as follows:  A - < ^ - ^ ( £ ) - ( ^ ] r f , . where the value of p* from equation (4.26) has been substituted. Then, given  (4-29)  Chapter 4. The Model Computations  104  which follows from the definition of 7+ in equation (4.21), we obtain Pm =  r +  .  •  (4.31)  The final problem concerns the solution of the remaining group (iii) elements. In general, this would require another iterative solution of a similar set of nonlinear equations to those of group (i). However, we adopted the following assumptions which allow us to obtain a direct solution: • Group (iii) elements form molecules only with group (i) elements, and not among themselves. In practice, the heavy elements making up group (iii) almost never associate with each other and this assumption holds quite accurately. • Molecules involving group (iii) elements have only one atom of the respective element present in the molecule. This assumption is largely justified on empirical grounds; molecules violating this assumption may exist but would have very low abundances and are not observed in stellar spectra. The first assumption above decouples all the abundance equations for each group (iii) element. The second assumption ensures that each of these abundance equations is linear in the partial pressure p of each of the group (iii) elements. Since the values of the partial k  pressures for the group (i) and (ii) elements are already known, a direct solution of each group (iii) k is immediately obtained. P  v  •  For example, the equilibrium solution considers the following four species involving Ti: these are Ti, T i , TiO, and TiS. Then the abundance equation for Ti is given by +  'an =  p  =  ^z(PTi + Pn + PTio + Pns)  p  = I(  PRI +  ^ l M  P* \  .  (4.32)  v  ?5 P* V  +  Pe  ^ KTxO  +  » Pe  |  K  TiO  +  ^££) K iS T  +  (4.33) )  (4.34) K J TiS  Chapter 4. The Model Computations  105  and thus, PTi =  j  •  (4.35)  This completes the solution of the equation of state. Overall, the order of the linearized equation of state matrix has been.reduced from 25 to 15, and then to 8 by the introduction of the fictitious metal. This corresponds to a factor of approximately (25/8) ~ 30 saving in timing. A direct solution of the 3  remaining elements (i.e. those not included explicitly in the linearization) is immediately obtained once the linearization step has been solved. The solution obtained remains exact within the limits of the assumptions made. In practice, errors resulting from violations of these assumptions should be at most of order 1-2%. Test results comparing the full and reduced solutions typically showed much smaller differences. In any case, these errors are negligible compared with the accuracy of Tsuji's [97] polynomial fits to molecular equilibrium constants, which depend upon many poorly determined dissocation energies and, in some cases, uncertain partition functions.  4.3  The Opacity of the Stellar Gas  The standard bound-free and free-free opacities are included, in addition to bound-bound opacities for the CN, TiO, and H2O molecules. The continuous opacities are similar to those of Auman and Woodrow [11]. The opacities for CN and TiO are represented by straight means, while the H2O opacity was represented by either a straight or an harmonic mean. All molecular lines are assumed to be formed in pure absorption. The details of the opacities included are summarized in Table 4.2. Straight mean opacities do not accurately describe absorption due to molecular bands but suffice for the purpose of this study, which was to demonstrate the viability of the complete linearization method to cool, extended stellar atmospheres. In general, mean opacities will overestimate the true opacity since frequency intervals with low opacities  Chapter 4. The Model Computations  Table 4.2: Model Opacities Opacity  Reference  H bound-free, free-free  Mihalas [69]  H  bound-free, free-free Gingerich [41]  -  H e free-free -  polynomial fit of Carbon et al. [25] to the calculation of John [52], as quoted by Kurucz [64]  Hj free-free  Somerville [93]  H^" bound-free, free-free Kurucz [64] C bound-free  Henry [45], as quoted by Kurucz [64]  Na, K , Mg, Ca, Al, Si . bound-free  Auman and Woodrow [11]  H Rayleigh scattering  Dalgarno [34]  He Rayleigh scattering  Dalgarno [34]  H2 Rayleigh scattering  Dalgarno [34]  e~ Thomson scattering CN red (straight mean)  Johnson et al. [58]  TiO (straight mean)  Collins [33]  H2O (straight mean)  Auman [9]  H 0 (harmonic mean)  Auman [9]  2  Chapter 4. The Model Computations  107  that may carry significant flux ('flux windows') are given low weight. An alternative considered in this work was to use an harmonic mean opacity instead. This has the advantage that the flux windows will be given a high weight. Furthermore, the harmonic mean becomes exact at depth. However, in the optically thin limit, the cooling of the gas by emission in molecular bands is more accurately described by a straight mean than by an harmonic opacity. In reality, neither the straight mean nor the harmonic mean properly describe the radiative transfer in the molecular bands. A more accurate treatment of opacities would be provided by the use of opacity distribution functions or opacity sampling methods. This study did attempt to estimate the effect of the treatment of molecular opacities upon atmospheric structure for the H2O molecule, which provides the dominant opacity over much of the infrared region of the spectrum. Some models were calculated which varied only in the representation of the H2O opacity, using either a straight mean or an harmonic mean treatment.  4.4  Convection  The model equations described assume radiative.equilibrium, and the models to be discussed in section 4.7 were calculated under this assumption. However, late-type stars have convectively unstable regions in their outer layers. Convective energy transport dominates the radiative flux in the atmospheres of late-type main sequence stars. The efficiency of convective transport is generally much reduced in the atmospheres of latetype giant and supergiant stars due to the lower densities. Convection energy transport only becomes dominant at depth in the atmosphere. As long as the convective flux H  conv  is only significant in the deepest parts of the atmosphere with T of the atmosphere with T  RM  of convection.  RM  ^> 1, the outer parts  < 1 and the emergent flux will not be affected by the neglect  108  Chapter 4. The Model Computations  To check the importance of the (neglected) convective transport, the convective flux was calculated using the temperature structure of our radiative models after full convergence had been attained. If the convective flux obtained is small, then our neglect of convection in the model calculations is justified. However, regions where the convective flux is significant will be inaccurately described by our radiative models. The convective quantities were calculated using a slightly modified version of the standard local mixing length theory as described by Mihalas [69], and the necessary thermodynamic variables were calculated using the formulae given by Kurucz [64]. In particular, the specific heats were calculated using  where E is the internal energy of the gas, and p refers, as it does throughout this work, to the gas pressure. The total pressure p , ignoring turbulent pressure, is tot  Ptot  = P  +  Prai  (4.38)  with the radiation pressure given by p  Tai  = — / K„du. c Jo  (4.39)  The above formulae for the specific heats follow from Kurucz [64]. Kurucz assumed that (dp ad/dT) oc T , from which 4  T  p  (Sg?)  (4.40)  follows. However, this result is true only for optically thick regions of the atmosphere where the radiation field is locally controlled and thermodyamic equilibrium is approached. It also applies for homologous transformations of the entire atmosphere. In the optically  Chapter 4. The Model Computations  109  thin outer layers, the gas temperature T is almost entirely decoupled from the radiation field, and (dp /dT) is not well-defined. For localized optically thin perturbations, rai  (dp ad/dT) T  p  p  « 0. For this work, we adopted the formula  dT J  T  p  from which both of the above limits are recovered. Ignoring radiation pressure contributions, the adiabatic gradient is given by [64] d rad\  (dpX  P  ~  \dmp) s ~ PC  dT) \dp)  2  s  p  P  T  _  (dp  \6T  (4.42)  while the radiative gradient V#, which equals the actual gradient V for our radiative models, is  This was evaluated numerically using, second order 3-point differences. The atmosphere is unstable against convection when V.R > V ^ , according to the Schwarzschild criterion, in which case the standard mixing length theory is invoked to estimate the convective flux. Following Mihalas [69], we consider the gradient V s of a convective bubble and define a convective efficiency 7 by  ' - v T ^ '  (4  -  44)  We assume the radius of the (spherical) convective bubbles equals the mixing length I. For optically thick bubbles, we then adopt the estimate _ pC v p  where  T  R M > B  = XRMPI  .  conv  is the Rosseland mean optical thickness of the bubble, v  conv  . is the  mean convective velocity of the bubble, and cr is the Stefan-Boltzmann constant. For the case of optically thin bubbles, equation (6-285) of Mihalas [69] has been modified since  Chapter 4. The Model Computations  110  a better estimate of the volume emissivity of the bubble is given by AnABupp using the Planck mean opacity «p. This yields pC v p  7.*. =  where r  PiB  = Kppl.  1  conv  -J-fT—  (4-46)  Linear interpolation between these limiting cases gives the result,  _^  ib  (I±i^-) .  (4  .  47)  Again following Mihalas [69], define Q = l-  (4  (^T)=-M) \omi  Jp  v  D  —  /  -  48)  p  Energy considerations can be used to estimate the convective velocity. Assume that half of the work done on the convective bubble by buoyancy forces eventually ends up as kinetic energy of the element. Mihalas [69] then obtains v^„= -r\]\gQh{V-V ) l  (4.49)  B  where h = p t/pg is the pressure scale height and g = GM+/r is the local gravity. 2  to  The  mixing length parameter a  = l/h is not determined by the 'theory' and must  conv  be additionally supplied. Usually, an arbitrary value near unity is adopted; a value of a  conv  = 1.6 was used throughout this study. In terms of a , conv  TRM.B  =  PXRM<x h  (4.50)  T ,B  =  pKpa h  (4-51)  P  conv  cmv  and v  cm  = a y/lgQh(V cmv  Substitution of the above value for v pC a p  7  *  =  cmv  /!  .  -S^TW^MV  conv  -  V ).  into equation  ~ ' / l  - v )[ fl  +lr —  (4.52)  B  (4.47)  P I B  T  R M I B  gives  \  )  V - V = VB^V;  s  ( 4  -  5 3 )  Chapter 4. The Model Computations  111  and thus lev^r  (  3  r  _ \  PtB  pC OL p  =  (v - v ) - (v A  v) B  (4.54)  1,B /  conv  We obtain [(V-V^)-(V-VB)]  2  (4.55)  2D,  V - V  B  where 16y/2(rT pC a  3  p  (4.56)  I P,B  TRM,E  conv  This quadratic equation in V — Vjg has the solution V - V  B  = D + (V - V ) - i/[D + (V- V )] -(V 2  A  A  -v y. A  (4.57)  which determines the convective bubble gradient. Finally, the desired convectivefluxis given by H  conv  = ^-a C Tv (V cmvP  p  conv  - V ). B  (4.58)  The quantity H , in analogy to the radiative case, refers to the Eddington flux; the canv  physical convective flux is 47r H . conv  4.5  Convergence Criteria and Iterative Control  An iterative control procedure was used to make the solution of the linearized equations more robust. If a trial solution is close to the exact solution, thefirst-ordercorrection obtained from the linearized system is quite accurate, and repeated iteration will yield rapid (~ quadratic) convergence to the solution. However, less accurate estimates yield less accurate corrections, and sufficiently poor trial estimates that exceed the radius of convergence of the linearized equations will result in immediate divergence if iterated. The remedy employed in this study was to control the flow of the iteration of the linearized system depending upon the accuracy of the current solution estimate, instead of proceeding to iterate blindly. Clearly, it is desirable to retain estimates close to the exact  Chapter 4. The Model Computations  112  solution, while rejecting those further away. The exact solution is not known beforehand, of course, and therefore we found it useful to construct a scalar quantity which provides an estimate of the 'goodness' of a particular trial solution of the linearized system of equations. The choice of this 'goodness' estimator is quite arbitrary, but it should vanish for the exact solution and increase in a continous manner away from the solution. We adopted the magnitude of the maximum component of the normalized residual vector £ of the model equations as our goodness indicator. In the notation of section 3.3, the linearized system of equations is R Ax = b  (4.59)  where the correction vector is A x = ( A j i , . . . , A J N , A T , Ap, Ar). The components Aji, A T , Ap, and A r are themselves vectors of dimension D and contain the depth variation of these variables. Each row of the left-hand side matrix R of equation (4.59) refers to one of the transfer equations for frequencies i — l , . . . , i V , or to one of the constraint equations. The magnitude of the right-hand side vector b, when suitably normalized, provides an estimate of the goodness of the current trial solution x and is zero if, and only if, x is an exact solution of the model equations. Therefore, we define the residual of each (transfer or constraint) equation by the vector  - =p b  -  (4 60)  with norm  Then, the scalar value ||e|| =max{|e |, M  = 1,..., iV + 3, d = 1,..., £ }  (4.62)  was chosen as our goodness of solution indicator and used to control the iteration strategy.  Chapter 4. The Model Computations  113  Blind iteration of the linearized equations will result in divergence given sufficiently poor solution estimates (i.e. large ||e||) due to the neglect of significant higher-order terms. In this situation, the linear derivatives usually underestimate the actual changes, and thus the linear corrections obtained generally overestimate the required correction. A common strategy is to employ a line search retaining the direction of the linear correction vector while scaring the amplitude to minimize the goodness indicator ||e|| (cf. Fletcher [40]). However, line searching requires a relatively large number of evaluations of ||e||, and is impractical in our case where each such evaluation requires considerable computational effort. We adopted a simpler, but robust, strategy. If the goodness indicator for the i + 1 iteration estimate is approximately less than that of the previous iteration z, < c||e||«  (4.63)  then we proceed and update the solution estimate x( ) directly. Here c is a constant ,+1  chosen slightly greater than unity (we used c = 1.05) to prevent possible infinite loops. These could occur if the machine precision is reached before the (too stringent) convergence criterion is met, in which  A choice of c > 1 then allows an  escape from the convergence loop. If instead this iteration yields a worse result, H  ( m )  >c||e||«  (4-64)  then the linear correction vector Aa;W is successively multiplied by the scaling factor a, with 0 < a < 1, until I e ( x « + a " A x « ) || < || e ( x « ) | |  (4.65)  holds after n trials. The procedure just outlined is straight forward, but still somewhat inefficient in terms of overall computational effort. An initially poor trial solution can result in a series of  Chapter 4. The Model Computations  114  poor linear corrections, each requiring several of the above sub-iterations to correct. It proved economical to put global limits on the size of the linear corrections as well. For our models, user specified limits ATu , Apu , and Afum were used to define the initial m  m  scale factor . J AT 3 = min < —^ Q  Apum Afum 1 ,— ,— ,1 >  Um  ^Pmax  I L\l  max  where AT , Ap , max  max  and Af  max  <-*•! max  (4.66) I  are the components of maximum amplitude of the  respective correction vectors A T , A p , and A f . The limits were set by means of the ATHENA program commands SET D T M A X , SET D P M A X and SET D R M A X , which default to the value of 0.1. Note that it is not necessary to restrict AJ in this manner, n  since the transfer equations are already linear in J , and so any discrepancy between the n  linear correction and the actual nonlinear solution must arise from nonlinearities in T, p, and r. A judicious choice of these limits ensures that the correction vector A x remains small enough to ensure the validity of the linear estimate. In rare cases when the equations are very nonlinear, the scaling algorithm previously discussed may operate to further reduce the value of a. In summary, the sequence of solution estimates x  (i+i)  =  x  x  (i+i)  =  x  x  (»+i) =  x  (i+i)  =  (») ^ «  W  +  A x  +  a / 5 A x  (0  « + a /3Ax 2  x  x  W  +  a  (i)  (4.67)  «/3 W A x  continues until n is found for which equation (4.65) is satisfied. We used a = 0.35 to compute the models in this study. This iterative control procedure proved very successful at preventing initial divergence of the linearized equations, and at guiding the iteration into the linear regime where  Chapter 4. The Model Computations  115  quadratic convergence is eventually attained. As a final refinement, the value of the i; + 1 iteration estimate was actually updated using the expression xf + aAxf  if Ax^ > 0 (4.68)  xf exp ( aAxf /xf  )  if Axf < 0  for each depth component d of the correction vector. These two variants are equivalent to first-order, but the latter form has the advantage of always providing a positive estimate for s  t+1 d  ^. This prevents program aborts due to unphysical negative variables from  occurring before the iterative control strategy has a chance to work. The iteration of the linearized equations continues until either ||e|| is less than the specified tolerance, in which case convergence to the solution is assumed, or the number of iterations exceeds the maximum specified on the command. 4.6  The Production of Converged Models  The solution of the spherical symmetric model equations is implemented in the ATHENA computer code by means of a command language structure. A summary of the most important commands is given below. INIT  This command generates an initial estimate of T(r), p(r) and 7*(T) through the atmosphere.  SAVE  file  This saves the current model parameters, arrays and flags into the specified file for later use.  RESTORE  file  This restores a previously saved model from the specified file.  Chapter 4. The Model Computations  FORMAL  116  This command specifies that a formal solution of the spherically symmetric radiative transfer problem is to be performed. Given T(r), p(r) and r(r), the moments J , H , K and the Eddington factor f are generated. u  MOMENT n  u  v  v  Given /„, improved values of J (r), T(r), v  (T),  and r(r) are calculated by the linearization  P  of the full system of moment transfer equations, radiative and hydrostatic equilibrium, and the radius equation. The maximum number of iterations to be attempted is given by ra.  CONVERGE m ra  This command automates the FORMAL + MOMENT loop, and will perform a maximum of m FORMAL solutions, each allowing a maximum of ra MOMENT iterations.  PTCORR ra  Given  J (T), v  /„, and r(r), improved  values of T(r) and (T) are calculated by P  linearization of the equations of radiative and hydrostatic equilibrium. The integer n specifies the maximum number of iterations to be attempted.  JTCORR n  Given (T), f„, and r(r), improved values P  of J (T) and T(r) are calculated by linearization V  of the moment equations of transfer and the equation  Chapter 4. The Model Computations  117  of radiative equilibrium. Again, n specifies the maximum number of iterations to be attempted.  SET flag value  Sets the value of the specified flag or variable.  OUTPUT  Produces complete output tables for the current model.  The actual solution procedure varies considerably depending upon the model details. The inclusion of molecular opacities was found to substantially increase the amount of effort required to converge models. For example, a series of models were calculated at effective temperatures of 3500K, 4500K, and 5500K for a variety of gravities and extensions. These included all of the usual continuous bound-free and free-free sources of opacity, but no molecular opacities. The command sequence INIT FORMAL CONVERGE 8 8 was sufficient to yield convergence for these models. Typically, convergence occurred within 5 FORMAL iterations, and with a maximum of 6 (much less for the later iterations) MOMENT iterations per FORMAL solution loop. Models with molecular opacities proved to be more difficult. In this case, iterations of the completely linearized moment equations often failed to converge, probably due to perturbations in the pressure scale arising from the nonlinear behaviour of the opacity XRM(T)  of the molecular gas. Convergence was obtained by the use of a sequential partial  linearization of pairs of the variables in turn. First, radiative and hydrostatic equilibrium were enforced by linearizing T and p for a given radiation field «/„ and radial structure r. This step was performed by the PTCORR command and always converged readily. Next,  Chapter 4. The Model Computations  118  the transfer equations were solved subject to the constraint of radiative equilibrium by linearizing J and T while p and r were held fixed. This system is quite well-behaved v  since temperature-dependent opacities largely cancel out, provided the Rosseland mean optical depth scale is used as the independent variable for the transfer equations and radiative equilibrium is expressed in the form of equation (2.17). This linearization was implemented in the command JTCORR and proved to be considerably more robust than the complete linearization provided by the MOMENT command. With the above changes, it proved possible to converge all the models which were examined. Typically, the command sequence would be to do a FORMAL solution (to get a new J from T, p, r), followed by a PTCORR sequence (to improve the estimate v  of T, p given J , r), followed by one or more JTCORR iterations (to update «/„, T from v  p, r) and closing the loop with another FORMAL solution. This method of sequentially sweeping through the variables always yielded slow improvement as measured both by the residuals of the equations and by the amplitude of the changes. Eventually, when the solution estimate was sufficiently good, the iteration sequence could be reduced simply to a FORMAL solution followed by a full MOMENT iteration loop. The final solution was then attained using a CONVERGE command, which simply automates the FORMAL + MOMENT cycle. As an actual example, the command sequence to converge a model with parameters T /log g/M eff  bol  — 3000/0.0/-3.0, and including the molecular opacities of CN, TiO, and  H 0 , was as follows: 2  RESTORE previous model PTCORR 10 JTCORR 1 FORMAL PTCORR 10  Chapter 4. The Model Computations  119  JTCORR 1 FORMAL PTCORR 10 JTCORR 1 FORMAL PTCORR 10 JTCORR 1 JTCORR 2 JTCORR5 PTCORR 10 MOMENT 1 MOMENT 5 FORMAL PTCORR 10 MOMENT 6 FORMAL MOMENT 8 CONVERGE 5 .4 Another persistent problem occurred in the very opaque, far ultraviolet continuum which arises from the strong resonant bound-free transitions of the abundant metals. For such frequencies, the values of J„ can range over nearly 20 orders of magnitude through the atmosphere. This results in the cubic spline solution of the transfer equation becoming unstable and yielding oscillatory solutions for J„ in the outer regions of the atmosphere. In terms of direct effects upon the emergent flux or the temperature structure, these oscillations are totally negligible due to the extremely small magnitude  Chapter 4. The Model Computations  120  of J in the affected regions. However, oscillations of sufficient amplitude eventually rev  sult in zero values of J„, a consequence of updating very large negative changes in 8J  V  in accordance with equation (4.68). This causes the tridiagonal solution of the discrete transfer equation to fail due to the presence of a zero diagonal element. Two modifications were made to circumvent this problem. First, a flag was introduced to allow the user to indicate the choice of solution method in FORMAL. In addition to the previously implemented cubic spline method, the less accurate but more stable method of centered finite differences was provided as an alternative to solve the transfer equation. The latter method, in which second derivatives are simply replaced by their corresponding second order finite difference analogs, is much less apt to yield oscillatory solutions in regions of large gradients. Flagging the problematic frequencies in the ultraviolet then alleviated the oscillatory behaviour of the FORMAL solution. While this approach would also work in principle for the moment equations, an alternate solution was sought to avoid the large number of program changes required to replicate the linearized coefficients and implement the alternate finite difference method. The method actually used updated values of J , ordered by decreasing radius, and then v  passed these values through a 5-point median filter for each frequency. This removed the oscillations and zero values of J for all but the most extreme situations. In such cases, v  which typically occurred during initial attempts to converge the more difficult models, the calculation of another FORMAL solution remedied the difficulty.  4.7  Model Results and Discussion  A series of trial spherically symmetric model atmospheres were calculated using the method just described for a variety of different opacities. Initially, a large number of models were converged using analytic test opacities. A further series of models were then calculated assuming realistic continuous opacities on a 30 point frequency grid. Finally,  Chapter 4. The Model Computations  121  several physically realistic models that included molecular opacities were calculated, and the results of these models are presented in this section. Unless specifically stated otherwise, all further references to models shall refer to those described below and summarised in Table 4.4. These models included the standard continuous bound-free, free-free, and scattering opacity sources, and the molecular bound-bound opacities of CN, TiO and H2O. The opacities due to CN and TiO were included using straight means over wavelength intervals. The opacity due to H2O was included in some models using straight means, and in other models using harmonic means. A microturbulent velocity of 2.0 km s  -1  was  assumed for the calculation of the H2O opacity. A grid of 106 frequency points was used. This grid was chosen to represent the gross structure of the molecular opacities, and to adequately sample the Planck function at the high frequency end at the inner boundary and the low frequency tail at the outer boundary. Failure to provide sufficient frequency coverage yields errors in the frequency quadratures and thus affects the accuracy of the total luminosity. A Rosseland mean optical depth grid of 80 points spanning the range log r  RM  == —5  to +2 was initially used, along with a corresponding impact parameter (angle) grid of 30 rays. However, it was found that the more extended models were optically thick in the strongest molecular bands, even at the outer boundary. Therefore, an expanded grid of 87 depths covering log T  RM  = —6 to +2 was used for the extended models. These models  also used an expanded grid of 45 impact parameters. Models were computed using the IBM 3081/63 system at the University of British Columbia Computing Centre. This computer has a floating point speed of about 3 mflops (flop = floating point operation per second). Representative timings for the various commands are given in Table 4.3. The details of the calculated models are summarized in Table 4.4. Each model is characterized by the model parameters T /\og g/M eff  bol  (where  Chapter 4. The Model Computations  122  M is the stellar luminosity specified as a bolometric magnitude), and a brief description hol  of the molecular opacities included. An equivalent alternate set of stellar parameters would be i * , M*, R ; the stellar luminosity, mass and radius respectively. All abundances t  are solar and are taken from Allen [2]. A complete list of the model parameters is given in Table 4.5, including actual model radii  and effective temperatures T as denned €ff  below in the text. Model output tables are presented in Appendix II. The units of all physical quantities appearing in tables and figures in this section are given in the cgs system unless explicitly stated otherwise. Table 4.3: Command Timings on the IBM 3081 Command  CPU time (s)  FORMAL  390  PTCORR  (per iteration)  65  JTCORR  (per iteration)  104  MOMENT  (per iteration)  108  Grid size: 87. depths x 106 freqs. x 45 angles Models were iterated until the residual ||£|| denned in section 4.5 was < 2 x 10 . -5  Average timings for various commands executed on the IBM 3081 are given in Table 4.3. Overall, these models took approximately three hours of CPU time to converge. The stellar radius, defined to be R* = r{r = To = 1), was not constrained to remain at R* (the specified model radius) during convergence'of these models. As a consequence, the final R varies slightly from R* and T , the actual effective temperature, differs m  eff  accordingly from the nominal T . Values of T calculated from eff  eff  Chapter 4. The Model Computations  123  Table 4.4: Converged Models with Molecular Opacities Model ID  Model Parameters T /logg/M tff  Molecular Opacities  Model Grid Sizes Frequency  bol  Depth Angle  01310191  3000/0.0/-3.0  CN, TiO H 0(sm)  106  87  45  02310191  3000/2.0/-3.0  CN, TiO, H 0(sm)  106  87  45  03310191  3000/0.0/-3.0  CN, TiO, H 0(hm)  106  87  45  04310191  3000/1.0/-3.0  CN, TiO, H 0(hm)  106  80  30  05310191  3000/2.0/-3.0  CN, TiO, H 0(hm)  106  80  30  06310191  3500/1.0/-3.0  CN, TiO, H 0(hm)  106  80.  30  07310191  3500/2.0/-3.0  CN, TiO, H 0(hm)  106  80  30  M  2  2  2  2  2  2  2  Note: hm= harmonic mean, sm= straight mean are listed in Table 4.5. Similarly, the actual gravity log g differs slightly from the specified log g; these values are also given in Table 4.5 for the current models. Temperature profiles r(r  RAf  ) as a function of Rosseland mean optical depth for a  sample of these models are displayed in Figure 4.1. These profiles have been scaled by the factors T /T eff  cff  to correct for the drift away from the fiducial stellar parameters.  This diagram illustrates the expected lowering of the surface temperature of the extended 3000/0.0/-3.0 model as compared with the compact 3000/2.0/-3.0 model. This can be seen more clearly by reference to Figure 4.2 where the detailed temperature structure of the surface layers is shown on an expanded optical depth scale. The extended model  Chapter 4. The Model Computations  124  Table 4.5: Converged Models: Summary of Parameters Model ID  M*/M  R*/RQ  R*/Ro  01310191  0.612  1225.  129.5  124.8  3055.4  0.03  02310191  61.2  1225.  129.5  129.4  3000.1  2.00  03310191  0.612  1225.  129.5  126.6  3033.5  0.02  04310191  6.12  1225.  129.5  129.4  3000.1  1.00  05310191  61.2  1225.  129.5  129.4  3000.2  2.00  06310191  3.30  1225.  95.1  93.6  3529.0  1.01  95.1  95.1  3500.0  2.00  Q  ttt  logy  *  07310191  33.0  M =1.989 x l O  1225.  g, L =3.826 x l O  33  @  33  Q  ergs/cm /s 2  R =6.9599xlO cm, from Allen [2] 10  Q  01310191 (3000/0.0/-3.0) is about 150K cooler on the outer boundary than the corresponding compact model 02310191 (3000/2.0/-3.0). The effect of the differing treatment of the H 0 opacity on the temperature structure is seen by comparison of the model 2  03310191 (3000/0.0/-3.0, harmonic mean H 0 ) with model 01310191 (3000/0.0/-3.0, 2  straight mean H 0 ) . Temperature differences of up to ~ 100K appear near optical depth 2  log  = —1.5 where the number density of H 0 reaches its maximum. 2  The pressure structure of the atmosphere is shown in Figure 4.3, where log P(T  RM  )  is  plotted for several models. Similarly, the behaviour of the electron pressure logp (r ) e  is shown in Figure 4.4.  HM  Chapter 4. The Model Computations  125  The atmospheric extension, defined by r(T )/R+ — 1, where R* is the actual stellar RM  radius, is plotted in Figure 4.5. The total extension of the atmosphere for the extended 3000/0.0/-3.0 models is about 0.25. By contrast, the 3000/2.0/-3.0 models are essentially compact, with an extension < 0.01. The Rosseland mean opacity profiles, XRM{ RM),  of selected models is shown in Figure  T  4.6. The oscillatory behaviour is due to the appearance of molecular opacities at various depths in the atmosphere. The prominent peak in XRM  near r  RM  = 2.5 (log T M R  = 0.4)  is due to the sharp maximum in number fraction of CN at this depth. The broad peak around logr  HAf  = —2 is due to the presence of TiO, which dissociates at depths greater  than this. The effect of H2O, abundant in the surface layers, is to extend the tail of this peak out to log TRM  = —4. The use of straight mean opacities in these models enhances  the prominence of these opacity peaks. These peaks would be lower in amplitude if a more accurate treatment of the molecular bound-bound opacities had been employed. An artifact is apparent in Figure 4.6 near log T  R  M  ~ —4 where the profiles of  XRM{TRM)  for the straight and harmonic mean opacities for the compact models intersect. The harmonic mean opacity should always be less than the corresponding straight mean. The problem is due to the choice of the four frequency points adjacent to the low frequency boundary. The opacity routines (in particular for the calculation of the opacity of H2O) truncate the calculation at wavelengths beyond ~ 10//. As a result, the calculated opacity for these four frequencies is very low. The Rosseland mean gives high weight to points of low opacity and so these four boundary frequencies can at times distort the calculation of the mean. Models with straight mean opacities are particularly susceptible due to the overall high opacity through the infrared region. It appears that the Rosseland mean calculation for the models using a straight mean H 0 opacity is affected above 2  log  = —3. At the outer boundary the Rosseland mean can be in error by 50%. The  net result is to introduce a small distortion of the outer portion of the optical depth scale  Chapter 4. The Model Computations  126  in the models using a straight mean H2O opacity. The Rosseland mean calculation for models using the harmonic mean H2O opacity is barely affected (the error is < 1% in XRM),  and the optical depth scale remains correct. The ratio of the flux-mean opacity XH to the Rosseland mean opacity is displayed in  Figure 4.7. This ratio, 7 =  appears as the multiplier in the radiation pressure  XH/XRM,  term of the equation of hydrostatic equlibrium. This ratio should approach unity at depth, as is confirmed by this figure. We have calculated the convective flux for our (radiative) models in the regions where the atmospheres are unstable against convection according to the Schwarzschild criterion. Figure 4.8 shows the resulting ratio of convective to total flux,  that would result  H /H, conv  if convection were allowed to develop in the radiative equilibrium (RE) models calculated. The corresponding mean convective velocities are shown in Figure 4.9. Typically, there is a convective region of variable extent present in the outer layers, between r  RM  1 0 - 1 0 . The convection is, however, extremely inefficient with -3  -1  H /H conv  ~  ~ 1 0 , and -6  the effect on the atmospheric structure is totally negligible. The onset of the primary convective zone occurs around T  ~ 1.  RM  For the low gravity (logy = 0) models, the convection does not become efficient until much deeper, with  > 0.01 only for  H /H conv  T  RM  >  15. This corresponds to a  monochromatic optical depth of r « 5 at the most transparent frequency (~ 1.9/t). v  Convection at this depth will have a negligible effect on the structure of the transparent regions of the atmosphere, and on the emergent flux distributions. In this case, we would expect that both the structure of the outer atmosphere ( T  R M  > 1) and the emergent flux  should be accurately described by our R E models. However, for the high gravity (log g = 2) models, the effect of convection does become (barely) significant. In this case, there is an outer peak with Deeper in the atmosphere,  H /H conv  ~ 0.2 for  H /H conv  falls again to ~ 0.03 near  r  RM  =  T  RM  ~ 5.  10, and then rises  Chapter 4. The Model Computations  127  steeply as in the low gravity case. The most transparent frequency at T  RM  = 5 occurs  at 1.74// for which r„ « 1.3 and exp(—r„) = 0.27. Therefore, the anticipated effect of the neglect of convection could be as large as 5% at the most transparent frequencies. Although, this is overall a small effect, it is clear that convection should be included in the model solution for accurate models in this case. The treatment of convection described previously essentially follows the standard local mixing-length theory as formulated by Mihalas [69]. This assumes that an optically thin, rising convective bubble has radiative losses proportional to KpAB,  where  Kp is the Planck mean opacity. The overall temperature dependence of the cooling is assumed to be that of AB = (4<r/7r)T AT. By contrast, the behaviour of the actual 3  cooling function $ j = J °° K B du, coo  Q  u  v  as illustrated in Figure 4.10, exhibits oscillatory  behaviour through most of the atmosphere. The destruction of sources of molecular opacity with increasing optical depth (and temperature) actually results in a decreasing cooling function over much of the atmosphere. However, the corresponding heating function  Eheat —  J °° K J du, 0  v  v  displays a matching temperature dependence. As a result, the  net cooling due to a temperature perturbation A T is still, to first order, given by KpAB, as assumed in the mixing-length model. Our models do not display evidence of the molecular cooling induced instabilities of the kind discussed by Muchmore [74]. The heating function appears to closely follow the cooling function (both are basically determined by the temperature structure of the opacity), so that the net radiative cooling remains a monotonically increasing function of temperature. Thus, our models are stable against such radiative instabilities for small temperature perturbations. The normalized residuals for the equation of radiative equilibrium (e ), the equation RB  of hydrostatic equilibrium (e ), and the radius (depth) equation (e ) are shown in HB  DB  Figures 4.11-4.13 rspectively. The rms normalized transfer equation residual, is defined  Chapter 4. The Model Computations  128  by 1/2  (4.70) where the summation is over the grid of N frequencies. The weights Wi used were  "i=(V^y  /4  (4.7i)  with J  The rms residuals e  TBrm  f B a  . = max{J }^.  (4.72)  i  , are presented in Figure 4.14. This particular choice of weighting  was used to ensure that e  TBTm  , was representative of frequencies where the radiation had  significant intensity. The residuals for all equations in the system were < 1 0  -6  throughout  the atmosphere for these models, demonstrating full convergence to the exact solution. The ratio of the actual luminosity L of the converged models to the prescribed model luminosity L* is presented in Figure 4.15. This quantity serves as a check on model convergence and overall consistency. In all cases, the ratio L/L+ is within ±1.5% of unity. The error appears to be dominated by quadrature truncation errors resulting from the numerical evaluation of the radiation field integrals on a finite frequency grid, instead of the semi-infinite interval [0, oo]. The 106 point frequency grid employed for the present series of models has its shortward boundary at 10.8/z = 926A, just longward -1  of the Lyman continuum edge. The quadrature error in the evaluation of the total flux H on this grid at the inner boundary of the atmosphere is also approximately 1%, mainly due to the lack of points in the Lyman continuum. The addition of such points, with associated extremely high opacities, would likely exacerbate the transfer equation instabilities noted in section 4.6. Therefore, frequencies in the Lyman continuum were deliberately excluded from the grid. The resulting errors in the evaluation of H impose a limit on the accuracy of the calculation of the luminosity L of about ± 1 % .  Chapter 4. The Model Computations  129  Figure 4.16 shows the emergent flux H for several models as a function of wavenumv  ber. A striking aspect of these flux distributions is the enhanced strength of the TiO bands for the extended models. Comparison of the flux shortward of the flux peak at 0.6/z for the extended 3000/0.0/-3.0 model and the compact 3000/2.0/-3.0 model shows -1  that the flux emitted by the extended model in this spectral region is similar to that of a compact model with effective temperature about 200K cooler. The 'photospheric radius' Rp evaluated as a function of wavenumber for the extended S  model 01310191 (3000/0^0/-3.0) is presented in Figure 4.17. Here Rp, is defined as the offset distance for which a ray drawn parallel to the symmetry axis of the star (i.e. an impact parameter as defined by H K K in their formal solution) has a total line-of-sight (monochromatic) optical depth of unity. The quantity Rp is indicative of the size of S  the resolved stellar disk, as seen by an observer at infinity. It is presented as a ratio to the model radius  This quantity provides an area in which direct observational  confirmation of atmospheric extension may be possible by means of stellar interferometric studies as reported by Beckers [14]. A series of greyscale representations of the full two-dimensional variation of several variables with optical depth and frequency are presented in Figures 4.18-4.20. These include the ratio of mean radiation intensity to the Planck function, J /B , v  v  the ratio  of the monochromatic opacity to the Rosseland mean opacity, XVIXRM, and the value of the Eddington factor /„. These figures permit a qualitative presentation of atmospheric structure in a very compact manner, which provides a useful diagnostic check. For example, it is evident at a glance that the radiation field thermalizes at depth («/„/B —> 1 v  and f  v  1/3) for all frequencies as expected.  The distribution of the main sources of opacity are shown in Figures 4.21 and 4.22 for the extended model 01310191 (3000/0.0/-3.0) and the compact model 02310191 (3000/2.0/-3.0) respectively. These figures display the log number density of significant  Chapter 4. The Model Computations  130  opacity sources as a function of Rosseland mean optical depth. These are presented as number densities since the strength of absorption features is determined by the number of absorbers along the line-of-sight. This can be obtained from these profiles by integration over optical depth. The ionization structure of the atmosphere for the same extended and compact models (01310191 and 02310191 respectively) is illustrated in Figures 4.23 and 4.24. The log number fraction of charged species are shown as a function of optical depth. Finally, the series of Figures 4.25-4.34 show the distribution of the abundant elements H, C, N, 0, and Si with Rosseland mean optical depth for these two models. These profiles are also represented in the form of log number fractions.  Chapter 4. The Model Computations  131  Figure Captions Figure 4.1. Temperature profiles T(r ) for selected models. Note: hm= harmonic RAf  mean, sm= straight mean. Figure 4.2. Temperature profiles T ( r  RAf  )/T  e//  for the outer atmospheres of selected  models. Figure 4.3. Pressure structure logp(r ) for selected models. Units are dynes/cm . 2  HM  Figure 4.4. Electron pressure logp (T ) for selected models. Units are dyne's/cm . 2  e  RM  Figure 4.5. Atmospheric extension r(r )/R RM  — 1 for selected models.  t  Figure 4.6. Rosseland mean opacity profiles log X R M { J  R  )  M  for selected models. Units  are cm /gm. 2  Figure 4.7. Ratio of flux-mean opacity to Rosseland mean opacity, log XH/ XSM, for selected models. Figure 4.8. Ratio of convective to total flux, log H (T )/H, CONV  for selected models.  RM  Figure 4.9. Profiles of mean convective velocities log v (r ) for selected models. conv  RM  Units are cm/s. Figure 4.10. Profiles of cooling functions log § OOI{T C  )  RM  = log J °° 0  KB V  v  dv for selected  models. Units are ergs/gm/s. Figure 4.11. Profiles of the normalized residuals e  RB  of the equation of radiative equi-  librium for selected models. Figure 4.12. Profiles of the normalized residuals e equilibrium for selected models.  HE  of the equation of hydrostatic  Chapter 4. The Model Computations  132  Figure 4.13. Profiles of the normalized residuals e  DB  of the radius (depth) equation for  selected models. Figure 4.14. Profiles of the RMS normalized residuals e  TETm  , of the equations of radia-  tive transfer for selected models. Figure 4.15. Ratio of actual to prescribed model luminosity, L/L„, for selected models. Figure 4.16. Emergent flux distributions log H for selected models. Units are ergs/cm /s/^ . 2  _1  v  Figure 4.17. Ratio of photospheric to model radii, Rp,(u)/Rt,  for extended model  01310191. Rp is a measure of the stellar disk radius as seen by a distant observer. S  See text for details. Figure 4.18. Ratio of the mean radiation intensity to the Planck function, log J jB , as v  a function of optical depth T  RM  u  for the extended model 01310191 (3000/0.0/-3.0).  On this 10-level greyscale, white (level 1) corresponds to a J /B v  v  value about 0.4,  the light grey background (level 4) to about 1, and black (level 10) to about 5. The value of J jB v  v  has been truncated at 5 to preserve dynamic range.  Figure 4.19. Ratio of the monochromatic opacity to the Rosseland mean opacity, log  XV/XRM,  as a function of optical depth TRM for the extended model 01310191 (3000/0.0/-3.0). On this 10-level greyscale, white (level 1) corresponds to  XV/XRM  values < 0.05,  the light grey dominant background at depth (level 4) to values near 1, and black (level 10) to values > 10 . 4  Figure 4.20. The Eddington factor /„ as a function of optical depth TRM  for the ex-  tended model 01310191 (3000/0.0/-3.0). On this 10-level greyscale, white (level 1) corresponds to f < 0.34 while black (level 10) corresponds to /„ > 0.65. The v  increment between levels is approximately 0.03 in f . v  Chapter 4. The Model Computations  133  Figure 4 . 2 1 . The distribution of opacity sources in the extended model 01310191. The log number density of each species is plotted as a function of optical depth r . RM  Units are c m . -3  Figure 4 . 2 2 . The distribution of opacity sources in the compact model 02310191. The log number density of each species is plotted as a function of optical depth r . RM  Units are c m . - 3  Figure 4 . 2 3 . The ionization structure of extended model 01310191. The log number fraction of each species with respect to the electron density is plotted as a function of optical depth T . RM  Figure 4 . 2 4 . The ionization structure of compact model 02310191. The log number fraction of each species with respect to the electron density is plotted as a function of optical depth T R ^ . Figure 4 . 2 5 . The distribution of hydrogen in extended model 01310191. The log number fraction of each species is plotted as a function of optical depth T  RM  .  Figure 4 . 2 6 . The distribution of hydrogen in compact model 02310191. The log number fraction of each species is plotted as a function of optical depth r . RM  Figure 4 . 2 7 . The distribution of carbon in extended model 01310191. The log number fraction of each species is plotted as a function of optical depth T  .  RM  Figure 4 . 2 8 . The distribution of carbon in compact model 02310191. The log number fraction of each species is plotted as a function of optical depth T . RM  Figure 4 . 2 9 . The distribution of oxygen in extended model 01310191. The log number fraction of each species is plotted as a function of optical depth r . RM  Chapter 4. The Model Computations  134  Figure 4.30. The distribution of oxygen in compact model 02310191. The log number fraction of each species is plotted as a function of optical depth r  R M  .  Figure 4.31. The distribution of nitrogen in extended model 01310191. The log number fraction of each species is plotted as a function of optical depth  TRM.  Figure 4.32. The distribution of nitrogen in compact model 02310191. The log number fraction of each species is plotted as a function of optical depth r . RM  Figure 4.33. The distribution of silicon in extended model 01310191. The log number fraction of each species is plotted as a function of optical depth r . RM  Figure 4.34. The distribution of silicon in compact model 02310191. The log number fraction of each species is plotted as a function of optical depth TJU^.  8f  ioooo  Models  I-I  0>  >-« o  8000 -  ST —  —  —  -  3000/0.0/-3.0 sm H 0  -  3000/2.0/-3.0 sm H 0  -  3000/0.0/-3.0 hm H 0  2  2  2  3500/2.0/-3.0 hm H 0 2  /  6000-  / /  4000  2000  0  -6  -2 log r  0 RM  t  Chapter 4. The Model Computations  136  CM i  o o o o I CM  CM  CM  CO  n:CM  I  E E E E CO en  o  O  x:  O  O  o  CO CO CO CO 1  1  o (D O O oi "D O O o o o o o  O o  i  CD o i  ^ »  O o o o o ID  CO CO CO CO  LO i  CD  O O O  o o  LO CO  O  o o  CO  O O  O O O  LO CM  Figure 4.2: Temperature profiles T ( r  CM  R M  O  o  LO  ) of the outer layers  i  d 6o| Figure 4.3: Pressure structure l o g ^ r ^ )  Figure 4.4: Electron pressure logp (r e  RJwr  )  Chapter 4. The Model Computations I  I  139  J_  CN  O O O CM o | W  CM  E E E CO CO -C CO CO  J/5  "D O  I  Q d o o o  o oi o o o  CO CO  CO 1  O  q d o o o  CO  CN CO  o  I  CD LO CN  o  b  O  CN  LO  •  •  o  LO  o O  o o  o o o  LO •  o o  i  •  Figure 4.5: Atmospheric extension r ( r  itM  )/i?» — 1  Chapter 4. The Model Computations  Figure 4.6: Rosseland mean opacity profiles log  XRM{TRM)  Chapter 4. The Model Computations  141  CN  o CM o CM o CM o CM o CM  e  X X X X  E E E E sz sz O O o o  CO CO  O CO CO CO CO CO _G0  CD "D O  1  O d  1  1  O o oi d  1  o o  oi oi  "•••^  Q o o o o O o o o o  O o o o LO CO CO CO CO CO  CN i  CO  o  I  CD CO  O  CN M  X  ,  H  i  X Bo|  Figure 4.7: Ratio of flux-mean opacity to Rosseland mean opacity \o%xnlx  Chapter 4. The Model Computations  CN  o X  o o CM  I  CM  X  o CM  W  CN  N  I  I  E E E E o o p in  CD  "a o  o  CO CO CO CO  p o o o o  o p  o  o o o o o o  o o o  c\i o c\i  I  CO CO CO CO  CO CN  CN  00  CD i  I  Figure 4.8: Ratio of convective to total flux log  H (r )/H  I  |_|/AU00|_|  I 60|  conv  RM  Chapter 4. The Model Computations I  L  CN  / I  \ O O O O X  CM  CD TD O  X  E  E  E  O  O  O  CO  CN  CM • X <* CM  X  CO  i  E  si  O  CO CO CO CO -«!^  o  1o  o  o  d  CM' d  oi  o o o  o o o  o o o  o o o  i  CO CO CO CO  CD CD  LO  CO  Auooyy  6o|  CN  i  Figure 4.9: Profiles of mean convective velocities log vcm^{rRM')  12.5 o o o  Models 3000/0.0/-3.0 sm H 0 2  12.0  3000/2.0/-3.0 sm H 0 2  3000/0.0/-3.0 hm H 0  era  2  (3  S  O  3000/2.0/-3.0 hm H 0  11.5  2  3 O  $  I  C  11.0 0 5  ^ c  ? O  era * —•  10.5-  ss  CO O  •e<  CO  O  10.0 9.5  -6  -4  -2 log r  0 RM  Chapter 4. The Model Computations I I  o o o CM  X X  CM  CN  E E E CO CO  O O O CO CO CO _C0 CO  O  1 1  o  d  "-•^  o o o  i  1 1  o  o C—N O O o o o o o  CO CO CO  CO I  00  I  o  CO I  I  3d  9 6o[  Figure 4.11: Normalized radiative equilibrium residuals e  T  Chapter 4. The Model Computations  O O  o  X  x  CM  CN  X  CM.  E E E CO  CO  q q  \  -C  q  CO co cd  / X/  I  I  I  CD P o p "D d oi d O o o o  o o o o  o o  CO CO CO  CO i  00  I  o  CN  I  I  3H  CO  9 6o|  Figure 4.12: Normalized hydrostatic equilibrium residuals e  Chapter 4. The Model Computations I  147  CN  o CM o CM O CM X  CN  X  X  I  E E E CO  CO  O O o  CO CO CO 1 1  _C0  o CD o O oi d  "O O  d  *">»«,  o o o o o o o o o  i  CO CO CO  CD CD i  00  CO  I  i  i  9 60| Figure 4.13: Normalized radius (depth) equation residuals e  z  i  CO O  Chapter 4. The Model Computations  I  o o CM  J  o CM CM  x x x E E E CO CO -C in CD "D O  p o p CO CO CO I  I  p o p  I  d oi d o o o o o o  o o o  CO CO CO  1 CO i  00  O i  SUJJ31  CM i  CO i  9 60|  Figure 4.14: RMS normalized transfer equation residuals e  a  Chapter 4. The Model Computations I '  o CM O CM o CM O CM o CM X  X  X X X  E E E E E CO  CO  si O O o O o  CO CO CO CO CO in CD  o  |  o  o  o  o  o  o o  o o o  o o o  O o  d  o o  CN d o o  CN CN  LO CO CO CO CO CO  CO  o  CN  O  O O  CD CO  Figure 4.15: Ratio of actual to prescribed model luminosity  CO CD  L/L,  CO  CO  "H  CD  ID  "tf  xn|j; ;U96J9UU3 6 O |  Figure 4.16: Emergent flux distributions log H  v  y /  sd  y snipej Ojjai|dsoioi|d psziieuuoN Figure 4.17: Ratio of photospheric to model radii  Rp,(v)/R*  Figure 4.18: Ratio of the mean intensity to the Planck function log J j'B v  v  Figure 4.19: Ratio of the monochromatic to the Rosseland mean opacity logXiz/x  Figure 4.20: The Eddington factor  f (r ) u  RM  Chapter 4. The Model Computations  CN  (Aijsuep  O  CN  "si"  i  i  UOJJ.09|8  CD  00  i  i  / Aijsuep jsqiunu) 6o|  Figure 4.23: The ionization structure of model 01310191  Chapter 4. The Model Computations  CD o :5 CO  +-•  o  o  l _ -J-»  **"^  oi  CO o o c o o CO "•*-» CO 0 N TD O o 2 •  •  CN  (Aiisuep  UOJIO9|8  / Aiisuep jeqiunu) 6o|  Figure 4.24: The ionization structure of model 02310191  Chapter 4. The Model Computations  159  CC CD O  CN  CN i  i  CD i  S9!09ds J;O uojioejj; J9quunu 6o|  00 I  Figure 4.25: The distribution of hydrogen in model 01310191  Chapter 4. The Model Computations 1  160  o  CN  CN  CO  03  s e p e d s jo uoi+yoejj. jeqwnu 6o| Figure 4.26: The distribution of hydrogen in model 02310191  Chapter 4. The Model Computations  161  CC  CO  o  CM  CM i  ssjOGds  i  CD  jo uojioejj. J9qtunu 6o|  00 I  Figure 4.27: The distribution of carbon in model 01310191  Chapter 4. The Model Computations  162  1  o  CN  CN  CO  00  sejoeds jo uojioejj. jequunu 6o| Figure 4.28: The distribution of carbon in model 021310191  Chapter 4. The Model Computations  CN  CN  SGjoeds jo  CO  00  uoiioejj jeqiunu 6o|  Figure 4.31: The distribution of nitrogen in model 01310191  Chapter 4. The Model Computations  co o  5 CN  ^ O  CM i  <sr i  CD  00  i  S9!09ds ±o uojjoejj jequunu  i  6o|  Figure 4.32: The distribution of nitrogen in model 02310191  Chapter 4. The Model Computations  167  co O Q ^  CN  O  CN i  ^ i  CO  00  i  i  S9!09ds j.o uojjoejj J9qujnu 6o| Figure 4.33: The distribution of silicon in model 01310191  Chapter 4. The Model Computations  168  co o 5.5 CM  O  CM i  ^J" i  CO  GO  i  sapeds jo uojioejj. jeqiunu 6o| Figure 4.34: The distribution of silicon in model 02310191  i  Chapter 5  Conclusions  5.1  Discussion  5.1.1  The Complete Linearization Method: A n Overview  The development of the solution procedure and implementation of the ATHENA computer code for the modelling of extended atmospheres of cool stars using the complete linearization method (CLM) has been demonstrated. This task actually involved two problems: the successful .application of the C L M to late-type atmospheres where molecular opacities are important; and the modification of the plane-parallel C L M to a spherically symmetric geometry. Both aspects of this project were significant since this work represents, to the best of our knowledge, the first published application of the CLM to any late-type model atmosphere, either plane-parallel or spherically symmetric. Additionally, few spherically symmetric (SS) models have been produced at all for late-type atmospheres. The use of the C L M brings many advantages. It is fundamentally not just a temperature correction procedure, although this does form the core of the work for the current LTE models. However, the generality of the CLM permits a straightforward extension to non-LTE problems where the equations of statistical equilibrium must additionally be solved to obtain level populations. The CLM could also be used to produce fully selfconsistent coupled stellar atmosphere and (steady-state) wind models. This would require the addition of a velocity variable to the model system and a replacement of hydrostatic equilibrium by an equation describing momentum transfer in the stellar material.  169  Chapter 5. Conclusions  170  The great power of the C L M derives from its simplicity. The underlying physics remains clear, which is often not the case with more formal methods of solution. Consequently, the introduction of extended physics (such as non-LTE) into an existing model proceeds in an obvious manner. In this study, the C L M has been implemented in spherical symmetry using an optical depth scale for the independent variable, as is standard for plane-parallel models. This implementation has been achieved by successfully linearizing the radius variable in a manner which does not destabilize the transfer equation. The spherically symmetric implementation described in this work has the further advantage of reducing to the corresponding plane-parallel model equations in the compact atmosphere limit. In this limit of zero atmospheric extent, the depth equation defining the radius scale becomes irrelevent and decouples from the rest of the system while 17(7*) —» 1 and the other equations reduce exactly to the Auer and Mihalas [7] implementation of the C L M . This greatly simplifies differential studies of extension effects since the same model code can be used to compute both compact and extended model analogs. This eliminates possible artifacts due to model inconsistencies in the different cases. 5.1.2  The Effects of Convection  The models in this study were all calculated assuming radiative equilibrium (RE). However, the effect of convection was estimated by calculating the convective flux that would result given the (fixed) temperature structure of these radiative models. For the low gravity log g = 0 models, convection was found to be significant only at depths with TRM  > 15. Therefore, the temperature structure of the outer atmosphere ( T  R M  < 5) and  the emergent flux should be accurately described by our current RE models. However, for the higher gravity logg = 2 models, convection does become marginally significant with an expected contribution of up to ~ 5% of the emergent flux at the most transparent  171  Chapter 5. Conclusions  frequencies. A consistent treatment of convection should be included in the solution to obtain accurate models in this case. At depth, the incorporation of a consistent treatment of convection would reduce the temperature gradient of the models. As a result, some artifacts occur due to the unrealistically large temperature gradient at depth. The most serious of these is a large radiation pressure gradient at the very bottom of the atmosphere, which for the two extended models calculated exceeds the Eddington limit at depths immediately adjacent to the inner boundary. 5.1.3  The Treatment of the H 0 Opacity 2  Our models used the approximation of mean opacities to represent molecular boundbound absorption. Water vapour absorption was modelled using both straight mean opacities, which more accurately represent thermal emission in the surface layers, and harmonic mean opacities, which are exact at depth. A more detailed treatment of the H 0 2  opacity is needed for accurate models. Differential comparisons of straight mean with harmonic mean H 0 models show that the greatest change in temperature structure (~ 2  120K) occurs near l o g r  HM  —1.5, where the number density of H 0 is near maximum. 2  The emergent flux shows little overall dependence upon the treatment of the H 0 opacity, 2  except in the regions of the strong infrared bands. 5.1.4  The Independent Variable Dilemma  Model calculations were attempted using both column mass and Rosseland mean optical depth scales. The column mass models converged well as long as the opacity was not overly temperature sensitive. However, when the opacity was a strong function of temperature, column mass based models inevitably suffered wild temperature oscillations at depth when the model solution was iterated. Basically this is because the transfer  Chapter 5. Conclusions  172  equation couples the radiation intensity to the optical depth in the gas; for a column mass scale at depth, neither the intensity nor the optical depth are known and therefore any iterative improvement will end up varying both (which is disastrous). Models using a Rosseland mean optical depth scale have the enormous advantage in that the temperature at depth is a known function of the independent variable. Furthermore, at depth the transfer of radiation is described by the diffusion equation, and this yields the mean intensity directly in terms of the function B (T) of the (known) temU  perature. Therefore, the use of the Rosseland mean depth scale fixes the temperature structure and the intensity of the radiation field at the bottom of the atmosphere leaving only the outer boundary to be iterated to convergence. Still, the outer boundary is troublesome. The Rosseland mean depends critically upon details such as the frequency points included in the grid and the treatment of the molecular opacities. This sensitivity of the Rosseland mean is shown clearly by the error introduced by an incorrect opacity calculation for four points at the extreme lowfrequency end of the grid of 106 points. Furthermore, since the outer atmosphere is optically thin at most frequencies the radiation field is not coupled to the gas and the optical depth is not a particularly useful variable. The physically meaningful variable in the outer atmosphere is the column mass, which is proportional to the pressure. On the other hand, the Rosseland mean optical depth is a very useful independent variable at depth. The dilemma of the modeller then is to reconcile these opposing constraints. One approach for future models may be to use the Rosseland mean of the continuum opacity sources only as an independent variable. This seems much less likely to cause trouble since the continuum bound-free and free-free opacities are much less temperature sensitive than are the molecular opacities presently included in the Rosseland mean calculation. Another promising idea is to construct a composite variable that approaches the Rosseland mean optical depth deep in the atmosphere, while acting like, a column  Chapter 5. Conclusions  173  mass for small optical depths. 5.1.5  Numerical Inconsistencies  The stellar luminosity was conserved to within ±1.5% in our models, as shown in Figure 4.15. While this may seem poor compared to other models that conserve flux to better than 0.1%, the flux and luminosity are secondary quantities in our models. Models that accurately conserve flux (cf. Scholz [88]) are generally designed to explicitly enforce flux constancy in their solution method. By contrast, fluxes do not appear in our model equations, with the sole exception of H supplied as an inner boundary condition. Instead, v  our models enforce the more basic equation of radiative equilibrium, which is mathematically equivalent to the conservation of flux (actually r H in the spherically symmetric 2  u  case), and our solutions yield mean intensities. Fluxes were obtained by differentiation of these mean intensities. The residuals of the transfer equations and radiative equilibrium were very small (~ 10 °) for our converged models, and it is clear our models accurately -1  solve the discrete systems. The implication is that the errors in luminosity arise from inconsistencies between the discrete and continuous model equations, and also possibly between the different discrete forms of the formal and moment solutions. It should be noted that the models of Scholz [88], while iterated to a flux constancy of 10~ , only obey radiative equilibrium 4  to within ±2%. Thus, the level of internal consistency of our models (~ 1.5%) is entirely comparable to that of the models of Scholz [88] and also of Bessell et al. [19]. There appear to be two effects present. First, examination of Figure 4.15 shows a slow rise in.L/L* of ~ 1% from the inner boundary outward to depths of T  R M  ~ 1. The  gentler rise for the 3500/2.0 model agrees very well with the profile of l/u>, where u> (ideally = 1) estimates the frequency quadrature error and is defined by (5.1) n=l  174  Chapter 5. Conclusions  with quadrature weights as defined by equation (3.116). Thus, for the 3500/2.0 model at least, the variation in L/L+ appears to be largely a result of quadrature errors due to the finite frequency grid. Our models used a frequency grid with a high frequency boundary at 10.8/z , just longward of the Lyman continuum edge. At the inner boundary, the -1  numerically integrated flux is underestimated by about 1%. The monochromatic fluxes supplied for the inner boundary condition are scaled to yield a total luminosity equal to the specified  Moving outward from this boundary, the flux spectral distribution shifts  to lower frequencies and better matches our grid so that the numerical quadrature errors become quite small. Therefore, we should expect the integrated flux to increase by ~ 1% over the range of optical depth log T  R M  = 2 to 0, as confirmed by the 3500/2.0 model.  However, the cooler 3000K models show a sharper, greater amplitude rise in L/L+ than can be accounted for by the variation in UJ alone. The second effect evident in Figure 4.15 is the sudden drop in luminosity of up to 2% around log r  RM  ~ —0.5, terminating the previous rise. The cause here is less obvious. The  possibility that molecules are responsible, though, is suggested for two reasons. First, the hotter 3500K model, for which molecular opacities should be less important than in the copier 3000K model, shows a much smaller drop in L/L+. Second, the drop for the higher gravity log 5 = 2 models occurs deeper in the atmosphere, nearer to log T  R M  = 0. This  echoes the shift in the distribution of molecular opacities with gravity seen in Figures 4.21-4.22 and particularly in Figure 4.10, where molecular opacities dominate the cooling function for log r  RM  < 0.5. A probable cause of this lack of luminosity conservation lies  in the finite difference form of the transfer equation. We use Ar = v  Xv  — — A T  R  M  =  (5.2)  XV& RM T  XRM  to represent the monochromatic optical depth step. Normally \  v  is a slowly varying  function of depth. However, in regions where molecular opacities first appear, the spectral distribution of the total opacity can change dramatically with depth, and then Xv may  Chapter 5. Conclusions  175  vary significantly over a single depth interval. Under these circumstances, the first order differencing used may become inaccurate, resulting in errors in the mean intensity and flux. It is also possible that rapid changes in the spectral characteristics of the opacity may be responsible for the larger variations in L/L* at depth in the 3000K models. At present this remains speculative; further testing is needed for confirmation. Another artifact affects the values of the Eddington factor f . This is seen in Figure v  4.20. At  2.0/i , fv _1  initially decreases inward from the outer boundary as expected and  approaches the diffusion limit value of 1/3, but then rises again to a plateau of about 0.38 near  logTRju  = 0 before finally thermalizing at logr  RM  ~ 1. Again it is likely that  the source lies with inaccuracies in the solution of the transfer equation in this, region. The problem occurs in regions where J„ is decreasing very steeply at rates up to 50% per depth point. The finite differencing is almost certainly inaccurate for such large changes over a depth interval. Since a similar finite difference equation is used to solve the transfer equation along impact rays in the formal solution, the values of I found here will be v  equally inaccurate. The result will be systematic errors in the moments J , H , and K u  v  v  calculated by the formal solution and, therefore, corresponding errors in the Eddington factor f as well. Further study is also required to substantiate this explanation. v  It seems possible, and certainly desirable, to reduce these internal inconsistencies to 0(1O~ ), and this will be given consideration in the immediate future. 3  5.2  Comparison with Other Models  A comparison of our models with the plane-parallel (PP) models of Brown et al. [23], the PP and spherically symmetric (SS) models of Scholz and Tsuji [89], and the PP and SS models of Schmid-Burgk et al. [87] is presented. The PP models of Brown et al. were calculated using a modified version of ATLAS [64] incorporating an opacitysampled treatment of molecular bands and convective energy transport using the local  Chapter 5. Conclusions  176  mixing length theory [64]. The models of Scholz and Tsuji were calculated using the code of Schmid-Burgk [86]. These models were calculated assuming radiative equilibrium and employed mean opacities for the molecular bands. Schmid-Burgk et al. calculated a variety of PP and SS models which included the molecular opacities of CH, CN, CO, OH, H2O, TiO, MgH, SiH and CaH calculated assuming the JOLA approximation of Tsuji [95]. They also calculated models with only the opacity of the H2O molecule included. All of the models of Schmid-Burgk et al. were calculated assuming radiative equilibrium. We present detailed comparisons of the temperature and pressure structure of our models, the models of Brown et al., and the Scholz and Tsuji models for two cases: T l log g =3000/0.0 and 3500/2.0. We also present the 3000/0.0 B0 (PP) and B3 (SS) etf  models of Schmid-Burgk et al. , with all molecular opacities included. The detailed temperature structure of the surface layers as a function of Rosseland mean optical depth is shown for the above 3000/0.0 models in Figure 5.1. The entire atmospheric temperature structure is shown in Figure 5.2 for the same models. The temperature profile of our model has been scaled by the factor T /T eff  eff  = 3055.4/3000  to compensate for the drift of the model radius J?» off the specified fiducial radius R . m  The models of Schmid-Burgk et al., originally on a 1.2p optical depth scale, have been put onto a Rosseland mean scale by using our T I . ^ T R M ) relation. The corresponding pressure profiles for these models are given in Figure 5.3, where the Scholz and Tsuji models (originally calculated with log g = —0.70) have been shifted by A log p = +0.70 to approximate the gravity of our models. Consequently, the Scholz and Tsuji temperature and pressure profiles should be interpreted with caution, since this scaling procedure will not be correct for pressure dependent opacity and equation of state changes. Our SS model is not directly comparable with that of Scholz and Tsuji; their model has about twice the atmospheric extension (~ 50%) of our 3000/0.0/-3.0 model. A better  Chapter 5. Conclusions  177  comparison with our model would be a temperature structure intermediate between the Scholz and Tsuji PP and SS curves. We do not obtain the pronounced drop in temperature near log T  RM  ~ —1.5 seen even in the PP Scholz and Tsuji model. The surface  layers of our model are much warmer (by ~ 600K) than the profile given by the average of the PP and SS models of Scholz and Tsuji [89] which we assume represents a model with comparable atmospheric extension (25%) to our model. Furthermore, the models of Scholz and Tsuji were calculated for a lower gravity than our log g = 0 and this makes comparison more ambiguous. However, our temperature profile is quite similar in shape, but ~ 100-200K cooler in the surface layers than the model of Brown et al. Our temperature profile is also close to that of the B3 SS model of Schmid-Burgk et al., a model which has a very similar atmospheric extension to our 3000/0.0/-3.0 model. The reduction in the temperature gradient at the very bottom of the atmosphere in the Brown et al. model is due to the onset of convection. The pressure structure of our model is in general agreement with the Brown et al. model, and also in reasonable agreement with the PP B3 model of Schmid-Burgk et al.. However, our model pressure is about 0.5 dex lower than these around logr The steep drop in pressure above  logTR  M  RM  ~ —1.5.  ~ —1.5 notable in the Scholz and Tsuji models  does not occur in our model. Again, the interpretation of these Scholz and Tsuji models is difficult because they were calculated for a considerably lower gravity. Analogous figures are presented for the 3500/2.0 models. Figure 5.4 shows the detailed temperature structure of the surface layers for our SS model, the PP model of Brown et al., and the models of Scholz and Tsuji (which were calculated for the substantially lower gravity, log g = —0.43). Schmid-Burgk et al. did not calculate models for a temperature of 3500K. As might be expected, the more compact 3500/2.0 models show less variation in temperature structure, with the Brown et al. and the Scholz and Tsuji PP and SS models deviating from each other by < 100K in the outer atmosphere. The temperature  Chapter 5. Conclusions  178  structure of the entire atmosphere is shown for the 3500/2.0 models in Figure 5.5. Our 3500/2.0 SS model has a temperature profile similar in shape to the other models, but consistently warmer by ~ 150K throughout the outer atmosphere ( l o g r ^ > 0). This is probably due to the omission of CO opacity in our model. The CO molecule is the dominant coolant of the surface layers in this temperature regime. Finally, the pressure profiles of our model and the PP model of Brown et al. are shown in Figure 5.6 and are seen to be in reasonable agreement. The models of Scholz and Tsuji were not included in this figure since these were calculated for the much lower gravity of log g = —0.43. 5.3  Future Research  5.3.1  Short Term Improvements  One of our first priorities will be the production and publication of a grid of spherically symmetric models. These will probably be pure hydrogen models including only a few continuous opacities (e.g. H~ and neutral H bound-free and free-free absorption, H Rayleigh scattering, and Thomson scattering). The intent is to provide a series of easily reproduced standard models that will facilitate the comparison of different model techniques.  ,  Another area of immediate concern is the equation of state calculation. The solution procedure described in Bennett [17] works well, but it has been implemented using Tsuji's [97] polynomial fits to molecular equilibrium constants. Some of this data is now quite dated, and Tsuji's method does not provide for easy updating. A better approach is given by Irwin [50] who provides polynomial fits to the logarithm of the partition functions as a function of temperature. The temperature variation of the equilibrium constants can then be expressed as a polynomial with coefficients that are explicit algebraic functions of the partition function polynomial coefficients and of the dissociation energy of the molecule. This is a very useful arrangement since it is generally the dissociation energies  Chapter 5. Conclusions  179  which are the most poorly determined quantities and, therefore, the most subject to change. The representation of the opacities is another limitation of the current models. We need to include more sources of opacity and to improve our treatment of bound-bound molecular absorption in the solution of the radiative transfer. In particular, the vibrational bands of the CO molecule, the atomic lines in the blue and ultraviolet, the CN violet system, the CH G band, MgH bands, and a treatment of VO should be considered before the models can begin to realistically represent emergent fluxes and broad band colours. Straight mean opacities overestimate band strengths. Accurate modelling requires a more realistic treatment of molecular opacities. We anticipate the construction of models using opacity distribution functions (ODFs) in the near future. The computation of such models will require the use of several hundred frequency points, several times the number used in the current models. Since the computational effort scales linearly with the number of frequencies, the computing time for these models will be increased accordingly, probably to around 20 hours of IBM 3081 time per converged model. This estimate is probably pessimistic. Economy can be gained by initial computation of the equation of state and ODF grids once over the entire (T,p) parameter space, for each choice of model abundance and microturbulent velocity. The number of depth points can be reduced by decreasing the depth of the inner boundary ( r  RM  = 100 in the  current models), and perhaps by increasing the optical depth grid spacing in the surface layers where most frequencies are optically thin. Any such reduction yields significant computational.savings since the model timing scales as the cube of the number of depths. Most of the time taken for calculation of the current models is spent not in the linearization solution (which takes only ~ 2 times as long to compute as comparable plane-parallel models) but in the formal solution. This step has a timing ~ 5-10 times  Chapter 5. Conclusions  180  worse than the plane-parallel case due to the much larger number (about 45 for spherical symmetry compared to 6-8 for plane-parallel models) of angles, or impact parameters, needed to accurately evaluate the radiation field moment integrals. The formal solution procedure can be economized as well. The sole purpose of this step is to provide a current value of the Eddington factor /„ to close the system of moment equations. In the initial stages of the linearization solution, the dependent variables T, p, r, and J„ are all poorly determined, and it is almost certainly not worthwhile to engage in substantial computational effort merely to compute 'accurate' values of f , as is presently done. A v  much faster formal solution, on a coarser grid, yielding a rougher estimate of /„ should suffice until the final convergence of the model equations is at hand. Only then need f be calculated using the final depth-angle grids. This procedure alone could probably v  halve the formal solution time. Some work remains to be done on improving the consistency of the numerical results obtained by the formal solution as compared to the moment transfer equations. The present scheme of using the formal solution moment quadratures to evaluate J at depth v  and H at the surface, and obtaining the other moment by differentiation or integration v  using moment equation 2.8 works adequately but is not fully satisfactory. The convergence of the formal solution/moment equation loop is slowed by consistency problems in these moments in the outer layers of the atmosphere. One approach that will be examined is to replace the numerical formal solution by an analytic solution obtained by series expansion at depth-angle points for which  T„<C1,  Since much of the impact pa-  rameter grid is optically thin at many frequencies, this could also result in a substantial improvement in timing. Altogether, the above steps could yield a factor ~ 5 improvement in overall computational speed. This' would permit spherically symmetric models using ODFs to be computed with timings comparable to our present preliminary models, each of which  Chapter 5. Conclusions  181  required about 3 hours of CPU time on an IBM 3081 for full convergence. Another improvement that will be considered in the near future is the incorporation of convective energy transport in the models in a fully consistent manner. This will require the generalization of the equation of radiative equilibrium to include a convective flux gradient in addition to the radiative terms. It should then be possible to linearize the resulting equation of energy balance and proceed directly with the complete linearization as in the present radiative case. One difference is that the new energy balance equation would produce a tridiagonal matrix when linearized, instead of the present diagonal matrix obtained from the radiative equilibrium equation, due to the presence of gradients requiring a 3-point finite difference representation.  5.3.2  Non-LTE Models  In the long term, an area of considerable interest is the study of non-LTE effects in extended stellar atmospheres. The effect of atmospheric extension might be expected to amplify departures from LTE, as compared to plane-parallel models, due to the increased amount of stellar material in a low density; radiation-dominated environment. Non-LTE effects may affect the ionization equilibrium of the outer atmosphere [11], and possibly also the molecular equilibrium, as well as the level populations. The necessary equations of statistical equilibrium, and rate equations for determining the molecular abundances, can be readily incorporated into the present linearization scheme. However, two major modifications must be made to implement a non-LTE code. First, changes to the equation of state routines would be required to allow for departures of the ionization equilibrium from the LTE Saha distribution. Implementation of this should be routine. More substantial effort would be required if departures from molecular equilibrium, e.g. due to non-LTE photodissociation, were to be included. This would entail the addition of a reaction network and rate equations in order to determine  Chapter 5. Conclusions  182  molecular abundances, in addition to the usual statistical equilibrium equations. Second, the Rybicki formulation of the model solution presently used is not suitable for non-LTE work, since the computing time scales as the number of constraint equations. This number is already large even for the relatively simple case of a several multi-level atom non-LTE problem. The problem becomes even larger if reaction rate equations are included, and especially if molecular levels are considered in non-LTE. The direct approaches then become completely infeasible. However, the method of Anderson [4] used to solve a large (~ 100 constraints) non-LTE model of early-type atmospheres appears applicable to late-type atmospheres as well. This method groups transitions with similar physics together into blocks, recognizing that the atmospheric structure depends upon the integral properties of the radiation field and not upon the details of the fine structure (an approach analogous to the replacement of the detailed structure of bound-bound molecular transitions with opacity distribution functions). Because of the difficulties, little work has been done towards non-LTE modelling of late-type atmospheres since the pioneering study by Auman and Woodrow [11]. It is hoped our models can provide a base for future non-LTE studies of late-type atmospheres, and particularly of giant stars with extended atmospheres. However, considerable effort will be required before the state of maturity present in the modelling of early-type atmospheres is attained.  5.3.3  Self-Consistent Atmosphere and Wind Models  The author's main interest lies in the study of atmospheric dynamics. The generalization of the current models to encompass a steady-state, spherically symmetric outflowing wind requires the addition of another fundamental variable, the wind velocity v, and the modification of the equation of hydrostatic equilibrium to account for the transfer of momentum from the radiation field to the stellar gas. Also, the equation of radiative  Chapter 5. Conclusions  183  equilibrium must be generalized to describe overall energy conservation in the radiation field and the stellar material, and now must consider not just the energy density of the radiation field, but also the kinetic energy of the gas, and the flux of quantities such as the internal energy of the gas. The system of equations can be closed by the addition of the equation of continuity. It should then.be possible, assuming a radiatively driven wind, to construct combined atmosphere and wind models that include a fully consistent treatment of radiative transfer, energy balance, and momentum transfer to the stellar wind. These models will still require the use of some ad-hoc assumptions in the treatment of grain formation and growth in order to model late-type giant and supergiant winds. Nevertheless, such models should help in elucidating details of the mass loss process from evolved stars that is so ubiquitous, and yet so poorly understood. 5.3.4  Time-Dependent Dynamical Models  Giant stars of spectral type later than about M6 are almost always variable. Many of these are categorized as Mira variables on the basis of the amplitude of their photometric variations. The underlying cause of the pulsation responsible for Mira variability remains unknown. However, the determination of the temperature structure of the outer atmosphere (r  RM  < 1) of the (static) spherically symmetric models in this study was very  sensitive to numerical perturbations. Scholz and Tsuji [89] report a similar effect that results in a sudden temperature drop over a small optical depth range in the outer layers of their models. Muchmore [74] has proposed that radiative cooling instabilities may be present in the surface layers of K and M giant stars. All of this suggests that the outer atmospheres of late-type giants are only marginally stable at best, and may be radiatively unstable for the cooler stars. This radiative instability could be an underlying cause of Mira variability. An example of just such an instability was given by Woodrow and Auman [109], who  Chapter 5. Conclusions  184  evolved time-dependent carbon rich models for several thousand time steps. These models employed a simplified grey treatment of radiative transfer coupled with time-dependent hydrodynamics and grain formation. Woodrow and Auman's 2400K model, after an initial perturbation, eventually relaxed to a periodic solution. In this model the pulsation resulted from a radiative instability driven by the extreme temperature sensitivity of the opacity of the surface layers (due to the condensation of graphite grains). While the condensation of grains in the atmosphere is now considered unlikely for late-type giants, a similar but smaller increase of opacity occurs in the surface layers of these stars due to the formation of polyatomic molecules (mainly H2O in M stars) for T < 3000K. eff  It is now feasible to contemplate the development of time-dependent model atmospheres of Mira variables, incorporating spherically symmetric hydrodynamics and the complete radiative transfer problem. Realistic time-dependent models are needed to definitively address the importance of the opacity-driven radiative instability hypothesis proposed above. These models should evolve the (one-dimensional) hydrodynamics of an extended, spherically symmetric LTE atmosphere consistent with an accurate treatment of the radiative transfer; The time-independent transfer equation would suffice for this purpose since the thermal and dynamical time scales of fluid elements in the atmosphere are much greater than the fight transit times. However, the presence of an atmospheric velocity field complicates the transfer problem since frequency derivative terms must now be retained in the transfer equations to account for Doppler shifted absorption and emission [70]. Handling this Doppler shifted radiative transfer represents a challenging problem. The time evolution of such models would require considerable computational effort but should be feasible with present-day computing capability. These models would represent the first self-consistent treatment of hydrodynamics and radiative transfer for a late-type stellar atmosphere, and would almost certainly provide fundamental insight into the nature of Mira variability.  Chapter 5. Conclusions  185  Figure Captions Figure 5.1. Temperature profiles of the surface layers for various 3000/0.0 models. (PP= plane-parallel, SS= spherically symmetric). Figure 5.2. Temperature profiles for various 3000/0.0 models. Figure 5.3. Pressure profiles for various 3000/0.0 models. The models of Scholz and Tsuji were calculated for logg = —0.70 and have been shifted by Alogp = +0.70. (PP= plane-parallel, SS= spherically symmetric). Figure 5.4. Temperature profiles of the surface layers for various 3500/2.0 models. (PP= plane-parallel, SS= spherically symmetric). Figure 5.5. Temperature profiles for various 3500/2.0 models. Figure 5.6. Pressure profiles for various 3500/2.0 models. (PP= plane-parallel, SS= spherically symmetric).  o  LO CO  o  O  CO  o  o O  ID CM  csi  1  o  o  ^~  T  ID  O _  Figure 5.1: Temperature profiles of the surface layers: 3000/0.0 models.  Chapter 5. Conclusions  o o  O  o o  o o  00 CD Figure 5.2: Temperature profiles: 3000/0.0 models J_  o o  o o CN  Chapter 5. Conclusions  188  CN  o ^  CO  CN ^  £ °>  CO  O  r CO  <D —•*—</) T3 ^ _^ r- ^ O 03 CO  2  O  d  o o o  1?  3  1  c §< 7o! fc O tl o o CD GO GO CO S  i  CO  ^  A\  CO CO  CN  CN i  d Bo| Figure 5.3: Pressure profiles: 3000/0.0 models.  References  [1] Ahlberg, J.H., Nilson, E.N., and Walsh, J.L. 1967, The Theory of Splines and Their Applications, Academic Press, New York. [2] Allen, C.W. 1976, Astrophysical Quantities, 3rd ed., Athlone Press, London. [3] Alexander, D.R., Augason, G.C. and Johnson, H.R. 1989, Ap. J. 345, 1014. [4] Anderson, L.S. 1985, Ap. J. 298, 848. [5] Auer, L.H. 1971, J.Q.S.R.T. 11, 573. [6] Auer, L.H. 1984, in Methods in Radiative Transfer, ed. W. Kalkofen, Cambridge Univ. Pr., Cambridge, p.237. [7] Auer, L.H. and Mihalas, D. 1969, Ap. J. 158, 641. [8] Auer, L.H. and Mihalas, D. 1970, M.N.R.A.S. 149, 65. [9] Auman, J.R. 1967, Ap. J. Suppl. 14, 171. [10] Auman, J.R. 1969, Ap. J. 157, 799. [11] Auman, J.R. and Woodrow, J.E.J. 1975, Ap. J. 197, 163. [12] Avrett, E.H. and Krook, M . 1963, Ap. J. 137, 874. [13] Baud, B. and Habing, H.J. 1983, Astr. Ap. 127, 73. [14] Beckers, J . M . 1985, in Mass Loss from Red Giants, ed. M . Morris and B. Zuckerman, D. Reidel, Dordrecht, p.57. [15] Bell, R.A., Eriksson, K., Gustafsson, B. and Nordlund, A. 1976, Astr. Ap. Suppl. 23,37. [16] Bell, R.A. and Rodgers, A.W. 1969, M.N.R.A.S. 142, 161. 192  .References  193  [17] Bennett, P.D. 1983, M.Sc. thesis, University of British Columbia. [18] Bennett, P.D. and Kost, C.J. 1989, OPDATA Reference Manual, TRIUMF Internal Report TRI-CP-87-04. [19] Bessell, M.S., Brett, J.M., Scholz, M . and Wood, P.R. 1989, Astr. Ap. Suppl. 77,  [20] Bohm-Vitense, E. 1958, Zs. f. Ap. 46, 108. [21] Brett, J.M. 1989, M.N.R.A.S. 241, 247. [22] Brown, J.A., Johnson, H.R., Alexander, D.R. and Wehrse, R. 1987, B.A.A.S. 19, 705. [23] Brown, J.A., Johnson, H.R., Alexander, D.R., Cutright, L.C. and Sharp, C.M. 1989, Ap. J. Suppl. 71, 623. [24] Brown, J.A. 1990, private communication. [25] Carbon, D.F., Gingerich, O. and Latham, D. 1969, in Low Luminosity Stars, ed. S.S. Kumar, Gordon and Breach, New York, p.435. [26] Carbon, D.F. 1973, Ap. J. 183, 903. [27] Carbon, D.F. 1974, Ap. J. 187, 135. [28] Carbon, D.F. 1984, in Methods in Radiative Transfer, ed. W. Kalkofen, Cambridge Univ. Pr., Cambridge, p.395. [29] Carlson, B.G. 1963, in Methods of Computational Physics, Vol. 1, Academic Press, New York, p . l .  . . <•  [30] Cassinelli, J.P. 1971, Ap. J. 165, 265. [31] Chandrasekhar, S. 1934 M.N.R.A.S 94, 444. [32] Chapman, R.D. 1966, Ap. J. 143, 61. [33] Collins, J.G. 1975, Ph.D. thesis, Indiana University.  References  194  [34] Dalgarno, A . 1962, Spectral Reflectivity of the Earth's Atmosphere III. The Scattering of Light by Atomic Systems, Geophys. Corp. of America, G C A Technical Report 62-28-A. [35 Ekberg, U . , Eriksson, K . and Gustafsson, B . 1986, Astr. Ap. 1 6 7 , 304. [36 Eriksson, K . , Gustafsson, B . , J0rgensen, U . G . and Nordlund, A. 1984, Astr. Ap. 132,  37.  [37; Feast, M . W . 1981, in Physical Processes in Red Giants, ed. I. Iben and A . Renzini, D. Reidel, Dordrecht, p.193. [38 Feautrier, P. 1964, Compt. Rend. 2 5 8 , 3189. [39 Fieldus, M.S., Lester, J . B . and Rogers, C. 1989, in press. [40 Fletcher, R. 1980, Practical Methods of Optimization, Vol. 1, Wiley, New York.  [41 Gingerich, O. 1964, in Proc. First Harvard-Smithsonian Stellar Atmospheres Conf S.A.O. Special Report No. 167, 'p. 17. [42 Gustaffson, B . , Bell, R . A . , Eriksson, K . and Nordlund, A 1975, Astr. Ap. 4 2 , 407. [43; Gustafsson, B . and Glander, N . 1979, Physica Scripta 2 0 , 570. [44; Gustafsson, B . 1981, in Physical Processes in Red Giants, ed. I. Iben and A . Renzini, D. Reidel, Dordrecht, p.25. [45 Henry, R . J . W . 1970, Ap. J. 1 6 1 , 1153. [46; Hinkle, K . H . and Lambert, D.L. 1975, M.N.R.A.S: 1 7 0 , 447. [47; Hummer, D . G . , Kunasz, C V . and Kunasz, P . B . 1973, Comp. Phys. Comm. 6, 38. [48 Hummer, D . G . and Rybicki, G . B . 1971, M.N.R.A.S. 1 5 2 , 1. [49 Hundt, E . , Kodaira, K . , Schmid-Burgk, J . and Scholz, M . 1975, Astr. Ap. 4 1 , 37. [50 Irwin, A . W . 1981, Ap. J. Suppl. 4 5 , 621.  References  195  [51] Irwin, A.W. 1987, preprint. [52] John, T.L. 1968, M.N.R.A.S. 138, 137. [53] Johnson, H.R., Marenin, I.R. and Price, S.D. 1972, J.Q.S.R.T. 12, 189. [54] Johnson, H.R. 1982, Ap. J. 260, 254.  [55] Johnson, H.R. 1985, in Cool stars with excesses of heavy elements, ed. M . Jasche and P.C. Keenan, D. Reidel, Dordrecht, p.271. [56] Johnson, H.R., Bernat, A.P. and Krupp, B . M . 1980, Ap. J. Suppl. 42, 501. [57] Johnson, H.R. and Krupp, B . M . 1976, Ap. J. 206, 201. [58] Johnson, H.R., Marenin, I.R. and Price, S.D. 1972, J.Q.S.R.T. 12, 189. [59] J0rgensen, U.G. 1990, Astr. Ap. 232, 420. [60] Kahaner, D., Moler, C , and Nash, S. 1989, Numerical Methods and Software, Prentice-Hall. [61] Kosirev, N.A. 1934, M.N.R.A.S. 94, 430. [62] Krupp, B.M., Collins, J.G. and Johnson, H.R. 1978, Ap. J. 219, 963.  [63] Kurucz, R.L. 1969 in Proc. Third Harvard-Smithsonian Stellar Atmospheres Conf ed. O. Gingerich, MIT Press, p.375. [64] Kurucz, R.L. 1970, S.A.O. Special Report, No. 309. [65] Kurucz, R.L. 1979, Ap. J. Suppl. 40, 1. [66] Kurucz, R.L. and Peytremann, E. 1975, S.A.O. Special Report, No. 362. [67] Lucy, L.B. 1964, S.A.O. Special Report, No. 167, p.93. [68] Martin, P., Rogers, C. and Rybicki, G. 1984, Ap. J. 284, 317. [69] Mihalas, D. 1970, Stellar Atmospheres, W.H. Freeman, San Francisco.  References  196  [70] Mihalas, D. 1978, Stellar Atmospheres, 2nd ed., W.H. Freeman, San Francisco. [71] Mihalas, D. 1985, J. Comp. Phys. 57, 1. [72] Mihalas, D. and Hummer, D.G. 1974, Ap. J. Suppl. 28, 343. [73] Morgan, W.W. and Keenan, P.O. 1973, Ann. Rev. Astr. Ap. 11, 29. [74] Muchmore, D. 1986, Astr. Ap. 155, 172. [75] Nordlund, A. 1974, Astr. Ap. 32, 407. [76] Pecker, J.-C. 1965, Ann. Rev. Astr. Ap. 3, 135. [77] Peraiah, A. 1984, in Methods in Radiative Transfer, ed. W. Kalkofen, Cambridge Univ. Pr., Cambridge, p.281. [78] Peraiah, A. and Grant, I.P. 1973, / . Inst. Math. Applic. 12, 75. [79] Peytremann, E. 1974, Astr. Ap. 33, 203. [80] Querci, F. and Querci, M . 1975, Astr. Ap. 39, 113. [8.1] Querci, F., Querci, M . and Tsuji, T. 1974, Astr. Ap. 31, 265. [82] Reimers, D. 1981, in Physical Processes in Red Giants, ed. I. Iben and A. Renzini, D. Reidel, Dordrecht, p.269. [83] Rybicki, G.B. 1971, J.Q.S.R.T. 11, 589. [84] Sauval, A.J. and Tatum, J.B. 1984, Ap. J. Suppl. 56, 193. [85] Saxner, M . and Gustafsson, B. 1984, Astr. Ap. 140, 334. [86] Schmid-Burgk, J. 1975, Astr. Ap. 40, 249. [87] Schmid-Burgk, J., Scholz, M . and Wehrse, R. 1981, M.N.R.A.S. 194, 383. [88] Scholz, M . 1985, Astr. Ap. 145, 251. [89] Scholz, M . and Tsuji, T. 1984, Astr. Ap. 130, 11.  References  197  [90] Scholz, M . and Wehrse, R. 1982, M.N.R.A.S. 200, 41. [91] Simonneau, E. and Crivellari, L. 1988, Ap. J. 330, 415. [92] Sneden, C , Johnson, H.R. and Krupp, B . M . 1976, Ap. J. 204, 281. [93] Somerville, W.B. 1964, Ap. J. 139, 192. [94] Strom, S.E. and Kurucz, R.L. 1966, J.Q.S.R.T. 6, 591. [95] Tsuji, T, 1966, Publ. Astr. Soc. Japan 18, 127. [96] Tsuji, T. 1968, in Low Luminosity Stars, ed. S.S. Kumar, Gordon and Breach, New York, p.457. [97] Tsuji, T. 1973, Astr. Ap. 23, 411. [98] Tsuji, T. 1976, Publ. Astr. Soc. Japan 28, 543. [99] Tsuji, T. 1978, Astr. Ap. 62, 29. [100] Tsuji, T. 1986, Ann. Rev. Astr. Ap. 24, 89. [101] Vardya, M.S. 1966, M.N.R.A.S. 134, 347. [102] Watanabe, T. and Kodaira, K. 1978, Publ. Astr. Soc. Japan 30, 21. [103] Wehrse, R. 1975, Astr. Ap. 39, 169. [104] Wehrse, R. 1981, M.N.R.A.S. 195, 553. [105] White, N.M. and Wing, R.F. 1978 Ap. J. 222, 209. [106] Whitelock, P.A., Pottasch, S R . and Feast, M.W. 1987, in Late Stages of Stellar Evolution, ed. S. Kwok and S.R. Pottasch, D. Reidel, Dordrecht, p.269. [107] Wing, R.F. 1971, in Proceedings of the Conference on Late-Type Stars, ed. G.W. Lockwood and H.M. Dyck (Kitt Peak National Observatory Contribution 554), p.145.  References  198  [108] Wing, R.F. 1981, in Physical Processes in Red Giants, ed. I. Iben and A. Renzini, D. Reidel, Dordrecht, p.41. [109] Woodrow, J.E.J, and Auman, J.R. 1982, Ap. J., 257, 247.  Appendix A  The Linearized Model Equations  A.l  The Model Equations  The spherically symmetric model atmosphere equations are summarised below. These equations have been put into the form  f(x) = 0  (A.l)  of section 3.3.  A. 1.1  Transfer Equation  fd+iqd+i 1 (qdXd + qd+iXd+i)Ar (^t1) J d + 1 .(<ldXd + qd+iXd+i)&T 24.q d  d  d+1  • - -•• ,]r..'^ 1  - fdqd  .{<ldXd + ?d+iXd+i)Ar  (q _xXd-\ + gdXd)Ar _iJ \RJ  d  d  d  - i^~^ ^ + 1d+iXd+i)A.T + (q . -i d  d  d  d lXd  + ^dX^Ard.i] {-)  J  d  , [ fd-iqd-i 1 &f-i/ . -s 1 fr -i\ + 7 ; ; -^TT- " - 777 Ud-iXd-i + 9dX<j)Ar _ I — - ) J _ i \q -iXd-\ + qdXd)&T -i 24g _i J\ RI 2  d  A  <i  d  d  T  1  d  d  + 777 KQdXd + q iXd+i)&T I ——- j 24 q \ V it / 1 £d d+  d  B  d+1  d+  +  1 2 g^****  +  9d  +^ +) 1  :<i  1  / + 9dXd)Ar _!] ^—J  Ar<i +  d  \  •  2  £  d  + ^ —(gd-xXd-i + 9dXd)Ar _ f ^ ) 5d-i = 0 24 <7d_i VK / 2  d  199  1  (A.2)  Appendix A. The Linearized Model Equations  A.1.2  Transfer Equation: Inner Boundary Condition  fD-iqD-l 7 : ; ~ TT {qD-lXD-l + qDXDj&TD..!  1 ' (D-1,  .  24 (Jrj-l .  fDqD  -  1  1 ( D-l \ (—5-) r  A  -  I f D\ r  u  D  x  J  \ K /  L*  /dB\  128*O-R T XD  +  \dT) ~  2  D  A.1.3  D  Transfer Equation: Outer Boundary Condition  ,  -  /2?2  ,  . ( 9 i X i + 92X2)Ari  ^|  t  e  +  ^  1  /f2\  2  2  z)ATl  12 gi  2  (i) ^ + ^ 2  |  t  J Vii/  o + 92X2)Ar 1  [  1  ,  1  where the extension parameter d' is JI  _  _  D  1+A  H  n(l + A)  r  Ti is the density scale height, and K l  A . 1.4  Q) S 1 2  :  I l_exp(-r 6xl)][GH(^ + 5 G ^ ) ] Q ) V = 0 +  +  2  u — ~(fcXi + ftXajAr! I —) J 24q J \-R/  . ( 9 i X i + 92X2)ATI +  -  16/-.  rrr  \dpj  y  Radiative Equilibrium JV  ^2/WndindX.ndJnd  - 1 = 0.  n = 1  JV  ^  WndtndXndBnd  J  T  l fr24  t  2  ^—j  + qDXD)AT _  c  2  J V -R /  D  ,  + , 9 D X D ) A T £ , _ I + ^ 1 2— ? (iD-lXD-l  (qD-iXD-i  . - x + qDXD)AT ^  {QD-IXD-I  D  D  Appendix A. The Linearized Model Equations  A. 1.5  201  Hydrostatic Equilibrium 2  -V+(-V  Pd - Pd-i  Td-iJ  gAma-x - ^ A T , ^ ) = 0  (A.8)  \r J d  where Amd-i =  + -)  2 VXd-i  Xd/  Ar  - ^  - *)  12 VXd-i XdJ  w  Ar^  (A.9)  and  (A.10)  R  GM.  9 dp  (A.ll)  R?  (A.12)  f  dr  ATTCR ' 2  and  Pd \drj A.1.6  l  (A.13)  di  XRM,d  L*-yxi  (R  gn  (1 + A)  Xi  AircGM (l  \ri  (A.14)  + A)  t  Radius or Depth Equation  1 /fid-iPd-iXd-i  2V  T^  +  d  , P-dPdXd ^ (r -i  T  d  - r) d  d  _\_(  12  fid-iPd-iXd-iSd-i  fidPdXdHd  Pd-iPd-iXd-i d-i Y  kTl,  Td-i , +  A.1.8  dx*  Hydrostatic Equilibrium: Outer Boundary Condition Pi ~  A.1.7  Y,=  ~ rdf  ~ kAr^ = 0.  (A.15)  Radius Equation: Outer Boundary Condition  r-i where R =  R =  (A.16)  0  and the factor £ is defined such that r(r ) = i?* 0  Appendix A. The Linearized Model Equations  A.2  202  The Linearized Equations  The model equations are linearized in terms of the independent variables T, p, r, and jv = (r/R) J . As shown in section 3.3, the resulting linear system is of the block matrix 2  v  form / T Tt,  n 0  . •• .. 0 n  0  T  0  0  Wi  w  0  0  • ••  2  0  •• T 2  N  ..' .'. W ..  N  0  TT,  v,  si, \ S  u  2  v  2  s  u  N  v  N  SN  c  D  E  A  B  F  / Ajx \  Aj  2  / Ki  K  2  2  KN  AT Ap Ar  \  (A.17)  M  L  •• 0 H G o Q \0 \ J V Here, and throughout this Appendix, it will be assumed that all references to the correcP  tions A j ' i , . . . , AJN, AT, Ap, and A r refer to the scaled variables as defined in equations (3.179-3.182) and similarly, all linear coefficients are assumed to have been scaled as per equation (3.183).  A.2.1  Transfer Equation  The linearized equation of radiative transfer for frequency n, at depth d in the atmosphere, is given by the <i-th row of the matrix equation  T A j + U A T + V Ap + S Ar = K . n  n  n  n  n  (A.18)  n  Since T , U„, V , and S are all tridiagonal matrices (of order D, the number of depths) n  n  n  this can be expanded into the scalar equation T ,d,d-\ Aj„ f_i + T Aj n  i(  n>dd  + T d,d+l Aj  nd  nt  + U ,d,d-\ Ar<i_i + U AT + n  nidd  + V - Ap _ nidtd  1  d  1  + V Ap + V nidd  + -Sn^d-iAr^-i + S Ar n>dd  U  d  d  d  +S  nidid+  Ap  nidid+1  n>d>d+  iAr  nid+1  iAT i d+  (A.19)  d+1  d+1  =  K  nd  Appendix A. The Linearized Model Equations  203  These coefficients have the values, for 2 < d .< D — 1,  T ,d,d-1  A.20  n  T dd  Ol2  n>  T ,d,d+1 =  a  n  A.21  13  J  A.22  n,d+1  Td-x  Un,d,d-1  (i)/ (I)/ (§)J  0-22  U dd n>  a23  Un d+1 id>  (i)d  a32  —  +a33  (i)d  +1  +1  +a43  <Sn,d,d-l =  Trf  A.24  (i)d J  A.25  Td+1  +1  Pd-i  —  Ki,d,d+1 =  (8B\-  a42  Ki,d,d-i = ^n,dd  A.23  \dp)d+i » • a-t-i  a2  A.26  Pd  A.27  Pd+i \\PJd+i. * • a-t-u  ;A.28  a33 d  , i r<i_i ^51 + CiJJ^)+cJW±A]rd 4 1 \dr _ ) V^d-i/J ^  A.29  1  d  Sn,dd — Sn,d,d+1 =  1  '^d-i  «52 + Cl , / 93d \ , ass + c 1 -r I+c \or J 2  fda  d + 1  \  r  V^+i/J  3  d+1  S  r-d(A,30 A.31  d+1  -&nd = J>,  A.32  where an  €d-i , _ - \A' , _i(gd-iXd-i + gdXdjArd-! 24g £d+l / - , - \ * 7—=—; V A WA VldXd + qd+iXd+i)Ar [<ldXd + qd+iXd+i)&T 24.q •1 -fdqd (?d-i%d-i + 9dXd)Ar _ + (%xd + gd+iXd+OAr^ fd-iqd-i ^ T A_ (qd-iXd-i [+ ?dXd)Ar /d+l?d+l <i  0  1  A.33  d  A.34  d  K  d  ai2  d+1  d  _A . 12%  1  [ (gd-iXd-i + qdXd)^-!  + (q x + gd iXd i)Ar d  d  +  +  d  , (A.35)  Appendix A. The Linearized Model Equations  204  fdqd(r<i/R) J d - fd-iqd-i{rd-i/R) J ,d-i (qd-iXd-i + qdXd) &T -i id 2  2  n  qd-1  ^21  n  2  d  24,tl^- ( X)  " • '''"  !  1  Bn  ,)  (A.36) fdqd{rd/R) Jnd  —  2  «23  =  qd+1  fd+iqd+i(rd+i/R) J ,d+i 2  n  (qdXd + qd+iXd+i) &Td 2  &d  ( ri,d+1 - B ,d+l)  T  24<fc4-1  J  R  n  Ar Q) (J 2  12q  d  2  -5  n d  )  (A.37)  fdqd{rd/R) J d - fd+iqd+i(r i/R) J ,d+i (qdXd + qd+iXd+i) AT 2  qd  a2  n d  d 2  n  d+  n  2  d  fdqdirdlR) Jnd fd-iqd-i(r -i/R) J ,d-i + (qd-iXd-i + qdX.d) AT -i 2  2  d  n  2  d  d+l  ^ — Ar _x ( " ^ T ) (Jn,d-1 — B -l) 2Aq . id ( A r _ + Ar ) (^) ' (Jnd - 5 ) 12q d  n>d  d  d  a  d  (A.38)  n d  d  «31  =  «33  =  «32  =  (A.39) (9dXd +  d  24 1  qd+lXd+l) ( - ^ )  (A.40)  (^n,d+l - #n,d+l)  % + 1  (gd-iXd-i +  12?  qdXd)Ar -i + d  + qd+iXd+i)Ar  d  d  (^) (J„ -5 2  d  n ( i  id  ) . (A.41)  M<fc-iX«£-i + qdXd)^r -x ( - ^ p ) 2Aq  (A.42)  «43  ^ ( ^ d  (A.43)  «42  ^  a  4 1  d  d  +  qd^Xd^Ard^—)  ( ? d - i X d - i + gdXd)ATj_i + ( g ^ d +  qd+iXd+i)AT  d  (  (A.44)  Appendix A. The Linearized Model Equations  205  (A.45)  <*51  (A.46) +  «52  ci  u tq  +  d  =  ~Xd  [  **Xd) <*-i + (?dXd + gd+iXd+i)Ar ]  +  5 (A.47)  Ar  d  (^)  nd  (gd-lXd-1 + ?dXd) A r _ i 2  d  +^i!d-iAr i_i(J d_i — 5 d_i) <  (A.48)  n>  /d+lgd l«^n,d+l +  c =  OH.  2  (QdXd + qd+iXd+i) A T * 2  +^^+1 C  ni  (A.49)  Ar<i(J i i — 5 d i) n)< +  ni  +  nd 3  L(9dXd + gd iXd+i) Ar 1 12 £ d A T d ( J d — -B„d)  •R'  2  +  d  (A.50)  n  C  fdqp, nd rdy g (qd-iXd-i + gdXd) Ar _  +Xd-i  4  2  d  1  — — £d&-Td-l{Jnd  (A.51)  B ) nd  fdqd(r /R) Jrui - fd+iqd+i(rd+i/R) Jn,d+1 (?dXd + qd+iXd+i)A>T 2  b =  —  2  d  d  fdqd(rd/R) Jnd -  fd-iqd-i(rd-i/R) J ,d-i  2  +  2  n  (qd-iXd-i + gdXd)Ard_1 + 24*q  d+l  ( * qd  + ^^(qd-iXd-i 2  d +  9d+iXd+i)^r  (^ ) L  d  (Jn,d+i ~ B ,d+i) n  + gdXd)Ar _! ( ^ r ) {J ,d-i -  \ (<ld-iXd-i + TTTI2q L  d  n  B , -i) n d  + 9dXd)Ar _ + (q JCd + %+iXd+i)Ar <i  1  d  d  d  x  The  ratios a and 8  (^)Vnd--M. (A.52)  are defined (A.53)  Appendix A. The Linearized Model Equations  a  206  9d-i  qd+i  -l  (A.54)  d  qd  and their derivatives with respect to radius are da  d  5r _i  r  2  d  dp -i d  +  d  qd-i r -i qd-\ " 2  d  r  d  fdr  2 d  nfd-i  dB dr  qd+i r  2  ! qd+i 2  r +i  d  qd  x  da  d  3/d-i-i\  d  d  d  da  d+1  9r  d  r -i , qd-i '2 i T 2 qd ' 2 r I 2 qd+i. d+l r qd+i 2  dpd-i  fd+ird+l Zf - 1 d+1  U d+ir  d  d  d  q  dr  da i d+  d  dr  qd Jd+i  d+1  fd l Zf - 1 r  +  2 d  (A.59)  d  +  fdrl  (A.60)  fd-ir ^ 2  (A.61)  d  d  r  d+1  fdT  (A.58)  rf  d+1  d  dr  3/ -l  +  d  2  r  qd  d  d(3  2  2  d  (A.57)  fd-ir ^  d  d  d  (A.56)  3fd-i  +  2  d  2  d  d  dr  (A.55)  fd-ir _y d  2 r -i r -i '3fd 1 qd '2 2 fdr qd-i r r +i qd "2  dr -  A.2.2  2  q  d  2  Jd+l'd+l Zfd+i - 1  fd+ir  Zfd fr  +  d+1  d  (A.62)  d  Transfer Equation: Outer Boundary Condition  The linearized outer boundary condition of the transfer equation for frequency n is given by the first row of the matrix equation  T A j + U A T + V Ap + S Ar = K . n  n  n  n  n  (A.63)  n  Expanding these tridiagonal matrices yields, for d = 1,  TniiAj + V A nll  nl  Pl  +T  n l 2  Aj  n 2  + V Ap nl2  2  + tfnuATi +  tf„i Ar  + SnuAn + S  2  n l 2  Ar  2  2  =  K  nl  (A.64)  where  Tn n  —  d\  2  {^^j  Jnl  (A.65)  Appendix A. The Linearized Model Equations  T 12 n  u  —  nll  207  (A.66)  &13  «^n2  a 2 2  (jr)/"  8 1  (§ )1  +042  i  (If),  OB \ 6T) . I 1  (A.67)  + 4(  a  23  dB\ '  ( ^ ) + «33 2  T  . (A.68)  2  (A.69) (A.70) (A.71) (A.72) (A.73) where 0-12  °18  (?iXi + 92X2)Ari =  j~2 Q2 7—=—T ^TA 1X1 I" q X2)L±T\ 2 ( ? I X I + 92X2JAT1  12gi  +  q 2 X 2  ^  (A.74)  T l  £2  (A.75)  ^ — ( 9 l X l + 92X2)AT!  ^92 24q ' fiqi(ri/R)JJni - /2g2(r y /R) Jn2 \fiqi(ri/R) i ~ f2q2(r (91X1 + 92X2) — 9 i -• 7—;—: \2\ Ar! 2  22  2  2  n  2  2  &22  (A.76) +  |ri jiexp(-T 6xi) ' fiqi(ri/R) J 1  <  2  nl  ^23  = 92  Bi  GH'^J-^X  n  f q (r /R) J 2  2  2  2  n2  ( 9 i X i + 92X2.) Ari 2  6  . /r \» 2  5)  (A.77)  n2  'nl  208  Appendix A. The Linearized Model Equations  A T !i  Ct32  /"^lN  12?;- ( 0 1 X 1  + 92X2) (^J  exp(-r ;ixi) G  + \n i X  H  ic  An  «33  2 4 ? 2  ( ? l X l + ?2X2)  ^ ( 9 1 X 1 + 92X2)  «42  -  12q  Q)  43  «52  = =  F  24  to+?2  =  C  -  2  2  (A.79  n2  A n g )  2  d 1+ A  H  T  +  9tGj  1+ A (A.81  +^-(91X1 + 92X )An  B  2  nl  [ttr (  ^ (?lXl + 12?  ^ )  T  +  G,( ^ )  ] ( £ ) B (A.82 M  (A.83  9 2 X 2 ) ^ 1 ( ^2 ) #n2  ,iE , hq\J 2 n  "Xi  © ^(A.78  1+ A  )Ari  2  C  + gtGj  1+ A  ^ G)'  + (1 <*53  nl  (^n2-5 )  + i[l-exp(-n6xi)] a  B)  .(9iXi + 92X2) An  nl  3  1 + —6An(J  n 2  - £  )  (A.8'4  - —6ATi(J  nl  - B)  (A.85  24  2  (gixi + 9 2 X ) A n 1 - exp(-r ;ixi) 2(1 +A)  n 2  nl  12  2  2  (A.86  ic  1 - exp(-n6xi) ( 2(1 +A) + d  Co  -  -(m) ^(iTA)lQ) nl +  ff  b =  fMri/R) Jnl  -  2  !  (A.87)  f2q2(v2/R) Jn2 2  (91X1 + 92X2) A n  + ^ f ^ * + 92X2)An 2  1  (Jn2 - 5 )  1  n2  + i^;(?ixi + 92X2)An g ) - |[1 - exp(-nc;iXi)] G  H  2  (Jm - JB„i)  d 1+ A  +K  g ) ^ . g) B A.88) 2  l (  Appendix A. The Linearized Model Equations  209  where the quantities Gj and G' refer to the first derivatives of the respective functions. H  A.2.3  Transfer Equation: Inner Boundary Condition  The linearized inner boundary condition of the transfer equation for frequency n is given by the D-th row of the matrix equatii— T Aj n  + U A T + V A p + S Ar = K .  n  n  n  n  (A.89)  n  Expanding these tridiagonal matrices gives,i for row d = D T D,D-l nt  Ajn,D-l + Tn, DD&jnD *  . ,r + U DD  + U„ D,D-lATD-i m  )  t  A  T  d  nt  + K.JO.D-lApD-l + + Sn,D,D-lAr -l  V  Ap  niDD  (A.90)  D  + S DDAr  D  n>  =  D  K  nD  where (A.91)  T ,D,Dn  (A.92)  T DD n<  i,D,D-l  T . .1  (A.93)  D  (A.94)  D  - \  Un,DD  +  °-(§)  ^(§)  + D  D  (d B\ 2  +0822  {dT~ ) 2  D  •fdx\  V ,D,D-l  ,  I  (dt\  n  S ,D,D-\ n  S ,DD n  (|)D+a32(|)J^  j  022  051 L+  «52  + Cl C l  + ^  Q  \orD-x -r  9  \  (A.96)  d Dfcj *fcj] r  a  (A.95)  n  V ,DD  +  +  (A.97) ri  (A.98)  \dr  D  KnD  (A.99)  Appendix A. The Linearized Model Equations  210  where CD-  fp-iqp-i  Oil  (?i?-iXr>-i + g DXr»)Ar ,_i J  a  12  =  fpqp (qD-iXD-i  CD  + qpX.p)& P-i  = 9D-1  n  +  (qD-lXD-1 ^  — ATD_I (  D  (A.100)  + qDXD)Ar _  (A.101)  D  J ,D-l n  qDXD) ^TD-\ 2  j^ ) 1  (^",0-l  Bn,D-\)  —  ?Z)  CD  /r„\2  (A.102)  12,^-(i) fDqD(rp/R) JnD  ~ fp-iqD-\{rp-llR)  2  0,22  3D  t  D  - /D-l?D-l(^£)-l/' Rf  fpqp(rp/R) J P  24  (qD-iXD-i  + gz?XD)ATr,_i  12q  T  2  «21  —(qD-iXD-i  24gp_i  £  J ,D-l  2  n  + gz>xz?)Ar£,_i 2  {qD-lXD-1  ^ — A r p - i ( ~ ^ ) («7n,JD-l 24 L  -  -Sn.D-l)  ?rJ  (A.103)  ^STTCT^T^X , 2  AT£)_I  ^31  24?i? ~ (  &32 =  ;  -  l X £ >  -  1 +  ^DXD) ( - ^ )  (•'n.D-l - £n,Z>-l)  -J^ (gi?-iXr>-i + 9z?Xz>) (^)  (^nu - #„D)  i  cjD-i  ai  9 Z J  ,  - \A  .  frp-i\  I  ai = 5  ^52  =  ^82  ^92 =  ^(fl -iXi,-i+9 XD)AT ,_x ( ^ ) D  D  ^-(?JO-IXU-I +  12?JD  g  1  9z?XD)Ar j_i Z  ( t o - l X D - l + qPX.p)^TP-\  L,  (A.106)  (^^)  (^)  (A.107)  2  # n Z ?  -Bn.u-  (A.108) (A.109) (A.110)  128*<rB*T*XD 3L* (dB\ l28ir<rR?T p \dTJ DX  (A.105)  2  4  «42  (A.104)  D  (A.lll)  Appendix A. The Linearized Model Equations  fD-iqr>-lJn,D-l  -XD  Cl  211  R  )  [{q -\XD-\ + gDX£>) AT£,_i 2  D  + 24^r»-iAr£»_ (J 1  c  3_ — jB £)_i) 1  ni  (A.112)  nfl 4  R'  [(QD-IXD-I  +  qDXD) AT 2  D  — ~-t,D&T _i(J D  b =  fDqD{rD/R) JnD  ~  2  1  —  nD  B ) nD  (A.113)  fD-iqD-l{rD-l/R) Jn,D-l 2  (qD-iXD-i  qDXD)&T _i  +  D  X  A.2.4  ri|Z  1 fdB\ ~ ' — •  *  (A.114)  Radiative Equilibrium  The linearized equation of radiative equilibrium appears as row JV + 1 (third from the bottom) of matrix equation (A.17), JV  ^  W A j + C A T + D Ap + E A r = M n  (A.115)  n  T»=l  All of the matrices W , C, D , and E are diagonal and so the linearized radiative equin  librium constraint in scalar form is  X)  w  + C AT + D A  n,dd&jnd  dd  d  dd  Pd  + E Ar dd  d  =M  d  . (A.116)  n=l  Therefore, letting JV (KB)  d  =  ^  WndtndXndBntl  (A.117)  ndindXndJnd  (A.118)  n=l JV (Kj)  d  =  Y n=l  w  Appendix A. The Linearized Model Equations  2 JV  R  J TdJ  JV  ^2  Wnd  C,dd («J)  ("57^ J +  Snd ( L  r, — 1  (A.119)  WndCndXndjnd ^-f n=l  '  the coefficients can be expressed  212  Xnd  \  J, nd  dT  nd-  N  nd  6id  +  |  )  dT  + Xnd  Wind!  B  nd  \ T  d  (A.120)  JV  Ddd  £  =  Wnd  'nd  n=l  JKj)d  JV  E Edd  + n  d  ^U/nd.  'nd > Pd  (A.123)  _ M)d  x  (A.121) (A.122)  WndindXndJnd  n>d  A.2.5  l  d  W d M  ^ 5j  n=l  2{ J) K  Wnd  (A.124)  d  Hydrostatic Equilibrium  The linearized equation of hydrostatic equilibrium appears as row N + 2 (second from the bottom) of matrix equation (A.17), A A T + B Ap + F A r = L  (A.125)  The matrices involved here are all bidiagonal, and so the corresponding scalar form of this equation is A  d | d  _ A T _ + A AT 1  ( i  1  dd  d  + B^-iA^.! +  B A dd  Pd  ' + Fd^Atd-r + F Ar dd  d  =  L  d  (A.126)  Appendix A. The Linearized Model Equations  213  where  1 "2 \ d-l,  Ad,d-\  'IV  +J  T  1 2 \r -ij  i-dd  \r J  d  - 5  1  J V  d  f  Bd,d-i  9  DJ  R  9I  'R ' R + \ d-\. Jd  (A.128)  - 5 - —  dpd-i  V fyd  9  • Pd-i  (A.129)  'Pd  (A.130) (A.131)  d  Fdd  -it)**  -) Td)  _  =  d  x  ( dAm-d-i  T  Fd,d-i  L  (A.127)  -  dTc  \Td-l  Bdd —  Td  (A-132)  1  Pd-l -Pd + -  Td-lJ  \rdy  x gArrid-i -  (A.133)  ) Ar -i d  and  Arrid-! =  lURM,d-i+X ,d)& d-i T  RM  ~ i2(x ,d-i^d-i — X ,dXd)Arl_ RM  V  Y = d  RM,d XRM,I  9 = dp  r  dr  dr )  RM  1  (A.135) . d  GM R? U (7d-i+7d) 8ircR  (A.136)  t  (A.137)  2  d Id =  XH  (A.134)  ) .  (A.138)  The quantities Y , g, dp /dr, and jd remain fixed during the linearization. The derivatives d  r  of Am -\ above are d  dAm-d-i dTd-i  1 -2  fd  ,  XRM  Ard  d-i  _  i  1 - ^-xAr,.! 0  (A.139)  Appendix A. The Linearized Model Equations  dAm, dT dAm -i  _ 1  ( XHM\  214  9  2  1+  1  d  (dx*M\  2  i -•--y _ Ar _ o d  2  dp -i d  dAm _i  1  d  -2  / dXRM  , A  1  l + -y Ar _ 6 d  dp  d  A.2.6  (A.140)  -YdAr^ 0  d  d  i  d  1  (A.141) (A.142)  Hydrostatic Equilibrium: Outer Boundary Condition  On the outer boundary, the scalar equation reduces to the form A n A2\ + -BnApi + FnArx = L  (A.143)  r  with, A  1X  (A.144)  = R  1 + JT,(1 + A) F  11  L  x  = 2.(1 +A) ( £ )  ny  2 X  r  m  *  dp  2r  Pi  J,  dp  T  9XRM,I  1 + A V dr ) 1 (dp \ ]  (A.145) (A.146).  lm  2 r  = 71(1 +A)  \  r  (A.1.47)  and, as before, 9 = dr  A.2.7  =  (A.148)  2  4TTCR  I 7i  i?  2  7i  XH XRM / i  (A.149) (A.150)  Radius Equation  The linearized radius equation appears as (the last) row iV + 3 of matrix equation (A.17), QAT +HAp +GAr =P  (A.151)  Appendix A. The Linearized Model Equations  215  The matrices involved here are all bidiagonal, and so the corresponding scalar form of this equation is Q ^ - i A T _ i + Q AT + Hd^Apd^ + d  dd  HA  d  dd  + (?<*,<*_! Ar-d.! +  Vd  G Ar dd  = P  d  (A.152)  d  where Qd,d-i  0-21  Qdd  +  V dT  ^31 + 041  D>D  HM  32 + &42  FFL  G -i d>d  G  d  &22 + «42  H -\  dd  Pd  )  dp ) _ d  x  (A.154)  + a2 dp Jd 7  I  -s-  /d  ^51^-1  (A.157)  ^52^  (A.158)  b  (A.159) p p XRM,d/T ,  d  X -i (r -i 2T _ Xd (r - 2T X -i (r -i 2p -i X (r -i -  «22  - r ) 1-  d  d  d  =  (A.156)  Pd  \dp i  and p = pkT/p is the mean particle mass. Then, setting X = &21  (A.155)  Pd-i  dP; -u dp d  ( 9XRM\  V  X  dp df dp  + ^72  ( 0XRM\  V  (A.153)  T-  O41  d  - ±t±Y _ ){r _  _  <2  d  d  x  d  d  d  l  (A.160)  r d )  1  d  r) l + ^C^-^^Kr-d-i-r,)  t  (A.161)  d  d  d  «31  d  - r) 1  - ^^-i)^-!  d  -r )  (A.162)  d  d  «32  =  ai = 4  2p  d  X -i d  d  - (H  +  6  d  2X  Y )(r ^ - r )  (A.163)  n  d  d  d  2A,  d-i k  (r -i - r ) d  d  2XRM,d-\  1 IX (r -i - r ) 6 ^ - -Y^d-i 2XRM,d X -i 1 - ^(Ht-t - ^-Yi-Mrt-i - r) d  =  r) l  d  x  «42  ^51  d  1  d  +  d  d  d  )(?•<£-1 -  - r) d  r) d  (A.164) (A.165)  Appendix A. The Linearized Model Equations  1 + ^{H - ^Y )(r -i  2  d  Xd-i 052  a a  =  7 i  =  7 2  =  d  216  - r)  d  (A.166)  d  1 - ^(H^ - ^ - Y ^ r ^ - r )  2~ _ Ad l+ 2 Ad_i  d  (A.167)  liH.-^Y^r^-r,)  (A.168)  (ra-i - ra)  1  +  2^  ( r d  - 1  r d )  1 + -(#  1\ --j-Yd)^  d  (A.169)  - r) d  b = fcArd-! - ^~(r -i - r ) l-I(JET _ -^Xi_ )(r i-i-r ,) d  d  - y (<*-i ~ "> r  d r  <l  1  1  + k( * ~ T^X ^ H  0  A.2.8  1  7  4  r) d  (  (A.170)  K  Radius Equation: Outer Boundary Condition  On the outer boundary, the scalar equation reduces to the form  (A.171) with  Q u  (A.172)  = 0  (A.173)  #11  - IJL  (A.174)  R  Pi = 1-  R  (A.1.75)  Appendix B Model Output Tables  , This appendix gives condensed listings of the frequency-independent or frequency-averaged quantities of interest that are calculated by the models listed in Table 4.4. A sample of the tabular output generated by the OUTPUT command in the ATHENA model atmosphere code is presented on the following pages. For reasons of brevity, only every second depth point has been printed and only the first three pages of output included for each model. A glossary of the variables appearing in this output follows.  217  Appendix B. Model Output Tables  B.l  218  Glossary of Variables  TauRM  Rosseland mean optical depth Tj^.  Col Mass  Column mass m in units of g/cm .  r/R* - 1  Atmospheric extension r/R — 1, where i?» is the stellar radius.  T  Temperature T in degrees K .  n  Number density n in c m .  Pe  Electron pressure p in dynes.  P.gas  Gas pressure p in dynes.  P_rad  Radiation pressure p d in dynes.  Mu  Mean molecular weight p.  Rho  Density p in g/cm .  ChiRM  Rosseland mean (total) absorption coeffcient XRM in units of cm /g.  r  Radius of atmospheric layer in cm.  TE res  Transfer equation rms residuals e  RE res  Radiative equilibrium residuals  HE res  Hydrostatic equilibrium residuals €u .  DE res  Depth (radius) equation residuals e .  EPS rms  Total rms residual vector (all equations) e -  g_eff/g  Ratio of (gravity - radiative acceleration)/gravity.  Chi_H/RM  Ratio of flux mean to Rosseland mean opacity  Hconv/H  Ratio of convective to total flux H /H.  Lum/L*  Ratio of luminosity from total flux H to specified stellar luminosity L*.  Cp/Cv  Ratio of specific heats C /C .  H_p  Pressure scale height h.  Qconv  Convective quantity Q = — (dmp/dhiT) .  DELrad  Radiative gradient V# = (d In T/d In P)R.  2  t  -3  e  Ta  3  2  ,.  TBrm  B  DB  rms  XH/XRM-  cmv  p  v  p  Appendix B. Model Output Tables  DELad  Adiabatic gradient V  DELbub  Gradient of convective bubble V B .  TauRMb  Rosseland mean optical thickness of convective bubble.  Vconv  Mean convective velocity of bubble in cm/s.  Hconv  Convective flux H  B.2  conv  5  = (dlnT/dlnp) . s  in ergs/cm /s. 2  Converged Models: Output Tables  Condensed output of the converged models summarized in Table 4.4 follows.  Model 01310191  SS 3000./ 0 . 0 0 /  TauRM  Col Mass  r/R*  -3.00  Summary of p h y s i c a l  CN. T i O , H20 (sm)  - 1  T  n  Pe  P_gas  P_rad  Mu  1 3 5 7 9  361E + 00 4 .477E- 13 363E + 00 6 .588E- 13 360E + 00 9 .440E- 13 355E+00 1 .321E- 12 349E + 00 1 .809E- 12  1 .647E -04 1 .877E -04 2 . 196E-04 2 .622E -04 3 . 183E-04  1 1 13 15 17 19  9 .429E-02 9 .535E-02 9 .647E-02 9 .767E-02 9 .897E-02  1 . 343E + 00 2 .437E- 12 1 .337E+00 3 .235E- 12 1 . 332E + 00 4 .247E- 12 1 . 326E + 00 5 .531E- 12 1 .320E+00 7 .165E- 12  3 .906E -04 4 .829E -04 5 . 9S5E -04 7 . 403E -04 9 .092E -04  21 23 25 27 29  1.251E+00 . 1.652E+00 , 2 .201E+00 2 .977E+00 4 .126E+00  1 .004E-01 1 020E-01 1 .039E-01 1 .061E-01 1 .088E-01  1 .315E+00 1 .310E+00 1 .305E+00 1 .299E+00 1 . 293E + 00  12 1 1 11 1 1 1 1  1 . 102E-03 1 . 309E -03 1 .51 1E -03 1 .678E -03 1 . 773E -03  31 33 35 37 39  1 836E -05 3. 013E -05 5.. 488E -05 1. ,176E-04 3,.073E -04  5 .929E+00 8. 983E+00 1 474E+01 . 2.. 741E + 01 6. 035E+01  1 . 124E-01 1 . 172E-01 1. 242E-01 . 1. ,355E-01 1. ,534E-01  1 . 287E+00 3 .952E- 11 1 . 280E + 00 5 .806E- 1 1 1 .272E+00 9 .160E. 11 1 . 263E + 00 1 617E10 . 1 .255E+00 3,. 344E- 10  1 . 757E -03 1. ,601E -03 1. 293E . -03 8 . 733E -04 5 .459E -04  41 43 45 47 49  3..411E+14 6.. 268E+14 9,.705E+14 1...295E+15 1 561E+15  8. 201E -04 1 923E . -03 4. 048E -03 7 .. 770E -03 1, 367E . -02  1. 356E+02 . 1. .768E-01 2. 616E+02 2. 047E-01 4.. 243E + 02 2. 407E-01 5. 918E+02 2. 886E-01 7 ..430E + 02 3. 486E-01  1 . 252E + 00 1 . 249E + 00 1 . 246E+00 1 .244E+00 1 . 243E + 00  7 .089E- 10 1. .300E- 09 2. 009E- 09 2. 676E- 09 3,.221E- 09  4.. 500E -04 4..994E -04 7. 036E -04 1. 204E , -03 2.. 193E-03  51 53 55 57 59  3. 607E+03 3. 814E+03 4. 094E+03 4. 433E+03 4. 826E+03  1. 752E+15 , 1 850E+15 . 1 877E+15 . 1 938E+15 . 2.. 170E+15  2.. 367E -02 4. 16 IE -02 7 .098E -02 1 066E . -01 -01 1 719E .  8.. 724E + 02 4 .258E-01 9. 740E+02 5.. 369E-01 1 061E+03 . 7. 090E-01 1 186E+03 . 9. 687E-01 1 446E+03 . 1 ..365E+00  1 . 241E+00 1 . 241E+00 1 . 240E + 00 1 . 239E+00 1 . 239E+00  3 .611E3. 810E3. 864E3, 987E4 . 464E-  09 09 09 09 09  4.. 271E-03 8. 838E -03 -02 1 349E . 1 .147E -02 9.. 166E-03  61 63 65 67 69  5. 307E+03 5. 884E+03 6. 555E+03 7 . 324E+03 8. 108E+03  2. 494E+15 4 . 984E -01 2. 584E+15 . 2. 141E+00 2. 465E+15 9 . 406E+00 2. 262E+15 3. 749E+01 2. 060E+15 1 160E+02 .  1 2. 2. 2. 2.  . 827E+03 100E+03 231E+03 287E+03 306E+03  2. 004E+00 3. 034E+00 4. 680E+00 7 ..295E+00 1 097E+01 .  1 .239E+00 1. 238E . + 00 1.234E+00 . 1,219E+00 . 1.177E+00 .  5.. 129E-09 5..312E- 09 5. 051E- 09 4. 577E- 09 4, 025E- 09  1 455E . -02 4. 338E -02 -01 1 523E . 5. 757E -01 2. 044E+00  71 73 75 77 79  8. 8. 9. 9.  1 938E+15 . 1 860E+15 . 1 831E+15 . 1 823E+15 .  2. 2. 2. 2.  309E+03 310E+03 310E+03 310E+03  1 1 1 1  1, .124E+00 1 073E+00 , 1.051E+00 . 1 044E+00 ,  3. 616E- 09 3. 313E- 09 3. 195E- 09 3. 160E- 09  4. 452E+00 7. 251E+00 8. 603E+00 9 .024E+00  81 83 85 87  7 .946E-03 1 .247E-02 1 .950E-02 3 .029E-02 4 .648E-02  1 . 928E -01 1 . 846E -01 1 . 766E -01 1 . 688E -01 1 .614E -01  1 . 789E+03 1 .797E+03 1 .807E+03 1 .821E+03 1 .839E+03  2 .254E+10 3 .531E+10 5 .525E+10 8 .589E+10 1 .317E+11  1 .080E -08 1 .649E -08 2 . 539E -08 3 .931E -08 6 .077E -08  5 .568E-03 8 .759E-03 1 . 378E-02 2 . 159E-02 3 . 345E-02  8 .557E-02 8 .638E-02 8 . 719E-02 8 .800E-02 8 .882E-02  1 . 289E + 00 1 . 307E+00 1 .325E+00 1 .342E+00 1 .354E+00  11 13 15 17 19  1 .000E-05 1 .585E-05 2 .512E-05 3 .981E-05 6 . 310E.-05  7 .002E-02 1 .032E-01 1 .488E-01 2 .097E-01 2 .899E-01  1 . 542E -01 1 .475E -01 1 . 411E -01 1 . 351E-01 1 . 294E -01  1 .861E+03 1 .886E+03 1 .912E+03 1 .940E+03 1 .967E+03  1 .981E+11 2 .912E+11 4 . 180E+11 5 .868E+11 8 .074E+11  9 . 312E-08 1 . 407E -07 2 .090E -07 3 .047E -07 4 . 368E -07  5 090E-02 7 .580E-02 1 . 103E-01 1 .571E-01 2 . 193E-01  8 .966E-02 9 .052E-02 9 . 141E-02 9 .233E-02 9 .329E-02  1. 1. 1. 1. 1.  21 23 25 27 29  1 .000E-04 1 . 585E-04 2 .512E-04 3 .981E-04 6 . 310E-04  3 .940E-01 5 .279E-01 6 .993E-01 9 .185E-01 1.200E+00 .  1 . 239E -01 1 . 186E-01 1 . 135E-01 1 .085E -01 1 .035E -01  1 .995E+03 2 .023E+03 2 .050E+03 2 .078E+03 2 . 106E + 03  1 .092E+12 1 .457E+12 1 .921E+12 2 .512E+12 3 .268E+12  6 . 162E-07 8 . 581E-07 1 . 183E-06 1. 620E . -06 2 . 210E-06  3 .009E-01 4 .068E-01 5 . 438E-01 7 .207E-01 9 .502E-01  31 33 35 37 39  1 .000E-03 1 .585E-03 2 .512E-03 3 .981E-03 6 . 310E-03  1.566E+00 . 2 .050E+00 2,.704E+00 3 :620E+00 4 .961E+00  9.. 855E -02 9 . 349E -02 8 . 822E -02 8 . 262E -02 7 .650E -02  2 . 135E + 03 2 . 166E+03 2 . 198E + 03 2 . 234E+03 2..275E+03  4/244E+12 5 .527E+12 7 .253E+12 9 .652E+12 1 .314E+13  3..018E -06 4.. 145E-06 5.. 764E -06 8 . 178E-06 1. .197E-05  41 43 45 47 49  1 .000E-02 1 .585E-02 2 .512E-02 3 .981E-02 6 . 310E-02  7 .040E+00 1 051E+01 . 1 694E+01 . 3. 080E+01 6 . 580E+01  6 .963E -02 6 . 164E-02 5.. 195E-02 3..950E -02 2.. 317E-02  2 . 323E+03 2 . 382E + 03 2. 461E+03 2.. 574E+03 2.. 725E+03  1 .849E+13 2 .731E+13 4..337E+13 7.. 715E+13 1.604E+14 .  51 53 55 57 59  1 .000E-01 1.585E-01 . 2 .512E-01 3 .981E-01 6 .310E-01  1, 432E+02 , 2.. 687E + 02 4.. 269E + 02 5.. 869E + 02 7 .. 294E + 02  5..975E -03 -8. . 291E -03 -1 .904E -02 -2, .666E -02 -3 . 203E -02  2,. 880E+03 3. 023E+03 3.. 166E+03 3,, 309E+03 3.. 448E + 03  61 63 65 67 69  1 .000E+00 1.585E+00 . 2..512E+00 3..981E+00 6 . 310E + 00  8.. 502E+02 9. 445E+02 1 025E+03 . 1 140E+03 . 1 376E+03 .  -3.. 595E -02 -3. 876E -02 -4. . 108E-02 -4. 434E -02 -5. 055E -02  71 73 75 77 79  1.000E+01 . 1.585E+01 . 2. 512E+01 3. 981E+01 6..043E+01  1 717E+03 . 1 958E+03 . 2. 075E+03 2. 126E+03 2. 145E+03  -5. 844E -02 -6. . 354E -02 -6. 603E -02 -6. 720E -02 -6. 770E -02  81 83 85 87  7. 805E+01 9. 233E+01 9. 824E+01 1 00OE+02 .  2. 2. 2. 2.  -6. -6. -6. -6.  633E+03 997E+03 136E+03 176E+03  2. 150E+02 3. 100E+02 3. 514E+02 3. 639E+02  411E+01 . 665E+01 . 770E+01 . 802E+01 .  4 .825E7 .661E1 . 216E1 .914E2 .962E-  ChiRM 1 . 281E-04 1 . 304E -04 1 . 334E -04 1 . 390E -04 1 . 490E -04  1 .000E-06 1 .585E-06 2 .512E-06 3 .981E-06 6 .310E-06  787E -02 795E -02 797E -02 798E -02  Rho 14 14 13 13 13  1 3 5 7 g  151E+03 153E+03 154E+03 154E+03  Page 220  quantities  9 .269E1 .202E1 .571E2 .082E2 .822E-  Mode 1 01310191  SS 3000./ 0 . 0 0 /  TauRM  RE res  TE r  r  HE res  EPS rms  DE res  Hconv/H  Ch i_H/RM  g_eff/g  Lum/L*  6 2 9 1 1  329E827E945E921E380E-  08 10 11 11 11  9 9 9 9 9  971E -01 963E -01 954E -01 944E -01 936E -01  1 1 2 2 2  497E+02 837E+02 271E+02 641E+02 790E+02  0 0 0 0 0  000E+00 000E+00 000E+00 000E+00 000E+00  9 9 9 9 9  982E-01 981E-01 981E-01 981E-01 981E-01  1 3 5 7 9  10 -4 360E- 12 10 - 1 848E- 12 10 -2 754E- 13 10 1 292E- 12 10 1 639E- 12  2 3 3 3 2  820E562E695E150E315E-  11 11 11 11 11  9 9 9 9 9  932E -01 931E -01 933E -01 935E -01 939E -01  2 2 2 1 1  683E+02 385E+02 006E+02 608E+02 242E+02  0 0 0 0 0  000E+00 000E+00 000E+00 000E+00 000E+00  9 9 9 9 9  981E-01 981E-01 981E-01 981E-01 981E-01  1 1 13 15 17 19  401E883E843E682E349E-  10 1 11 1 11 9 11 2 1 1-6  383E176E032E529E134E-  12 12 13 13 13  1 8 7 4 3  342E509E513E485E208E-  11 12 12 12 12  9 9 9 9 9  944E -01 949E -01 955E -01 960E -01 964E -01  9 6 4 3 2  320E+01 828E+01 931E+01 542E+01 552E+01  0 0 0 0 0  000E+00 000E+00 000E+00 000E+00 000E+00  9 9 9 9 9  980E-01 980E-01 980E-01 980E-01 980E-01  21 23 25 27 29  352E460E144E028E004E-  1 1-1 11 -2 1 1-3 10 -5 10 -7  554E646E787E^ 289E314E-  12 12 12 12 12  4 6 8 9 9  171E192E766E864E640E-  12 12 12 12 12  9 9 9 9 9  969E -01 972E -01 976E -01 979E -01 982E -01  1 1 1 8 6  857E+01 374E+01 040E+01 122E+00 583E+00  0 0 0 0 0  000E+00 OOOE+00 000E+00 OOOE+00 OOOE+00  9 9 9 9 9  980E-01 980E-01 980E-01 980E-01 979E-01  31 33 35 37 39  15 - 1 748E- 15 5 360E- 1 1-1 15 1 224E- 14 -7 794E- 1 1-2 14 5 001E- 14 -2 520E- 10 -3 14 1 322E- 13 8 975E- 10 1 14 1 417E- 13 3 346E- 09 4  173E207E257E043E527E-  11 11 11 10 10  5 7 2 8 3  255E759E434E654E234E-  12 12 11 11 10  9 9 9 9 9  985E -01 988E -01 990E -01 993E -01 996E -01  5 5 4 5 4  588E+00 019E+00 873E+00 064E+00 742E+00  0 0 0 0 0  OOOE+00 OOOE+00 OOOE+00 OOOE+00 OOOE+00  9 9 9 9 9  979E-01 979E-01 978E-01 977E-01 976E-01  41 43 45 47 49  1 524E-2 183E- 09 1 307E-3 476E- 09 -5 841E- 09 1 157E- 1 078E- 09 -4 801E3 865E- 10 -2 612E-  10 10 11 11 11  2 3 5 1 3  096E332E595E034E710E-  10 10 10 10 11  9 9 9 9 9  998E -01 998E -01 998E -01 997E -01 995E -01  3 2 2 1 1  537E+00 687E+00 230E+00 793E+00 374E+00  0 0 0 0 0  000E+00 OOOE+00 OOOE+00 OOOE+00 OOOE+00  9 9 1 1 1  977E-01 986E-01 002E+00 010E+00 014E+00  51 53 55 57 59  498E623E101E082E059E-  11 11 11 10 11  7 1 6 2 1  856E471E678E225E845E-  11 10 12 10 10  9 9 9 9 9  992E -01 985E -01 978E -01 982E -01 986E -01  1 1 1 1 9  169E+00 105E+00 089E+00 002E+00 876E-01  4 1 1 3 0  154E- 11 186E-06 050E-05 194E-07 OOOE+00  1 1 1 1 1  013E+00 012E+00 010E+00 012E+00 013E+00  61 63 65 67 69  1 273E- 1 266E- 1 149E-8 046E-2 793E-  11 11 11 12 12  3 1 9 2 2  774E281E491E606E837E-  10 1 1 12 12 13  9 9 9 9 6  978E -01 933E -01 766E -01 115E -01 866E -01  9 1 1 1 1  968E-01 001E+00 004E+00 004E+00 001E+00  1 8 1 2 3  655E-06 752E-03 301E+00 453E+01 837E+02  1 1 1 1 1  009E+00 006E+00 004E+00 ' 002E+00 OOOE+00  71 73 75 77 79  -4 -1 -2 -2  13 13 15 17  4 1 3 2  773E421E867E382E-  14 3 179E -01 14 - 1 108E -01 15 -3 180E -01 15 -3 823E -01  1 1 1 1  000E+00 000E+00 000E+00 000E+00  6 8 0 0  478E+03 883E+06 OOOE+00 OOOE+00  1 1 1 1  000E+00 000E+00 OOOE+00 000E+00  81 83 85 87  1 3 5 7 9  1 1 2 3 6  000E-06 585E-06 512E-06 981E-06 310E-06  1 1 1 1 1  075E+13 067E+13 060E+13 053E+13 046E+13  4 2 2 3 3  661E 641E 934E 375E 16 1E  1T 15 15 15 15  2 1 9 5 2  139E536E239E093E936E-  12 6 607E- 07 0 OOOE+00 12 -2 951E- 09 -5 209E- 1 1 13 - 1 038E- 09 - 1 745E- 1 1 13 -2 003E- 10 - 1 067E- 1 1 13 1 439E- 10 -6 763E- 12  11 13 15 17 19  1 1 2 3 6  000E-05 585E-05 512E-05 981E-05 310E-05  1 1 1 1 1  040E+13 034E+13 028E+13 023E+13 018E+13  2 1 1 9 6  641E 978E 461E 985E 241E  15 1 15 1 15 5 16 1 16 -6  769E024E025E526E393E-  13 13 14 14 15  2 3 3 3 2  944E719E858E288E416E-  21 23 25 27 29  1 1 2 3 6  000E-04 585E-04 512E-04 981E-04 310E-04  1 1 1 9 9  013E+13 008E+13 003E+13 988E+12 943E+12  4 4 4 2 2  320E 354E 118E 788E 288E  16 16 16 16 16  - 1 761E-2 231E-2 373E-2 274E-2 086E-  14 14 14 14 14  1 8 7 4 3  31 33 35 37 39  1 1 2 3 6  000E-03 585E-03 512E-03 981E-03 310E-03  9 9 9 9 9  899E+12 853E+12 805E+12 755E+12 700E+12  2 3 5 8 1  695E 680E 047E 056E 531E  16 16 16 16 15  - 1 886E- 1 686E- 1 446E- 1 155E-7 939E-  14 14 14 14 15  4 6 9 1 1  41 43 45 47 49  1 1 2 3 6  000E-02 585E-02 512E-02 981E-02 310E 02  9 9 9 9 9  638E+12 566E+12 479E+12 366E+12 219E+12  3 9 2 6 8  534E 183E 552E 637E 227E  51 53 55 57 59  1 1 2 3 6  000E-01 585E-01 512E-01 981E-01 310E-01  9 8 8 8 8  064E+12 936E+12 839E+12 770E+12 722E+12  6 4 9 9 8  1 19E 14 6 597E901E 14 3 313E343E 14 9 164E808E 14 -7 650E460E 14 -7 735E-  14 14 15 14 14  61 63 65 67 69  1 1 2 3 6  000E+00 585E+00 512E+00 981E+00 310E+00  8 8 8 8 8  687E+12 661E+12 640E+12 611E+12 555E+12  1 1 3 3 1  1 12E 13 227E 13 852E 14 452E 14 361E 14  -3 -1 -1 -1 -2  914E623E276E064E661E-  14 8 199E- 10 -2 14 1 535E- 09 4 14 -6 648E- 1 1 2 14 -2 320E- 09 -1 15 - 1 926E- 09 -6  71 73 75 77 79  1 1 2 3 6  000E+01 585E+01 512E+01 981E+01 043E+01  8 8 8 8 8  484E+12 438E+12 416E+12 405E+12 401E+12  3 5 1 3 4  509E 526E 941E 128E 061E  15 16 15 15 15  1 -3 -1 -2 -3  827E839E472E086E373E-  16 3 940E- 09 16 1 331E- 10 15 -9 842E- 11 15 -2 599E- 11 15 -9 861E- 13  81 83 85 87  7 9 9 1  805E+01 233E+01 824E+01 000E+02  8 8 8 8  399E+12 398E+12 398E+12 398E+12  4 4 3 2  356E 357E 89 TE 381E  15 15 15 15  -3 -3 -3 -4  724E957E936E153E-  15 15 15 15  :  Page 221  Convergence checks  CN, T i O , H20 (sm)  -3.00  2 9 2 1  206E027E144E228E-  13 14 15 16  445E088E225E204E-  Model 01310191  SS 3000./ 0 . 0 0 /  TauRM  Cp/Cv  -3.00  Convective q u a n t i t i e s  CN, T i O , H20 (sm) DELrad  Qconv  H_p  DELbub  DELad  Page 222  Vconv  TauRMb  Hconv  Hconv/H  1 3 5 7 9  1 1 2 3 6  000E-06 585E-06 512E-06 981E-06 310E-06  1 .667E+00 1 648E+00 1 586E+00 1 503E+00 1 422E+00  2 1 1 7 5  687E+12 743E+12 150E+12 824E+11 568E+1 1  2 2 2 2 3  060E+00 362E+00 650E+00 879E+00 022E+00  6 1 1 2 2  759E-03 066E-02 488E-02 011E-02 554E-02  1 1 1 1 1  540E -01 500E -01 392E -01 239E -01 079E -01  0 0 0 0 0  OOOE+OO 000E+00 000E+00. 000E+00 OOOE+OO  2 2 2 3 3  657E-05 787E-05 983E-05 330E-05 930E-05  0 0 0 0 0  OOOE+OO OOOE+OO OOOE+OO OOOE+OO OOOE+OO  0 0 0 0 0  OOOE+OO OOOE+OO OOOE+OO OOOE+OO OOOE+OO  0 0 0 0 0  OOOE+OO OOOE+OO OOOE+OO OOOE+OO OOOE+OO  1 3 5 7 9  11 1.3 15 17 19  1 1 2 3 6  000E-05 585E-05 512E-05 981E-05 310E-05  1 1 1 1 1  356E+00 306E+00 269E+00 243E+00 224E+00  4 3 2 2 2  183'E+11 324E+11 783E+11 434E+11 204E+11  3 3 3 2 2  081E+00 073E+00 017E+00 936E+00 840E+00  3 3 3 4 4  068E-02 514E-02 8B6E-02 145E-02 347E-02  9 8 7 7 6  399E -02 328E -02 567E -02 056E -02 735E -02  0 0 0 0 0  OOOE+OO 000E+00 OOOE+OO OOOE+OO OOOE+OO  4 6 9 1 2  935E-05 575E-05 229E-05 349E-04 031E-04  0 0 0 0 0  OOOE+OO OOOE+OO OOOE+OO OOOE+OO OOOE+OO  0 0 0 0 0  OOOE+OO OOOE+OO OOOE+OO OOOE+OO OOOE+OO  0 0 0 0 0  OOOE+OO OOOE+OO OOOE+OO OOOE+OO OOOE+OO  11 13 15 17 19  21 23 25 27 29  1 1 2 3 6  000E-04 585E-04 512E-04 981E-04 310E-04  1 1 1 1 1  210E+00 199E+00 191E+00 185E+00 180E+00  2 1 1 1 1  048E+11 942E+11 869E+11 818E+11 783E+11  2 2 2 2 2  742E+00 644E+00 549E+00 456E+00 365E+00  .4 4 4 4 4  510E-02 600E-02 723E-02 817E-02 921E-02  6 6 6 6 6  549E -02 465E -02 455E -02 504E -02 602E -02  0 0 0 0 0  000E+00 000E+00 000E+00 000E+00 000E+00  3 4 7 1 1  120E-04 854E-04 601E-04 191E-03 859E-03  0 0 0 0 0  OOOE+OO OOOE+OO OOOE+OO OOOE+OO OOOE+OO  0 0 0 0 0  OOOE+OO OOOE+OO OOOE+OO OOOE+OO OOOE+OO  0 0 0 0 0  OOOE+OO OOOE+OO OOOE+OO OOOE+OO OOOE+OO  21 23 25 27 29  31 33 35 37 39  1 1 2 3 6  000E-03 585E-03 512E-03 981E-03 310E-03  1 1 1 1 1  177E+00 174E+00 173E+00 172E+00 173E+00  1 1 1 1 1  760E+11 745E+11 737E+11 735E+1 1 739E+1 1  2 2 2 2 1  276E+00 187E+00 097E+00 003E+00 903E+00  5 5 5 5 5  025E-02 148E-02 272E-02 434E-02 625E-02  6 6 7 7 7  746E -02 936E -02 180E -02 492E -02 902E -02  0 0 0 0 0  OOOE+OO OOOE+OO OOOE+OO 000E+00 000E+00  2 4 6 9 1  876E-03 394E-03 598E-03 702E-03 392E-02  0 0 0 0 0  OOOE+OO OOOE+OO OOOE+OO OOOE+OO OOOE+OO  0 0 0 0 0  OOOE+OO OOOE+OO OOOE+OO OOOE+OO OOOE+OO  0 0 0 0 0  OOOE+OO OOOE+OO OOOE+OO OOOE+OO OOOE+OO  31 33 35 37 39  41 43 45 47 49  1 1 2 3 6  000E-02 585E-02 512E-02 981E-02 310E-02  1 1 1 1 1  175E+00 180E+00 191E+00 216E+00 260E+00  1 1 1 1 1  749E+1 767E+1 795E+1 841E+1 894E+1  1 1 1 1 1  1 1 1 1 1  794E+00 671E+00 526E+00 365E+00 239E+00  5 6 6 7 6  896E-02 316E-02 868E-02 379E-02 970E-02  8 9 1 1 1  463E -02 294E -02 067E -01 315E -01 677E -01  0 0 0 0 0  OOOE+OO 000E+00 ODOE+00 OOOE+OO OOOE+OO  1 2 3 4 5  943E-02 627E-02 403E-02 160E-02 532E-02  0 0 0 0 0  OOOE+OO OOOE+OO OOOE+OO OOOE+OO OOOE+OO  0 0 0 0 0  OOOE+OO OOOE+OO OOOE+OO OOOE+OO OOOE+OO  0 0 0 0 0  OOOE+OO OOOE+OO OOOE+OO OOOE+OO OOOE+OO  41 43 45 47 49  51 53 55 57 59  1 1 2 3 6  000E-01 1 304E+00 585E-01 . 1 350E+00 512E-01 1 409E+00 981E-01 1 476E+00 310E-01 1 536E+00  1 1 2 2 2  938E+1 1 981E+11 034E+11 096E+11 162E+11  1 1 1 1 1  173E+00 129E+00 088E+00 056E+00 035E+00  6 8 1 1 2  949E-02 351E-02 145E-01 561E-01 243E-01  2 2 2 3 3  001E -01 308E -01 679E -01 063E -01 379E -01  0 0 0 0 0  OOOE+OO OOOE+OO OOOE+OO OOOE+OO OOOE+OO  9 2 4 1 2  894E-02 058E-01 599E-01 081E+00 444E+00  0 0 0 0 0  OOOE+OO OOOE+OO OOOE+OO OOOE+OO OOOE+OO  0 0 0 0 0  OOOE+OO OOOE+OO OOOE+OO OOOE+OO OOOE+OO  0 0 0 0 0  OOOE+OO OOOE+OO OOOE+OO OOOE+OO OOOE+OO  51 53 55 57 59  61 63 65 67 69  1 1 2 3 6  OOOE+OO 585E+00 512E+00 981E+00 310E+00  1 1 1 1 1  590E+00 632E+00 654E+00 662E+00 658E+00  2 2 2 2 2  246E+11 363E+11 527E+11 719E+11 922E+11  1 1 1 1 1  021E+00 012E+00 008E+00 004E+00 002E+00  3 6 8 5 3  843E-01 918E-01 505E-01 695E-01 827E-01  3 3 3 3 3  642E -01 833E -01 936E -01 969E -01 955E -01  3 6 8 5 0  843E-01 913E-01 486E-01 694E-01 OOOE+OO  5 1 2 1 1  544E+00 273E+01 107E+01 989E+01 913E+01  2 6 1 3 0  022E+02 187E+03 287E+04 995E+03 OOOE+OO  1 4 4 1 0  633E-02 691E+02 173E+03 278E+02 OOOE+OO  4 1 1 3 0  154E-11 186E-06 050E-05 194E-07 OOOE+OO  61 63 65 67 69  71 73 75 77 79  1 1 2 3 6  000E+01 585E+01 512E+01 981E+01 043E+01  1 1 1 1 1  631E+00 551E+00 395E+00 250E+00 196E+00  3 3 3 4 5  162E+11 471E+11 862E+11 361E+11 003E+11  1 1 1 1 1  004E+00 014E+00 055E+00 195E+00 547E+00  5 1 3 8 2  405E-01 246E+00 093E+00 487E+00 731E+01  3 3 2 1 1  845E -01 492E -01 679E -01 675E -01 078E -01  5 1 1 5 2  401E-01 158E+00 253E+00 573E-01 295E-01  3 1 4 1 6  776E+01 280E+02 754E+02 839E+03 587E+03  6 1 5 1 2  349E+03 062E+05 244E+05 233E+06 778E+06  6 3 5 1 1  1 655E-06 820E+02 647E+06 8 752E-03 1 301E+00 451E+08 030E+10 2 453E+01 613E+1 1 3 837E+02  71 73 75 77 79  81 83 85 87  7 9 9 1  805E+01 233E+01 824E+01 OOOE+02  1 1 1 1  200E+00 216E+00 223E+00 226E+00  5 6 6 6  583E+11 101E+11 329E+11 398E+11  1 2 2 2  950E+00 1 198E+02 299E+00 1 149E+04 439E+00 -1 381E+02 480E+00 - 1 160E+02  8 8 7 7  882E -02 17 1E -02 993E -02 950E -02  1 4 0 0  716E-01 845E-01 OOOE+OO OOOE+OO  1 2 2 2  438E+04 345E+04 783E+04 919E+04  6 7 0 0  925E+06 704E+07 OOOE+OO OOOE+OO  2 3 0 0  725E+12 737E+15 OOOE+OO OOOE+OO  81 83 85 87  6 8 0 0  478E+03 883E+06 OOOE+OO OOOE+OO  Model 02310191 TauRM  SS 3000./ 2:00/ - 1.00 Col Mass  r/R*  -1  Summary of p h y s i c a l  CN, T i O , H20 (sm) T  n  Pe  P_gas  P_rad  Page 223  quantities Mu  ChiRM  Rho  1 3 5 7 9  1 .000E-06 1 .585E-06 2 .512E-06 3 .981E-06 6 .310E-06  4..267E-03 6..296E-03 9,.213E-03 1 .323E-02 1 .856E-02  1 .476E 1 .433E 1 . 392E 1 .355E 1 . 320E  03 03 03 03 03  1 .912E+03 1 .909E+03 1 .913E+03 1 .920E+03 1 .930E+03  1 .612E+12 2 . 381E+12 3 .478E+12 4 .973E+12 6 .942E+12  6 .830E -07 9 . 321E-07 1 .278E -06 1 . 741E-06 2 . 350E -06  4 .253E-01 6 .276E-01 9 . 185E-01 1 . 318E + 00 1 .850E+00  1 .076E-01 1 .077E-01 1 .078E-01 1 .078E-01 1 .080E-01  1 . 545E+00 1 .622E+00 1 .681E+00 1 .728E+00 1 .763E+00  4..136E6 ,, 412E9 .709E. 1 .427E2..032E-  12 12 12 1 1 1 1  2 . 800E -04 2 .989E -04 3.. 375E -04 3 .956E -04 4 . 785E -04  1 3 5 7 9  1 1 13 15 17 19  1 000E-05 1 .585E-05 2 .512E-05 3 .981E-05 6 .310E-05  2 .545E-02 3..415E-02 4..493E-02 5 .811E-02 7 .404E-02 ,  1 . 288E 1 .259E 1 .231E 1 .205E 1 . 180E  03 03 03 03 03  1 .943E+03 1 .958E+03 1.975E+03 , 1 .994E+03 2 .014E+03  9 .459E+12 1 .260E+13 1 .643E+13 2 . 105E+13 2 .655E+13  3 . 149E-06 4 . 191E-06 5 . 547E -06 7 . 306E -06 9 . 583E -06  2 .538E+00 3 . 405E + 00 4 .480E+00 5 .794E+00 7 . 383E+00  1 .081E-01 1 .083E-01 1 .086E-01 1 .089E-01 1 .093E-01  1 .787E+00 1 .801E+00 1 . 808E+00 1 . 807E+00 1 .800E+00  2 .807E3..768E4. 932E6 .315E. 7 . 934E-  11 1 1 1 1 1 1 1 1  5 .940E -04 7 .529E -04 9 . 703E -04 1 .266E -03 1 .664E -03  1 1 13 15 17 19  21 23 25 27 29  1 .000E-04 1 .585E-04 2 .512E-04 3 .981E-04 6 .310E-04  9..319E-02 1.161E-01 . 1 .437E-01 1.771E-01 , 2..178E-01  1 . 156E 1 . 132E 1 . 109E 1 .086E 1 .062E  03 03 03 03 03  2 .037E+03 2 .061E+03 2 .087E+03 2 . 114E + 03 2 . 143E+03  3 . 305E+13 4 .070E+13 4 .975E+13 6 .051E+13 7 .344E+13  1 . 253E -05 1 .631E -05 2 . 118E-05 2 . 745E -05 3 . 558E -05  9 .293E+00 1. .158E + 01 .1 .433E+01 . 1.766E+01 . 2 .173E+01  1 .097E-01 1 .103E-01 1 . 109E-01 1 . 1 16E-01 1 . 125E-01  1 . 787E + 00 1 . 770E+00 1 .749E+00 1 .725E+00 1 .698E+00  9 . 807E- 1 1 1. .196E-10 1. .445E- 10 1. .733E- 10 2..071E- 10  2 . 200E 0 3 2 .907E . -03 3..825E -03 4 .990E -03 6 .427E -03  21 23 25 27 29  31 33 35 37 39  1 .000E-03 1 .585E-03 2 .512E-03 3 .981E-03 6 .310E-03  2,.685E-01 3..326E-01 4..161E-01 5..283E-01 6, 850E-01  1 .037E 1 .01 IE 9 .825E 9 .510E 9 . 155E  03 03 04 04 04  2.. 174E+03 2 . 207E+03 2.. 243E+03 2.. 282E+03 2.. 325E+03  8 .921E+13 1 .089E+14 1 .340E+14 1 .673E+14 2 .130E+14  4 . 620E -05 6 .029E -05 7 .944E -05 1 .063E -04 1 .455E -04  2 .678E+01 3.. 318E + 01 4..151E+01 5,.271E+01 6..835E+01  1 . 135E-01 1 .147E-01 1 . 161E-01 1 .179E-01 1 .201E-01  1 .669E+00 1 . 638E+00 1 .607E+00 1 .574E+00 1 . 543E + 00  2. 472E2. 962E3.. 576E4 ,374E, 5. 455E-  10 10 10 10 10  8 . 132E-03 1.005E . -02 -02 1. 207E . 1. .398E -02 1. 556E . -02  31 33 35 37 39  41 43 45 47 49  1.000E-02 , 1.585E-02 . 2..512E-02 3..981E-02 6,.310E-02  9..135E-01 1.262E+00 . 1 819E+00 . 2..748E+00 4..362E+00  8 .746E 8 . 266E 7..699E 7 .029E 6 . 240E  04 04 04 04 04  2.. 372E+03 2. 426E+03 2..488E+03 2.. 560E + 03 2,.645E+03  2 .783E+14 3..761E+14 5..286E+14 7.,761E+14 1 .193E+15  2 .055E -04 3 .027E -04 4 . 689E -04 7 .694E -04 1 .343E -03  9.. 116E+01 1. 260E+02 . 1 816E+02 . 2.. 743E+02 4.. 355E+02  1 . 229E-01 1 .266E-01 1 .316E-01 1 . 384E-01 1 .480E-01  1 .511E+00 1 .481E+00 1 .451E+00 1 .422E+00 1 .394E+00  6.,984E- 10 9 . 247E- 10 1. .274E- 09 1 833E. 09 2.. 761E- 09  -02 1.656E . 1. 684E . -02 1.637E . -02 -02 1, 526E . 1. .369E -02  41 43 45 47 49  51 53 55 57 59  1.000E-01 . 1.585E-01 . 2..512E-01 3..981E-01 6 .310E-01 ,  7..272E+00 1,272E+01 . 2,,315E+01 4..234E+01 7 .328E+01 ,  5.. 317E 4.. 246E 3..024E 1. 709E . 4 . 319E  04 04 04 04 05  2.. 745E+03 2.. 865E+03 3. 005E+03 3., 164E+03 3.. 345E+03  1 916E+15 . 3..212E+T5 5.. 574E+15 9. 686E+15 1.586E+16 ,  2 . 498E -03 4 .949E -03 1 .038E -02 2 . 251E-02 4 .954E 0 2  7.. 261E + 02 1 .616E-01 1. 270E . + 03 1 .810E-01 2.. 313E+03 2 .088E-01 4.. 231E+03 2 .480E-01 7.. 324E+03 3 .050E-01  1 . 366E + 00 4.. 347E- 09 1 . 340E+00 7.. 146E-09 . 08 1 . 317E + 00 1 219E1 .298E+00 2. 087E- 08 1 .280E+00 3..370E- 08  1. .183E-02 9 . 863E -03 8.. 190E-03 7 .. 377E -03 -03 7 .947E .  51 ' 53 55 57 59  61 63 65 67 69  1.OOOE+00 . 1.585E+00 . 2..512E+00 3, 981E+00 6 .310E+00 .  1.126E+02 . 1,508E+02 . 1.826E+02 . 2. 084E+02 2..354E+02  -6. .413E -1 ..423E -1 ..972E -2, .379E -2. . 787E  05 04 04 04 04  3,. 552E+03 3,. 769E + 03 4,.016E+03 4,. 322E+03 4.. 708E+03  2.. 296E+16 2..897E+16 3. 293E+16 3. 492E+16 3. 621E+16  1 .090E -01 2 . 265E -01 4 . 585E -01 9..074E -01 1.651E+00 .  1. .126E + 04 1.507E+04 . 1 826E+04 . 2..084E+04 2..354E+04  3 .882E-01 4 .989E-01 6 . 506E-01 8 . 792E-01 1 . 239E + 00  1 . 264E + 00 4..817E- 08 1 .253E+00 6. 028E- 08 1 .246E+00 6. 815E- 08 1 . 242E + 00 7 .204E- 08 1 . 240E+00 7 . 458E- 08  1, ,148E-02 2. 061E -02 4. 094E -02 -02 7 ,470E , 9 . 156E-02  61 63 65 67 69  71 73 75 77 79  1 000E+01 . 1.585E+01 . 2.,512E+01 3,.981E+01 6. 043E+01  2..768E+02 3..326E+02 3.,740E+02 3. 965E+02 4. 072E+02  -3.. 384E -4. . 126E -4. 651E -4. 943E -5,.091E  04 04 04 04 04  5.. 172E + 03 5.. 722E+03 6 ,. 361E+03 7 .094E+03 7. 846E+03  3, 876E+16 4. 210E+16 4, 259E+16 4. 049E+16 3. 760E+16  2 . 836E+00 7,.212E+00 2..752E+01 1.097E+02 . 3..572E+02  2., 768E + 04 3.. 326E + 04 3., 740E + 04 3..966E+04 4. 073E+04  1 . 807E + 00 2 .711E+00 4 .145E+00 6 .420E+00 9 .614E+00  1 . 239E + 00 1 .239E+00 1 .238E+00 1 .236E+00 1 .228E+00  8.. 809E -02 1. 396E . -01 3. 771E -01 1. .155E+00 3.. 275E+00  71 73 75 77 79  81 83 85 87  7. 805E+01 9 ,233E+01 . 9 . 824E+01 1 000E+02 .  4..111E+02 4.,130E+02 4..135E+02 4.,137E+02  -5,. -5. , -5.. -5..  150E 178E 188E 190E  04 04 04 04  8., 350E+03 8. 701E+03 8. 834E+03 8. 873E+03  3. 566E+16 3, 438E+16 3.. 391E+16 3.. 377E+16  :  7 .025E+02 1 072E+03 . 1 .247E+03 1. 301E+03 .  4.. 1 11E + 04 4.. 130E + 04 4.. 135E + 04 4,, 137E + 04  1. .234E + 01 1 .456E+01 1 .547E+01 1 .574E+01  1 .218E+00 1 .207E+00 1 .202E+00 1 .200E+00  7 .977E- 08 8. 662E- 08 8. 757E- 08 8. 308E- 08 7 .668E- 08 7 . 212E- 08 6. 890E- 08 6. 766E- 08 6.. 7 30E-08  :  6. 271E+00 9 ,639E+00 1. .130E+01 1, ,183E + 01  81 83 85 87  to to  CO  Model 02310191 TauRM  SS 3000./ 2.00/ r  -3.00  -8. .664E-5.. 305E-3..409E- 1.946E. -5..114E-  Ch i_H/RM  Hconv/H  Lum/L *  EPS rms  g_eff/g  7..733E- 09 3 .488E- 09 2,.487E- 10 8 .787E- 10 1 .802E- 09  9 .9g8E-01 9 .gg8E-oi 9 .998E-01 9 .998E-01 9 .gg8E-oi  4 . 168E+02 4 .507E+02 4 . 290E+02 3 .810E+02 3 .223E+02  0. .OOOE+OO 0 .OOOE+OO 0..OOOE+OO 0 .OOOE+OO 0 .OOOE+OO  9 .970E-01 9 .970E-01. 9 .970E-01 9 .970E-01 9 .970E-01  1 3 5 7 9  12 12 12 12 12  2 .641E- 09 2..672E- 09 09 1 406E. 4 .787E- 10 2.. 779E- 09  9 .998E-01 9 .998E-01 9 .9g8E-01 9 .998E-01 9 .998E-01  2 .628E+02 2 .083E+02 1.614E+02 1 . 230E + 02 9 .253E+01  0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 0 OOOE+OO 0 .OOOE+OO  9 ,g70E-01 9 .970E-01 9 .g70E-01 9 .970E-01 9 .g69E-01  11 13 15 17 '19  4 .002E -08 -3 .388E- 12 4 .804E -08 -8 .026E- 13 4 . 284E-08 8 .396E- 13 1 .652E- 12 3 . 100E-08 1 . 732E- 12 1 . 716E-08  3..834E- 09 4..6D2E- 09 4 . 103E-09 2 .970E- 09 1.643E. 09  9 .998E-01 9 .998E-01 9 .998E-01 9 .998E-01 9 .9g8E-01  6 .885E+01 5 .097E+01 3 . 765E+01 2 . 784E+01 2 .067E+01  4 .638E- 1 1 3 .522E-09 1 .972E-08 5 . 729E-08 1 . 184E-07  9 .969E-01 9 .969E-01 9 .969E-01 9 .969E-01 9 .96gE-01  21 23 25 27 2g  10 -8 10 -3 10 -2 10 -9 1 1-1  .070E -08 0 .OOOE+OO .641E -08 -5 .300E- 12 . 574E-09 1 .470E- 12 . 172E-09 1 .643E- 12 ..881E -08 6 .050E- 13  1 3 5 7 9  1 .OOOE-06 1 .585E-06 2 .512E-06 3 .981E-06 6 .310E-06  9 .024E+12 9 .023E+12 9 .023E+12 9 .023E+12 9 .022E+12  2..317E- 10 6..568E- 12 4. 812E- 12 7.. 139E-12 7 .055E. 12  1 1 13 15 17 19  1 .000E-05 1 .585E-05 2 .512E-05 3 .981E-05 6 .310E-05  9 .022E+12 9 .022E+12 9 .022E+12 9 .021E+12 9 .021E+12  9. 598E2.. 319E2,.764E2..472E1 .674E-  12 11 11 11 1 1  21 23 25 27 29  1 .000E-04 1 .585E-04 2 .512E-04 3 .981E-04 6 .310E-04  9 .021E+12 9 .021E+12 9 .021E+12 9 .020E+12 9 .020E+12  1.577E. 1 .181E6..363E3..673E3,.553E-  11 1.653E, 11 7 .752E12 8 .607E12 -3 .243E12 -4. . 360E-  10 1 1 12 1 1 1 1  31 33 35 37 39  1 .000E-03 1 .585E-03 2 .512E-03 3 .981E-03 6 .310E-03  9 .020E+12 2 .376E9 .020E+12 1.474E. 9 .019E+12 ' 1.027E9 .019E+12 5 .467E9 .019E+12 4..354E-  12 12 12 13 13  -3 -2 -1 -9 -2  1 1 4 . 225E-09 1 1-4 .722E -09 1 1-8 .432E -09 12 -6 .273E -09 12 -3 .414E -09  41 43 45 47 49  1 .000E-02 1 .585E-02 2 .512E-02 3 .981E-02 6 .310E-02  9 .018E+12 9 .018E+12 9 .017E+12 9 .017E+12 9 .016E+12  3 .724E3..539E3..599E4..676E6..524E-  13 13 13 13 13  -7 .558E-3..606E-7..931E-5. 894E-2. 019E-  14 13 13 13 13  51 53 55 57 59  1 .000E-01 1 .585E-01 2 .512E-01 3..981E-01 6 .310E-01  9..015E+12 9 .014E+12 9 .013E+12 9 .012E+12 9 .011E+12  1.595E, 1.319E. 6, 948E1 537E. 2..048E-  12 11 11 10 10  4,.790E-9..723E-1 ,.674E-6, 430E-8. 570E-  61 63 65 67 69  1 .OOOE+OO 1 .585E+00 2..512E+00 3 .981E+00 6 .310E+OO  9 .010E+12 9 .009E+12 9 .009E+12 9..008E+12 9..008E+12  71 73 75 77 79  1.000E+O1 . 1.585E+01 . 2..512E+01 3..981E+01 6..043E+01  81 83 85 87  7 .805E+01 . 9..233E+01 9,.824E+01 1.OOOE+02 .  1.047E. 2..516E3 .365E3..262E2 . 505E-  DE res  HE res  RE res  TE res  Page 224  Convergence checks  CN, T i O , H20 (sm)  10 -2 .757E -08 10 -2 . 790E-08 10 -1 .467E -08 10 4 .980E -09 10 2 .901E -08  -1 -4 -6 -6 -5  .906E.720E.582E.842E.137E-  12 13 14 13 13  4 .047E4..523E8,.076E6 .008E3 . 270E-  10 10 10 10 10  9 .998E-01 9 .998E-01 9 .998E-01 9 .999E-01 9 .99gE-01  1 . 546E+01 1 . 168E+01 8 .955E+00 6 .992E+00 5 . 580E+00  1 .930E-07 2 . 575E-07 2 .832E-07 2 .544E-07 1 .828E-07  9 .969E-01 9 .969E-01 9 .968E-01 9 .968E-01 9 .968E-01  31 33 35 37 3g  .111E.322E.068E.693E.939E-  14 12 12 12 13  5. 837E8,. 704E6. 806E3. 968E3.. 721E-  10 10 10 10 10  9 ,9ggE-01 9 .gggE-01 9 .gggE-01 9 ,gg9E-oi 9 .gg9E-oi  4 . 564E+00 1 .017E-07 3..827E+00 4 . 163E-08 3.. 285E + 00 1 .152E-08 2..875E+00 1.700E-09 . 2.. 554E + 00 3 881E-11  9 .968E-01 9 .968E-01 9 .968E-01 9 .967E-01 9 .967E-01  41 43 45 47 49  1 . 322E13 2. 524E -08 13 5.. 310E-08 4 .554E1 1-2. 807E -08 1 .930E11 -2. .054E -07 - 1.077E1 1 6.. 563E-08 -8. .703E-  11 11 11 10 11  2..417E- 09 5..086E- 09 2..690E- 09 1.967E. 08 6.. 289E- 09  1 .OOOE+OO 1 OOOE+OO 1 .OOOE+OO 1 .OOOE+OO 1 .OOOE+OO  2.. 297E + 00 2..077E+00 1.857E+00 . 1. 660E+00 . 1,545E+00 .  9 .966E-01 9 .966E-01 9 .965E-01 9 ,96gE-01 9 .994E-01  51 53 55 57 59  1.818E. 3..283E3,.133E5, 467E1 683E.  10 -4. 370E- 1 1 4.. 565E-07 7 .495E11 -2. 919E- 13 - 1, .407E-07 2 .906E1 .067E, 12 - 1. .349E-08 12 1 457E12 7 .320E. 12 - 1. .739E-08 -7 .047E12 9 . 616E- 13 2.. 219E-08 6 .772E-  11 12 12 12 12  4.. 372E- 08 1,348E, 08 1. 292E. 09 1,666E. 09 2.. 125E-09  1 .OOOE+OO 1 .OOOE+OO 9 ,9ggE-oi 9 .gggE-01 g .g99E-01  1. .351E + 00 0. OOOE+OO 1.. 1 12E + 002.. 133E-05 1 014E+00 . 1. 379E-02 . 9. 97gE-01 1.547E-01 , 9.. g47E-01 1 426E-01 .  1 .008E+00 1 .011E+00 1 .010E+00 1 .007E+00 1 .006E+00  61 63 65 67 69  9..007E+12 9..007E+12 9..006E+12 9..006E+12 9..006E+12  1 093E. 1 245E. 4.. 204E1 554E. 6, 421E-  13 13 14 14 15  653E. 864E917E834E233E.  14 2 . 138E-09 -8 .277E13 - 1. 168E-08 - 1.365E. 14 -6. . 865E- 10 - 1.811E14 2., 703E- 10 2 .792E14 4.. 996E- 1 1 2 .706E-  13 12 13 14 14  2..048E- 10 1.119E. 09 6..575E- 1 1 2. 589E- 11 4,. 786E- 12  9 .g99E-01 g .998E-01 9 ,9g4E-01 g..982E-01 9 ,g50E-01  9. g64E-01 1 OOOE+OO . 1 002E+00 . 1.003E+00 . 1 001E+00 .  3 .222E-02 3..723E-01 5.,600E+00 3. 607E+01 1 728E+02 .  1 .006E+00 1 003E+00 1 .002E+00 1 .001E+00 1.001E+00 .  71 73 75 77 7g  9..006E+12 9 .006E+12 9..006E+12 9..006E+12  3. 619E- 15 2. 558E- 15 2., 135E-15 1 538E. 15  6. 931E4. 724E4..129E4. 060E-  15 3.. 996E-12 5 .249E15 7,. 645E-13 1 .506E15 5. 552E -13 3 .376E15 -2, , 176E-14 - 1.923E.  15 15 16 17  3, 828E7 .327E. 5,, 322E2. 607E-  g 904E-01 9 .852E-01 g .827E-01 9. 819E-01  1 OOOE+OO . 1 OOOE+OO . 1 OOOE+OO . 9, 999E-01  4., 796E + 02 9 .701E+02 , 1, 265E , + 03 1, 365E . + 03  1 .OOOE+OO .  81 83 85 87  .876E.836E.809E.277E.701E-  9, 1 6. 2. 1  -6 .094E -09 -9 087E -09 -7 . 106E-09 -4. . 142E-09 3..885E -09  1 .231E5 .366E6 .837E2 .624E5 .594E-9 -1 -2 -2 -5  13 14 14 15  0..OOOE+OO 0. OOOE+OO 0,, OOOE + OO 0,.OOOE+OO 0..OOOE+OO  1 OOOE+OO 1 .OOOE+OO 1 OOOE+OO .  225 LO CVJ CM  O)  ^  X  > c  o u X  > c  o  o X  > c o u  >  c  n  ron 3  cr tu  C D 1—  •rCJ  cu  > c o  n  D  n  _i  LU Q  X)  o  to l_ _J LU Q  CM  o o o o o o o o o o + + + + +  O O O O o O O O O o  LU LU LU LU LU O O O O O  2 l_)  a  SS  a cj  o  o  o  o  o  o  o  o  o  o  e\j O  CO  CM  O  cu o  o o o o o  + L+U L+U L+U L+U  LU O O O  O O O O O O O O O O O O  LU O O O  o  O O O O  O O  LU LU LU LU O o o o O o o o o o o o  i  1  i  y~  s r ro  LU O O O  r^.  1—•  i  O O O OO o o o o o o  o O o o o o O o o o + + + + +  •  LU LU LU LU O i n CM s r r o r^- r o s r CD i n CO i n  o  + L+U L+U L+U L+U  t  LU LU LU LU LU CO CM CO CD s r r o C\J CVJ CO ID *~ i n CD  o  o o o o o  i  LU LU LU LU LU O O O O O  o  O O O O O o O O O o O O  i n -ro  o O o o t  o o o o o o o o o o + + + + +  LU O O O  O o o o o  o o o o  O  CO CD CO CD i n  in in in  o  + + + +  LU LU LU LU *— s t CD CD CD co CO CO CD CM O  o o o  LU LU LU LU o o o O o o o O o o o o  LU CD i-  ro  •  LU CO C\J CO  C\J CM CM  o o o o  r ~ CO CO CD •r-  O o o O »  1  o  o  LU LU LU ro co co ro co O CM CD  o  o  + + +  i —  LU h~ h~ ro  LU LU CO CO r o co  LU ro in CO  ro  r o i n CO  ro ro  •  i  •  i  i  "»  O O O O O  O O O O  CM r o CM CM CM  CM CM CM CM CM  LU O O O  LU LU O O O O O O  LU LU O O O O O O  LU LU LU LU LU >a- i n CD CD CD CO CM T — r o > o in O  LU LU LU LU LU i n i n 00 CO y- r ^ i n 00 CD «o T  O  O  O  m  o  Oi o O Oi Oi  ^  i n CD CO CO  CM r o CM CM CM  CM CM CM CM r o  CM CM CM CM CM  O O O O O  O O O O o  O O O O O  LU LU LU LU LU 00 r o CD i n n CD r~ r*~ o iCO CM r~- r^. o  LU LU LU CD O O CO > - f co c o  LU LU LU LU LU CD O •a- r» CM c o r o CD CD O O < -  LO t o  •a- * a  LU LU CM CO N o c o CD  o  o  O  ro CO  <-  LU LU LU LU LU CO r o CM CD CD CO o r ~ o 00 CM  r o i n CO —  O O O O O  CM •<3-  O  o  *»  -  CO r*.  y-  ro ro i n ro  CM  T-  t—T — T•a-  CM r o r o CM CM  O O •  •  oo  •  t  o t  LU LU LU LU CM r ~ O O O CO T - m •ain ^ t D CD  LU O O o  LU O O o  LU O O O  LU O O o  LU O O O  LU O O O  LU LU LU O CD O O Ul s r O r*. o  LU LU CO CM r*. co ro  LU LU O O O O o o  LU LU CO LO in CO  > —  i n LO LO  LU LU LU o T in o CD o r o co  LU LU r o CO CM O *~  o o o o o  o  o1 o1 o1 ot o (  o O o o o O o + + + + +  LU LU CO CD *— CD r o  ~  CM r o  O o o o o  o o o o o  o  o o o  LU CM CD •*}•  LU LU LU LU LU « a r o O r o yr o y- c o c o m oo s t o i n co  LU LU in i — r o a> CD 00  LU CO in co  LU LU LU O st o CD o c o in in st  CO CD  CM O  LU LU LU CD CD CD r o CD CD T - r o  -  O O O O O O o o o o + + + + +  O o O o O O o o o O + + + + +  o o o o o o o o o o + + + + +  LU LU y— s t y— O CM CM  O O O O O o O O O O + + + + +  LU LU LU LU LU O N CO l O co O CD CD CD CD CM v— y- y-  LU LU LU LU LU O ro CD cn ro O o o o •CM CM CM CM CM  o o o o o  in in in in in o o o o o  sf sf sf sf sf  LU LU LU LU LU O LO CM y- O O CO y- 00 v—  LU LU LU LU LU O LO CM y- O O CO y- 00 y-  O O O O O ' LU LU LU LU LU O i n CM y— O O 00 y- 00 yO m i n co r o  O  in  LO CD  ro  '-•-CMrocD  y-  CO LO N Ui  o  i n i n co r o  .-»-CMrocD  ' - ' - r o m e o  T -  o o o O O  1—  O O o O O o o o O o + + + + +  LU O CM CM  CM * -  LU O CO CD  LU  O O O o O O o O o o + + + + +  LU LU *— o s a ro CM CM  r o CM  LU LU CM o r-. CD LO  LU LU LU LU 1— O l O r*O) CM i n  LU LU r-N. r o CO o •a- c o  o O O Oo O o o o o + + + + +  O O O OO O O OOO + + + + +  in ^  LU co o N  LU LU LU i n CM i n co co i n f— r o r o  y-  o O O O o o o O O o + + + + +  LU LU LU i— O) CM CM oo i n co CO  CM CM CM  o O O O o  t— r^- c o CD y~  CD CD  -3" CD h~ r ^  CM CM CM CM CM  o o o O O  CO CO CD CO  CD CD CD CD CD  CM N . r o  LU CM CM 00  CM CM CM CM CM  C O M v  < — CD  ro  O O O O + + + +  CM <a- i n i n  LO i n CO CC CO  y-  LU LU LU CM «* CO CO r~r o •3-  LU LU CM CM ro ro ro  O  co r o r o r o  CD CD  LU in in ro  O  t-  LU r^. in O  LU LU CD CD CM CO •a-  o o O O O + + + + +  O  00  LU oo CM O  t-  ro ro ro r o  O  LU LU 1— c o c n *— CO CD  LU LU LU oy— iCnD CM O CO O  T- T -  CM CM r o  O  LU LU LU CD i n Tin co s r co *—  co r^. o CD i n s t CM h~ CD s f ro r o ro s t  to  O  r-  r*-  LU LU CM > — CD * CD CD  y—  CD ro  LU C— CO N  LU •aro O  LU O 00 CD  .LU LU LU LU LU  LU CO CM N  O o oO  LU LU >a r^. CD CO o in  y~  LU LU CM O r o (D  LU LU r o CM CM CO d) T -  LU CM ro ro  LU  CD CO  O O OO + + + +  o o o o o  LU LU O) C M r o «aro  LU LU LU LU * — >a- 00 * — O 00 «J- CM *~ o * ~ r o  CO 00 CD CD CD  r—  LU LU LO O) T ~ CD co  •f  o1 o1 o o o 1 1 1  o O O O O + + + + +  *—  LU  LU O O O  Oi O< O o1 o i 1  00 CO CO CO 00  LU LU LU LU in CD r o i n o co CD LO CM  o o1 o1 o1 o 1  CM CM CM CM  o O O O O + + + + +  CD CD ID CD  LU O O O  o1 o1 o1 o Oi 1  ro ro ro ro ro  in in i n m  LU O O O  O> O• o> O> O i  ro ro r o r o ro  T -  LU O O O  LU •sf *a  1^-  LU LU LU LU LU CO Ul CD r o O * — r o CO CD i— N. ao  ro s f s r  r - CM i- CD  LU O O O  LU LU CO 1— CO i n 00 CD CO CO CO CD  CM CM CM CM CM  LU LU LU LU LU CM y— CO LO r o •st 1— CM O o 00 h~ 1— r ~  LU in CD CD  LU O O O  LU CD O CD  i n LO CO CO h~  LU LU LU r o y~ 00 CO m r o CD *~ o  y—  LU 00 CO CO  CM CM CM CM CM  LU LU O CD r o CM  T —  + + + +  LU LU LU CD i n s r CM i n CO  O  O O O O O + + + + +  in in in in in  o O o o O o O o o o + + + + +  y— r -  O O O O O  CM CM CM CM CM  1^ *—  LU ro ro ro  , ,  CM CO CO CM CM  •st  LU LU LU •sr r o CM r o CM O  LU CD «ro  s r CD y- y-  ro  CM CM CM CM CM LU LU LU LU 1— CD 00 CO co •3co < t Co m  LU CD "3O  o O O O + + + +  CD  O• o1 o O• O 1 i  *— CM r o ^  LU LU LU LU LU o i n O CD CD CD CD o CD r *»  y-  o o o o O + + + + +  T—  •<a- TJ- -a- -<t  Oi Oi oi ot o 1  y-  CD  LU LU LU CD r^. CM O ro ro o O  o o  ++ ++ +  in ^• T—  -  co o> o o  O  + + + + + - + + + + + LU O O O  o cu O o o o O • 1 + + +  O O o O o  CM r o r o CO CM LU  in in in  o1 o O• o1 o1 1  LU LU LU LU CD in in CD O CO CO ro ro  o r o ^ LO m o o o O O + + + + +  CM  ro ro  ro ro  r o r o i n CD  O  o  o o o o o  --  LU LU LU LU LU C\J r o O CO e\j C\J O O CVJ C\J CO CO 1^  LU CO ro sr  o o o o o o o o o o + + + + +  T  O  o  i  O O O o O + + + + +  -—  O O  CM  1  r o r o r o C0 CM  LU ro CM •sf  LU O O O  CM  O O O O o + + + + +  LU LU LU LU LU CO CO CO CM o c o m CO o in ro O  LU O O O  O  ro ro ro r o ro  LU LU LU LU LU 1— i— CD CM i n i n CO i n ro CO •a- CM  LU O O O  O  O O O O O + + + + +  LU O O O  LU O O O  o  C\J rvj r o r o  O  LU LU LU LU . — r o CM CO CM •aO) i n CM *~ ~  LU O O O  O  ^ m in N cocococo  r o y-  O O O O O  LU in en ro  o O O O O o + + + + +  O  LU r^. sr LO  r o i n N cn r-~  T-  LU O O O  O O O o O t i • t i  O OO O O + + + + +  o  CM r o r o r o r o  LU CO O CM  LU LU LU LU LU O O O O O  ro  ro y— s t CO  LU LU CM ro  O LD CVJ y~ O O O o o + i • 1 1  co  r - CD y— ui co  y-  o O O O O o O O O O + + + + + O O O O O O O O O O  CM s t  LU LU LU CD O s r CD CD CM sr ro ro  r o i n i * ~ o> c o CD C D t o CD  LU LU LU LU LU CO CD CD s r o <- o o r^. i n CM CM s t  •ao1 ot o O Oi  y-  y—  r o i n r>- c n LOinioiOLO  LU LU LU O r o CD O ro ro O  o o o O  r o r o r o CM CM  r  i — i  T —  «—  •=t O O O o O i • • 1 i  "»  i  LU LU LU LU LU r o CVJ O CO i n O co  st st 1  LU LU LU LU CD ro ro r o LO m o i n LO o i n  (  T- S f  + + + + +  LU LU co ro CD O ro  T - ro in ai sr sr s t s t sr  r o y- r o r o r o  CO CO CO CO CD  cr  O o o O O  i  o  y- y-  \  CO CO CO f-. O o o O  o  LU LU LU LU r o O O CM CD i n c o O CM y- O O  >  1  o  + + + + +  o o o ro  y~  ' - n i O N C D ro ro r o ro ro  o  CD a> a> CD 00 O o O o o  o o  y~  n i n s 01 r o r u <\J CM  o  st st sf st ro  o o  ^  ro  o  CM 00 i -  > c o u  > - r i P M j  o o O O O o o o O O + + + + +  O  ro _i LU Q  fn i n s o)  O• Oi o o1 o i 1 LU LU LU CD 00 CO CM CO CO i- T- —  LU LU 1^. CD y- O 00 CO  LU *CD O  LU LU LU CM CM s t r ~ y- CM i^to  o1 Oi O Ot Oi i  L U LU LU CO "» 00 i n 00 O)  LU LU LU LU LU in O) r o CO ro O i n r*i n r o y~ CD r^-  ro r o ro ro r o  r o r o r o CM CM  O) O) CD CD CD  CD CD CD CD CO O o O O O  o o o O O + + + + + LU LU LU CO o to C CD CO O O y- , ~  LU LU O r-. y~ i n CM CM  O O O O O O O O OO + + + + + LU LU LU CD CD yy- y- CM CM CM CM  LU LU CM ro CM CM CM CM  + + + + +  LU Ul O ro  LU CO CO ro  LU CD CM st  LU O O in  LU O CO in  O O O O O O O O O O + + + + +  LU LU ro CM CM CM CM CM  LU O CM CM  LU LU CO CD  y~  LU i —  oo •sr  u> u> CD CD CD  o o O O O + + + + + LU LU LU LU LU r o CO CD CO s* r~ co CM r*CO 00 o »-  O O O O O O OO O + + + + + O  LU LU LU •sf s t CO  y— y—  r - y— y—  CM CM  CM CM CM  LU LU CD CM CM i n CM CM  ro CM CM CM CM o o o o o  LU LU LU LU LU O i n CM >- o y- CO -o i n LO cn r o  LU LU LU LU LU  O CO  o i n w - o O 00 * - 00 y~  LU LU LU LU LU O m CM y- O O 00 y- 00 y-  '-'-coroco  ' - ' - c M r o c o  o in in ro " - i - r o m c o  o  m LO o  ro  »—romr*~co  ^  m m N cn  roinr^co  romr^.cn  y— •>— *— y— y~-  CMCMCMCMCM  COrOrOrOro  s t s t s f s f s t  ^_  LU ro co CM  LU LU LU sa- r^. i n i n ro s t CD CD  CM S t CD CO  LU LU LU LU CM r o CO s t in r o y— y~ O O o  O  O  O  y- r - y-  O  _  y- -- y-  o1 o1 O O O + + +  O O OO + + + +  i n CO y- r o i n  CO ^ - -- y-  o o o o o o o o o o + + + + +  o o o o o o o o + + + +  CD CD O) CD CD  CD CD CD CD O O O O  LU CO co st  LU LU LU o oo CD CD O CO CD i —  LU LU LU LU LU CO CO r o r o i n 1— r~ CO o r o CO r o co i n  LU in o o  LU LU LU LU s t o s t 00 r o CD o o o o o  LU LU LU LU s t s f tD O y— 1^ CO CM r o o y- CM  LU LU *— ( • oo CD y- r o  LU LU If) N O yro n  CM CM  ro ro ro ro ro  O O O OO  r o r o r o r o CM  y-~ T — y— T~  LU LU CO yCM O)  LU LU LU LU O) r o CD i n 1^ O r o o s t CM CD  ro r o r o r o  *-  LU LU LU i — co t— y~ r o CO T - r o  o o o o o  CD  ^rotnr^.cD LO LO LO l O LO  CD CD CD CD CD O O O O O  + + + + +  LU LU O ro o ro i n  LU oo Ico S  LU LU y- •sr co i n co y-  o o o o o + + + + + LU S ro co  LU CO CD ro  LU LU O N 1— o r~- r o  + + + +  LU ro "» *—  CM CM CM CM CO  ro r o s t st i n  i n i n t o CD  o o o O o O o o o o + + + + +  o o o oo o o o oo + + + + +  o o o o o o o o + + + +  LU LU LU LU LU O CD s t *~ CD O h~ oo co ro ro ro ^ t i n to  LU LU LU ro r - CD i n <t co CO CD LO  o o o o o o o o o o + + + + +  o o o o o + + + + +  LU LU LU LU LU o i n CM i - o O 00 » - 00 » o in in ro  CD  LU LU s t ro N st s t ro  LU LU LU LU LU O LO CM y- r o O 00 <— CO s t  T-r-roroco  o LO i n co o .-i—CMroco  romr^cD  y— c o i n N UI  CDCDCOCOCD  f^r^P^.l^f^  LU LU LU LU COCOON  CM CM CM CM  o o o o  + + LU LU LU LU LO r o s j o o r o CM o CO CM 00 O s  oioi  ^ r o i n r ^ CO CO CO CO  Model 03310191 TauRM  SS 3000./ 0. 0/ - 3.00 r/R*  Col Mass  - 1  Summary of p h y s i c a l  CN, T i O , H20 (hm) n  T  Pe 1.,694E'-08 2,.412E-08 3. 477E-08 5, 065E-08 7 .426E-08  Page 226  quantities Mu  ChiRM  Rho  P_gas  P_rad  9.. 240E-03 1.330E-02 , 1.932E-02 . 2 .827E-02 4.. 152E-02  8..489E-02 8. 552E-02 8 616E-02 8 .681E-02 8,. 747E-02  1 .319E+00 1 .333E+00 1 .346E+00 1 .359E+00 1 .371E+00  8.. 202E- 14 1. .185E-13 1.726E, 13 2 .531E- 13 3..717E- 13  9.. 732E -05 1.055E . -04 1 . 137E-04 1 . 226E -04 1 . 335E -04  1 3 5 7 9  1 3 5 7 9  1 OOOE-06 . 1 ,. 585E-06 2.. 512E-06 3. 981E-06 6 . 310E-06  1. .317E-02 1 892E-02 . 2. 736E-02 3. 977E-02 5. 794E-02  1 1 1 1 1  928E -01 864E -01 799E -01 733E -01 668E -01  + 03 1. 788E . 1, 799E . + 03 1 811E+03 , 1.826E+03 . 1 842E+03 .  3..744E+10 5..354E+10 7 .. 724E+10 1.122E+1 . 1 1 633E+11 .  1 1. 13 15 17 19  1 .OOOE-05 1.585E-05 . 2.. 512E-05 3. 981E-05 6 ,. 310E-05  8. 419E-02 1 215E-01 . 1 ..735E-01 2. 446E-01 3. 402E-01  1 1 1 1 1  604E -01 541E -01 480E -01 421E -01 365E -01  1, 860E+03 , 1. 880E+03 . 1 900E+03 . 1. 922E+03 . 1.943E+03 ,  . 2..371E+1 1 1.091E-07 . 3, 418E+11 1.598E-07 4, 876E+11 2..323E-07 6. 864E+11 3,,341E-07 9 . 527E+1 1 4.. 747E-07  6..089E-02 8 .871E-02 1 .279E-01 1.821E-01 . 2 .556E-01  8 812E-02 8 .878E-02 8, 943E-02 9. 007E-02 9 073E-02  1 .381E+00 1 .389E+00 1 . 395E+00 1 . 399E+00 1 .401E+00  5. 437E7 .884E. 1.129E. 1.594E. 2..217E-  13 13 12 12 12  1 .475E -04 1 .659E -04 1 .904E -04 2 . 226E -04 2 .646E -04  1 1 13 15 17 19  21 23 25 27 29  1 .000E-04 1. .585E-04 2.. 512E-04 3..981E-04 6..310E-04  4. 667E-01 6.. 321E-01 8..467E-01 1,124E+00 . 1.484E+00 ,  1 1 1 1 1  31 1E-01 259E -01 209E -01 161E -01 1 13E -01  1 965E+03 , 1. 988E+03 . 2, 010E+03 2..033E+03 2. 057E+03  1.304E+12 . 1,762E+12 , 2,,354E+12 3.,117E+12 4..100E+12  6..660E-07 9.. 237E-07 1 268E-06 . 1 729E-06 2..346E-06  3 .539E-01 4 .836E-01 6 .534E-01 8 .750E-01 1 .165E+00  9 . 139E-02 9 .207E-02 9 .278E-02 9 . 355E-02 9 .439E-02  1 .403E+00 1 .403E+00 1 .401E+00 1 .399E+00 1 . 396E + 00  3..037E4 .104E5 .478E7 .242E9 .506E-  12 12 12 12 12  3 . 188E-04 3 .881E -04 4 . 752E -04 5 .822E -04 7 . 11 1E-04  21 23 25 27 29  31 33 35 37 39  1.OOOE-03 . 1.585E-03 . 2. 512E-03 3..981E-03 6.. 310E-03  1.952E+00 . 2,.567E+00 3..388E+00 4..505E+00 6..070E+00  1 1 9 9 8  065E -01 017E -01 682E -02 174E -02 637E -02  2. 082E+03 2.. 108E+03 2.. 136E+03 2.. 166E + 03 2 . 200E+03  5,,374E+12 7 .040E+12 9..249E+12 1.224E+13 , 1.639E+13 .  3..178E-06 4 .316E-06 5 901E-06 8 .167E-06 1 .152E-05  1 .545E+00 2 .049E+00 2 .727E+00 3 .660E+00 4 .977E+00  9 .532E-02 9 .639E-02 9 . 764E-02 9 .914E-02 1 .010E-01  1 .242E- 1 1 .392E+00 1 . 387E + 00 1 .622E- 1 1 . 382E + 00 2 .122E- 1 1 . 37.5E + 00 2 .793E- 1 1 .366E+00 3 .718E- 1  1 1 1 1 1  8 .625E -04 1 .034E -03 1 . 217E-03 1 .401E -03 1 .561E -03  31 33 35 37 39  41 43 45 47 49  1.000E-02 , 1.585E-02 . 2..512E-02 3..981E-02 6,. 310E-02  8,.342E+00 1.182E+01 , 1.753E+01 . 2. 813E+01 5.. 361E+01  8 7 6 5 4  057E -02 41 1E-02 663E -02 740E -02 421E -02  2 . 239E+03 2.. 285E + 03 2,. 345E+03 2.. 429E+03 2..573E+03  2..236E+13 3..139E+13 4..599E+13 7 .241E+13 . T..333E+14  1 .674E-05 2 .544E-05 4 .144E-05 7 .649E-05 1 .844E-04  6 .912E+00 9 . 903E + 00 1 .489E+01 2 .428E+01 4 .736E+01  1 .034E-01 1 .067E-01 1 . 113E-01 1 . 186E-01 1 . 325E-01  1 .356E+00 1 .343E+00 1 .327E+00 1 .306E+00 1 . 278E+00  1 1 1 1 10 10 10  1 .668E -03 1 .679E -03 1 . 550E -03 1 .218E -03 6 .887E -04  41 43 45 47 49  51 53 55 57 59  1.000E-01 . 1.585E-01 , 2..512E-01 3..981E-01 6..310E-01  1. .310E + 02 2 480E -02 2.. 855E+02 7 097E -03 4.. 803E + 02 -5 156E -03 6. 615E+02 - 1 297E -02 8.. 123E + 02 - 1 816E -02  2.. 786E + 03 2. 973E+03 3,. 138E+03 3.. 292E + 03 3. 432E+03  3..115E+14 6 ,570E+14 1 072E+15 . 1.427E+15 . 1.697E+15 .  6 . 105E-04 1 . 742E-03 3 981E-03 7 .857E-03 1. 374E-02 .  1 . 198E + 02 2 .697E+02 4 .643E+02 6 . 488E + 02 8 .043E+02  1 .592E-01 1 .918E-01 2 . 304E-01 2 . 790E-01 3 . 376E-01  1 . 259E + 00 6 .513E- 10 1 .367E- 09 1 .253E+00 1 .248E+00 2 .221E- 09 1 .245E+00 2 .951E- 09 1 . 243E + 00 3 .504E- 09  3 .881E -04 3 .993E -04 5 .903E -04 1 . 115E-03 2 .093E -03  51 53 55 57 59  61 63 65 67 69  1.OOOE+00 . 1.585E+00 , 2.. 512E + 00 3..981E+00 6 . 310E+00  9. 397E+02 1 041E+03 . 1. .124E + 03 1. 232E . + 03 1 452E+03 .  -2 -2 -2 -2 -3  197E -02 47 7E -02 699E -02 985E -02 541E -02  3.. 585E+03 3.. 783E+03 4. 054E+03 4.. 391E+03 4,, 784E+03  1.893E+15 . 1 997E+15 . 2, 020E+15 2. 052E+15 2. 240E+15  2., 340E-02 4 .084E-02 . 7..056E-02 1 .081E-01 1.672E-01 .  9 . 370E + 02 1 .043E+03 1 . 130E+03 1 . 244E+03 1 .479E+03  4 . 115E-01 5 . 170E-01 6 .811E-01 9 . 331E-01 1 .318E+00  1 .242E+00 1 .241E+00 1 . 240E + 00 1 . 239E + 00 1 .239E+00  3 .903E- 09 4 .115E- 09 4 .158E- 09 4 .223E- 09 4 .609E- 09  3 .998E -03 8 .253E -03 1 . 374E -02 1 . 249E -02 9 .526E -03  61 63 65 67 69  71 73 75 77 79  1 000E+01 . 1 ,.585E+01 2., 512E+01 3,. 981E+01 6 .043E+01  1 ..798E+03 2. 057E+03 2., 184E+03 2,. 240E+O3 2. 261E+03  -4 -4 -5 -5 -5  318E -02 853E -02 -02 1 17E 241E -02 294E -02  5,. 262E+03 5., 835E + 03 6 ,. 502E + 03 7 .. 265E+03 8. 043E+03  2. 551E+15 2. 653E+15 2. 536E+15 2.. 329E+15 2., 122E+15  4..531E-01 1.926E+00 . 8..540E+00 3..446E+01 1 079E+02 .  1 .853E+03 2 . 137E+03 2 .276E+03 2 . 336E + 03 2 . 356E + 03  1 .937E+00 2 .933E+00 4 . 528E + 00 7 .062E+00 1 062E+01  1 . 239E+00 1 .238E+00 1 .234E+00 1 .221E+00 1 . 182E + 00  5..247E- 09 5 .453E- 09 5..197E- 09 4..720E- 09 4.. 165E-09  1 . 377E -02 3 .989E -02 1 . 396E -01 5 . 252E -01 1 .859E+00  71 73 75 77 79  81 83 85 87  7 .805E+01 9 ,. 233E+01 9 . 824E+01 1 000E+02 .  2.. 267E+03 2., 270E+03 2. 271E+03 2. 271E+03  -5 -5 -5 -5  312E -02 321E -02 323E -02 324E -02  8., 564E + 03 8. 926E+03 9. 064E+03 9.. 104E+03  1 1 1 1  2..020E+02 2. 934E+02 3.. 337E + 02 3.. 459E + 02  2 . 360E + 03 2 . 361E + 03 2 . 361E + 03 2,. 361E + 03  1 . 366E+01 1 613E+01 1.715E+01 . 1 .745E+01  1 .133E+00 1 .085E+00 1 .064E+00 1 .058E+00  3 , 755E- 09 3..452E- 09 3..333E- 09 3..298E- 09  4..058E+00 6 .649E+00 7 .914E+00 . 8 .310E+00  81 83 85 87  996E+15 . 916E+15 . 886E+15 . 878E+15 .  5 .035E7 .002E1 .014E1 .570E2 .829E-  Model 03310191  SS 3000./ 0 . 0 0 /  TauRM  RE res  TE r  r  Page 227  Convergence checks  CN, T i O , H20 (hm)  -3.00  HE res  g_eff/g  EPS rms  DE r e s .  Hconv/H  Chi_H/RM  Lum/L*  2 9 8 6 2  229E- 08 518E- 12 750E- 1 1 022E- 11 349E- 12  9 9 9 9 9  983E -01 983E -01 983E -01 983E -01 983E -01  1 1 9 9 8  122E+02 041E+02 712E+01 015E+01 253E+01  0 0 0 0 0  OOOE+00 OOOE+00 OOOE+00 OOOE+00 OOOE+00  9 9 9 9 9  927E-01 927E-01 927E-01 926E-01 926E-01  1 3 5 7 9  12 12 12 12 12  4 5 3 1 1  657E319E489E846E389E-  9 9 9 9 9  983E -01 984E -01 984E -01 984E -01 984E -01  7 6 5 4 3  404E+01 490E+01 560E+O1 666E+01 844E+01  0 0 0 0 0  OOOE+00 OOOE+00 OOOE+00 OOOE+00 OOOE+00  9 9 9 9 9  926E-01 926E-01 926E-01 926E-01 926E-01  1 1 13 15 17 19  12 12 12 12 12  9 7 3 1 8  979E- 12 823E- 12 273E- 12 1 17E-11 986E- 12  9 9 9 9 9  985E -01 985E -01 986E -01 986E -01 986E -01  3 2 1 1 1  120E+01 501E+01 987E+01 572E+01 243E+01  0 0 0 0 0  000E+00 OOOE+00 OOOE+00 OOOE+00 OOOE+00  9 9 9 9 9  926E-01 926E-01 926E-01 925E-01 925E-01  21 23 25 27 29  181E630E387E752E573E-  9 9 9 9 9  987E -01 988E -01 988E -01 989E -01 990E -01  9 7 6 5 4  860E+00 875E+00 357E+00 216E+00 374E+00  0 0 0 0 0  OOOE+00 OOOE+00 OOOE+00 000E+00 OOOE+00  9 9 9 9 9  925E-01 925E-01 925E-01 925E-01 925E-01  31 33 35 37 39  327E- 07 0 OOOE+00 863E- 1 1 1 205E- 11 135E- 10 -6 561E- 12 286E- 10 -1 158E- 1 1 219E- 11 - 1 043E- 1 1  1 3 5 7 9  1 1 2 3 6  000E-06 585E-06 512E-06 981E-06 310E-06  1 1 1 1 1  075E+13 069E+13 063E+13 057E+13 051E+13  5 5 2 7 6  599E 357E 147E 798E 660E  10 14 14 15 15  1 1 6 3 1  460E023E312E358E538E-  12 12 13 13 13  -2 -9 -9 -6 2  11 13 15 17 19  1 1 2 3 6  000E-05 585E-05 512E-05 981E-05 310E-05  1 1 1 1 1  046E+13 O40E+13 034E+13 029E+13 024E+13  5 3 2 2 2  02'6E 152E 197E 013E 098E  15 15 15 15 15  6 2 1 5 2  223E430E066E212E570E-  14 14 14 15 15  4 5 3 1 1  861E553E643E927E449E-  10 10 10 10 10  -6 -4 -3 -3 -3  922E293E560E587E204E-  21 23 25 27 29  1 1 2 3 6  000E-04 585E-04 512E-04 981E-04 310E-04  1 1 1 1 1  019E+13 015E+13 010E+13 006E+13 001E+13  2 2 3 3 5  375E 15 534E 15 367E 15 429E 15 1 1 1E15  1 1 2 3 4  525E564E352E018E071E-  15 15 15 15 15  1 8 3 1 9  041E162E397E166E377E-  10 -2 1 1 -2 1 1-3 10 -2 11 -2  996E982E669E612E950E-  31 33 35 37 39  1 1 2 3 6  000E-03 585E-03 512E-03 981E-03 310E-03  9 9 9 9 9  970E+12 927E+12 883E+12 837E+12 789E+12  4 8 5 4 2  510E 430E 591E 895E 563E  15 4 958E15 5 169E15 3 031E14 -2 082E14 -4 382E-  15 15 15 15 15  1 1 2 3 3  233E701E492E913E714E-  10 -2 330E- 12 1 10 -5 002E- 13 1 10 4 447E- 12 2 10 1 698E- 1 1 3 10 3 453E- 1 1 3  41 43 45 47 49  1 1 2 3 6  000E-02 585E-02 512E-02 98iE-02 310E-02  9 9 9 9 9  736E+12 678E+12 61 1E+12 528E+12 409E+12  2 6 2 1 1  837E 039E 354E 204E 347E  13 13 11 11 11  51 53 55 57 59  1 1 2 3 6  000E-01 585E-01 512E-0T 981E-01 310E-01  9 9 8 8 8  234E+12 074E+12 964E+12 894E+12 847E+12  2 9 2 1 1  861E 131E 061E 802E 055E  61 63 65 67 69 -  1 1 2 3 6  OOOE+00 585E+00 512E+0O 981E+00 310E+00  8 8 8 8 8  813E+12 787E+12 767E+12 742E+12 691E+12  1 8 5 2 1  71 73 75 77 79  1 1 2 3 6  000E+01 585E+01 512E+01 981E+01 043E+01  8 8 8 8 8  621E+12 573E+12 549E+12 538E+12 534E+12  81 83 85 87  7 9 9 1  805E+01 233E+01 824E+01 000E+02  8 8 8 8  532E+12 531E+12 531E+12 531E+12  11 11 11 11 1 1  1 1 11 1 1 1 1 1 1  14 9 242E- 10 8 511E- 1 1 14 6 208E- 09 4 355E- 10 1 262E- 09 13 1 557E- 08 12 -7 409E- 08 -4 902E- 09 14 -7 364E- 08 - 1 871E- 09  8 5 1 7 7  890E- 11 961E- 10 496E- 09 1 12E-09 056E- 09  9 9 9 9 9  990E -01 991E -01 992E -01 993E -01 995E -01  3 3 3 3 4  780E+00 395E+00 246E+00 503E+00 702E+00  0 0 0 0 0  OOOE+00 000E+00 OOOE+00 OOOE+00 OOOE+00  9 9 9 9 9  925E-01 924E-01 924E-01 923E-01 921E-01  41 43 45 47 49  11 2 088E12 8 784E12 8 762E13 -4 245E13 -3 787E-  1 11 1 194E- 07 1 12 -8 118E- 08 13 -8 123E- 08 - 1 14 1 302E- 08 -7 14 3 099E- 09 2  1 7 7 1 2  151E777E781E248E968E-  08 09 09 09 10  9 9 9 9 9  997E -01 998E -01 998E -01 997E -01 996E -01  4 3 2 1 1  519E+00 045E+00 393E+00 822E+00 365E+00  0 0 0 0 0  OOOE+00 000E+00 000E+00 OOOE+00 OOOE+00  9 9 9 1 1  920E-01 931E-01 976E-01 008E+00 013E+00  51 53 55 57 59  217E 094E 267E 433E 576E  13 14 14 14 14  -2 934E- 1 371E-6 778E- 1 304E-3 387E-  14 7 954E- 10 1 838E14 - 1 674E- 10 6 700E15 -4 410E- 1 1 6 989E15 4 552E- 10 -6 282E15 -4 098E- 09 - 1 523E-  1 1 12 12 12 10  7 1 4 4 3  621E605E277E360E928E-  11 1 1 12 1 1 10  9 9 9 9 9  993E -01 986E -01 977E -01 981E -01 986E -01  1 1 1 1 9  145E+00 077E+00 077E+00 005E+00 868E-01  1 2 3 2 1  179E- 10 057E-06 671E-05 999E-06 548E-09  1 1 1 1 1  014E+00 013E+00 01 1E+00 012E+00 013E+00  61 63 65 67 69  1 5 1 1 2  953E 949E 423E 976E 269E  15 16 15 15 15  -2 -5 -6 -1 -1  16 3 470E- 09 -2 170E16 .7 089E- 10 -2 371E16 1 397E- 1 1- 1 788E15 1 286E- 11 - 1 059E15 6 256E- 12 -3 098E-  1 1 1 1 1 1 11 12  3 6 2 1 6  323E793E173E596E687E-  10 11 12 12 13  9 9 9 9 7  979E -01 939E -01 785E -01 192E -01 151E -01  9 1 1 1 1  963E-01 001E+00 004E+00 004E+00 001E+00  1 7 1 2 3  638E-06 719E-03 188E+00 210E+01 214E+02  1 1 1 1 1  010E+00 006E+00 004E+00 002E+00 001E+00  71 73 75 77 79  2 2 2 1  402E 405E 137E 307E  15 15 15 15  - 1 855E-2 150E-2 218E-2 171E-  7 1 1 1  558E352E708E316E-  14 3 781E -01 14 - 1 859E -02 14 -2 124E -01 15 -2 729E -01  1 1 1 1  OOOE+00 OOOE+00 OOOE+00 OOOE+00  3 0 0 0  958E+03 OOOE+00 OOOE+00 000E+00  1 1 1 1  OOOE+00 OOOE+00 OOOE+00 OOOE+00  81 83 85 87  - 1 026E-9 369E-6 864E-2 455E-9 T73E-  400E282E185E090E559E-  15 6 15 8 15 1 15 -1  470E873E767E678E-  335E- 08 163E- 09 303E- 09 301E- 1 1 095E- 12  13 -4 511E- 13 14 - 1 069E- 13 13 8 578E- 15 15 - 1 029E- 16  Model 03310191  SS 3000./ 0 . 0 0 /  TauRM  Cp/Cv  -3.00  CN, TiO, H20 (hm) Oconv  H_P  Convective q u a n t i t i e s  DELrad  DELbub  DELad  TauRMb  Page 228 Hconv  Vconv  Hconv/H  1 3 5 7 9  1 1 2 3 6  000E-06 585E-06 512E-06 981E-06 310E-06  T .660E+00 1 .598E+00 1 .524E+00 1 .450E+00 1 .384E+00  1 1 8 6 4  633E+12 174E+12 506E+1 1 259E+1 1 725E+11  2 2 2 3 3  580E+00 771E+00 945E+00 097E+00 218E+00  1 1 1 2 2  534E-02 773E-02 976E-02 196E-02 431E-02  1 ' 1 1 1 9  515E -01 411E -01 278E -01 133E -01 929E -02  0 0 0 0 0  OOOE+00 OOOE+00 OOOE+00 OOOE+00 OOOE+00  2 2 2 3 3  086E-05 349E-05 671E-05 109E-05 750E-05  0 0 0 0 0  OOOE+OO OOOE+00 OOOE+00 OOOE+00 OOOE+OO  0 0 0 0 0  OOOE+00 OOOE+00 OOOE+00 OOOE+00 OOOE+00  0 0 0 0 0  OOOE+00 OOOE+OO OOOE+00 OOOE+00 OOOE+OO  1 3 5 7 ' 9  1 1 13 . 15 17 19  1 1 2 3 6  000E-05 585E-05 512E-05 981E-05 310E-05  1 . 332E + 00 1 . 291E+00 1 .262E+00 1 .241E+00 1 .227E+00  3 2 2 2 2  690E+11 998E+11 536E+1 1 227E+1 1 018E+1 1  3 3 3 3 3  305E+00 362E+00 392E+00 404E+00 400E+00  2 2 3 3 3  667E-02 885E-02 074E-02 238E-02 392E-02  8 7 6 6 6  705E -02 724E -02 984E -02 451E -02 081E -02  0 0 0 0 0  OOOE+00 OOOE+00 OOOE+00 OOOE+00 OOOE+00  4 6 8 1 1  733E-05 275E-05 724E-05 264E-04 894E-04  0 0 0 0 0  OOOE+00 OOOE+00 OOOE+00 OOOE+00 OOOE+00  0 0 0 0 0  OOOE+00 OOOE+00 OOOE+00 OOOE+00 OOOE+00  0 0 0 0 0  OOOE+00 OOOE+00 OOOE+00 OOOE+00 OOOE+OO  1 1 13 15 17 19  21 23 25 27 29  1 1 2 3 6  000E-04 585E-04 512E-04 981E-04 310E-04  1 .216E+00 1 209E+00 1 204E+00 1 199E+00 1 196E+00  1 1 1 1 1  875E+11 778E+1 1 71 1E+1 1 666E+11 635E+1 1  3 3 3 3 3  385E+00 359E+00 324E+00 281E+00 230E+00  3 3 3 4 4  538E-02 700E-02 839E-02 OOOE-02 150E-02  5 5 5 5 5  831E -02 672E -02 579E -02 535E -02 529E -02  0 0 0 0 0  OOOE+00 OOOE+OO OOOE+00 OOOE+00 OOOE+00  2 4 7 1 1  906E-04 531E-04 129E-04 124E-03 769E-03  0 0 0 0 0  OOOE+OO OOOE+00 OOOE+00 OOOE+OO OOOE+00  0 0 0 0 0  OOOE+00 OOOE+OO OOOE+00 OOOE+00 OOOE+00  0 0 0 0 0  OOOE+00 OOOE+OO OOOE+OO OOOE+00 OOOE+00  21 23 25 27 29  31 33 35 37 39  1 1 2 3 6  000E-03 585E-03 512E-03 981E-03 310E-03  1 1 1 1 1  194E+00 1 616E+1 1 192E+00 • 1 605E+11 190E+00 1 602E+11 188E+00 1 604E+1 1 186E+00 1 612E+11  3 3 3 2 2  171E+00 103E+00 025E+00 934E+00 826E+00  4 4 4 4 5  317E-02 498E-02 685E-02 908E-02 164E-02  5 5 5 5 5  555E -02 609E -02 692E -02 805E -02 959E -02  0 0 0 0 0  OOOE+00 OOOE+00 OOOE+00 OOOE+00 OOOE+00  2 4 6 1 1  771E-03 306E-03 620E-03 004E-02 497E-02  0 0 0 0 0  OOOE+00 OOOE+00 OOOE+00 OOOE+00 OOOE+OO  0 0 0 0 0  OOOE+00 OOOE+OO OOOE+00 OOOE+OO OOOE+00  0 0 0 0 0  OOOE+00 OOOE+00 OOOE+00 OOOE+00 OOOE+00  31 33 35 37 39  41 43 45 47 49  1 1 2 3 6  000E-02 585E-02 512E-02 981E-02 310E-02  1 1 1 1 1  184E+00 182E+00 180E+00 180E+00 194E+00  1 1 1 1 1  627E+11 649E+11 683E+11 738E+11 831E+11  2 2 2 2 1  695E+00 531E+00 315E+00 011E+00 593E+00  5 5 6 7 8  514E-02 958E-02 642E-02 906E-02 874E-02  6 6 6 7 1  169E -02 473E -02 958E -02 909E -02 044E -01  0 0 0 0 0  OOOE+00 OOOE+00 OOOE+00 OOOE+OO OOOE+00  2 3 4 5 5  185E-02 103E-02 233E-02 316E-02 707E-02  0 0 0 0 0  OOOE+00 OOOE+OO OOOE+00 OOOE+OO OOOE+00  0 0 0 0 0  OOOE+OO OOOE+00 OOOE+00 OOOE+00 OOOE+00  0 0 0 0 0  OOOE+00 OOOE+00 OOOE+OO OOOE+00 OOOE+00  41 43 45 47 49  51 53 55 57 59  1 1 2 3 6  000E-01 585E-01 512E-01 981E-01 310E-01  1 1 1 1 1  247E+00 306E+00 375E+00 453E+00 520E+00  1 2 2 2 2  935E+11 003E+11 070E+11 143E+1 1 214E+1 1  1 1 1 1 1  291E+00 181E+00 113E+00 067E+00 041E+00  8 8 1 1 2  114E-02 654E-02 227E-01 701E-01 332E-01  1 1 2 2 3  551E -01 999E -01 463E -01 934E -01 294E -01  0 0 0 0 0  OOOE+00 OOOE+00 OOOE+OO OOOE+00 OOOE+00  7 1 4 1 2  825E-02 749E-01 342E-01 128E+00 598E+00  0 0 0 0 0  OOOE+OO OOOE+00 OOOE+00 OOOE+00 OOOE+OO  0 0 0 0 0  OOOE+00 OOOE+00 OOOE+00 OOOE+00 OOOE+00-  0 0 0 0 0  OOOE+OO OOOE+00 OOOE+00 OOOE+OO OOOE+00  51 53 55 57 59  61 63 65 67 69  1 i 2' 3 6  000E+00 585E+00 512E+00 981E+00 310E+00  1 1 1 T 1  578E+00 624E+00 651E+00 662E+00 659E+00  2 2 2 2 2  297E+11 412E+1 1 575E+11 775E+11 990E+11  1 1 1 1 1  024E+00 014E+00 009E+00 004E+00 002E+00  3 6 9 6 4  818E-01 923E-01 378E-01 753E-01 223E-01  3 3 3 3 3  582E -01 800E -01 923E -01 967E -01 960E -01  3 6 9 6 4  818E-01 917E-01 338E-01 746E-01 223E-01  5 1 2 2 2  736E+00 31 1E+01 354E+01 341E+01 100E+01  2 7 1 8 6  748E+02 156E+03 886E+04 184E+03 397E+02  4 7 1 1 6  504E-02 905E+02 417E+04 165E+03 082E-01  1 2 3 2 1  179E- 10 057E-06 671E-05 999E-06 548E-09  61 63 65 67 69  71 73 75 77 79  1 1 2 3 6  000E+01 585E+01 512E+01 981E+01 043E+01  1 1 1 1 1  636E+00 561E+00 410E+00 259E+00 198E+00  3 3 3 4 5  237E+11 553E+11 951E+11 456E+11 096E+11  1 1 1 1 1  003E+00 012E+00 049E+00 177E+00 502E+00  5 1 2 8 2  354E-01 205E+00 979E+00 067E+00 484E+01  3 3 2 1 1  863E -01 540E -01 764E -01 748E -01 116E -01  5 1 1 5 2  351E-01 126E+00 256E+00 763E-01 367E-01  3 1 4 1 6  742E+01 236E+02 586E+02 768E+03 312E+03  6 1 5 1 2  223E+03 004E+05 038E+05 183E+06 592E+06  6 3 4 8 1  538E+02 116E+06 821E+08 994E+09 310E+11  1 7 1 2 3  638E-06 719E-03 188E+00 210E+01 214E+02  71 73 75 77 79  81 83 85 87  7 9 9 1  805E+01 233E+01 824E+01 000E+02  1 1 1 1  198E+00 212E+00 219E+00 221E+00  5 6 6 6  667E+11 1 882E+00 8 173E+11 2 218E+00 - 1 395E+1 1 2 355E+00 -2 462E+1 1 2 395E+00 - 1  821E+01 157E+03 040E+02 573E+02  9 8 8 8  097E -02 304E -02 101E -02 051E -02  1 0 0 0  673E-01 OOOE+00 OOOE+OO OOOE+00  1 2 2 2  382E+04 267E+04 699E+04 834E+04  5 0 0 0  789E+06 OOOE+OO OOOE+00 OOOE+00  1 0 0 0  613E+12 OOOE+00 OOOE+00 OOOE+00  3 0 0 0  958E+03 OOOE+00 OOOE+OO OOOE+00  81 83 85 87  Model 04310191  SS 3000./  TauRM  1.00/  -3.00  r/R*  Col Mass  Summary of p h y s i c a l  CN, T i O , H20 (hm) T  - 1  Pe  n  P_rad  P_gas  Page 229  quantities ChiRM  Rho  Mu  1 3 5 7 9  1 1 2 3 6  000E-05 585E-05 512E-05 981E-05 310E-05  4 5 7 1 1  467E-02 907E-02 809E-02 031E-01 360E-01  1 1 1 1 1  541E -02 501E -02 462E -02 422E -02 383E -02  2 2 2 2 2  025E+03 033E+03 042E+03 053E+03 066E+03  1 2 2 3 4  550E+12 041E+12 686E+12 530E+12 627E+12  9 1 1 2 2  108E-07 187E-06 551E-06 034E-06 673E-06  4 5 7 1 1  331E-01 728E-01 574E-01 001E+00 320E+00  1 1 1 1 1  042E-01 043E-01 044E-01 045E-01 046E-01  1 1 1 1 1  340E+00 355E+00 371E+0O 385E+00 397E+00  3 4 6 8 1  449E- 12 594E- 12 1 14E-12 1 16E-12 073E- 1 1  3 4 5 6 7  693E -04 422E -04 313E -04 409E -04 757E -04  11 13 15 17 19  1 1 2 3 6  OOOE-04 585E-04 512E-04 981E-04 310E-04  1 2 3 4 5  789E-01 351E-01 087E-01 056E-01 340E-01  1 1 1 1 1  345E -02 306E -02 268E -02 229E -02 190E -02  2 2 2 2 2  081E+03 098E+03 115E + 03 135E+03 155E+03  6 7 1 1 1  049E+12 890E+12 028E+13 339E+13 748E+13  3 4 6 8 1  520E-06 644E-06 139E-06 140E-06 084E-05  1 2 3 3 5  738E+00 285E+00 002E+00 947E+00 201E+00  1 1 1 1 1  047E-01 049E-01 051E-01 053E-01 057E-01  1 1 1 1 1  408E+00 416E+00 423E+0O 428E+00 432E+00  1 1 2 3 4  414E855E429E176E155E-  11 1 1 1 1 1 1 1 1  9 1 1 1 1  404E -04 ' 139E -03 375E -03 651E -03 966E -03  1 T 13 15 17 19  21 23 25 27 29  1 1 2 3 6  OOOE-03 585E-03 512E-03 981E-03 310E-03  7 9 1 1 2  059E-01 386E-01 258E+00 703E+00 337E+00  1 1 1 1 9  150E -02 109E -02 066E -02 022E -02 742E -03  2 2 2 2 2  178E+03 202E+03 229E+03 258E+03 290E+03  2 3 3 5 7  288E+13 011E+13 991E+13 338E+13 225E+13  1 1 2 3 5  453E-05 964E-05 685E-05 722E-05 251E-05  6 9 1 1 2  880E+00 154E+00 228E+01 664E+01 285E+01  1 1 1 1 1  061E-01 067E-01 074E-01 084E-O1 098E-01  1 1 1 1 1  434E+00 435E+00 435E+00 433E+00 429E+00  5 7 9 1 1  450E177E508E270E714E-  1 1 1 1 1 1 10 10  2 2 3 3 3  317E -03 696E -03 090E -03 482E -03 845E -03  21 23 25 27 29  31 33 35 37 39  1 1 2 3 6  OOOE-02 585E-02 512E-02 981E-02 310E-02  3 4 6 1 1  257E+00 630E+00 747E+00 016E+01 599E+01  9 8 8 7 6  235E -03 686E -03 082E -03 404E -03 618E -03  2 2 2 2 2  328E+03 371E+03 421E+03 483E+03 562E+03  9 1 1 2 4  918E+13 386E+14 980E+14 909E+14 447E+14  7 1 1 2 4  576E-05 124E-04 731E-04 799E-04 850E-04  3 4 6 9 1  187E+01 536E+01 618E+01 974E+01 573E+02  1 1 1 1 1  117E-01 143E-01 181E-01 236E-01 320E-O1  1 1 1 1 1  422E+00 412E+00 399E+00 381E+00 358E+00  2 3 4 6 1  342E251E599E670E002E-  10 10 10 10 09  4 4 4 4 3  144E -03 340E -03 382E -03 212E -03 747E -03  31 33 35 37 39  41 43 45 47 49  1 1 2 3 6  000E-01 585E-01 512E-01 981E-01 310E-01  2 5 9 1 2  704E+01 042E+01 988E+01 840E+02 823E+02  5 4 3 1 6  661E -03 446E -03 012E -03 634E -03 016E -04  2 2 2 3 3  668E+03 813E+03 988E+03 180E+03 381E+03  7 1 2 4 6  233E+14 282E+15 397E+15 160E+15 013E+15  9 2 5 1 2  343E-04 071E-03 040E-03 212E-02 720E-02  2 4 9 1 2  664E+02 979E+02 888E+02 827E+03 807E+03  1 1 1 2 3  450E-01 664E-01 991E-01 460E-01 115E-01  1 1 1 1 1  330E+00 301E+00 279E+00 264E+00 253E+00  1 2 5 8 1  597E- 09 770E- 09 091E- 09 734E- 09 251E- 08  2 2 1 1 3  983E -03 156E -03 729E -03 871E -03 143E -03  41 43 45 47 49  51 53 55 57 59  1 1 2 3 4  OOOE+00 468E+00 154E+00 162E+00 642E+00  3 4 4 5 5  662E+02 249E+02 737E+02 145E+02 574E+02  -6 730E -05 -4 698E -04 -7 776E -04 -1 025E -03 - 1 282E -03  3 3 3 4 4  564E+03 721E+03 909E+03 145E+03 437E+03  7 8 8 8 9  408E+15 239E+15 747E+15 963E+15 074E+15  5 8 1 2 3  315E-02 875E-02 494E-01 505E-01 957E-01  3 4 4 5 5  645E+03 233E+03 721E+03 129E+03 559E+03  3 4 5 7 9  917E-01 740E-01 846E-01 434E-01 757E-01  1 1 1 1 1  248E+00 244E+00 242E+00 241E+00 240E+00  1 1 1 1 1  535E703E804E847E868E-  08 08 08 08 08  6 1 1 3 3  048E -03 046E -02 883E -02 085E -02 576E -02  51 53 55 57 59  61 63 65 67 69  6 1 1 2 3  813E+00 O00E+O1 445E+01 089E+01 020E+01  6 7 8 9 9  223E+02 287E+02 418E+02 214E+02 679E+02  -1 -2 -2 -3 -3  661E -03 247E -03 825E -03 217E -03 450E -03  4 5 5 6 6  777E+03 170E+03 605E+03 097E+03 648E+03  9 1 1 1 1  417E+15 020E+16 087E+16 095E+16 055E+16  5 9 2 7 2  763E-01 770E-01 443E+00 558E+00 385E+01  6 7 8 9 9  210E+03 277E+03 414E+03 214E+03 681E+03  1 1 2 3 4  312E+O0 804E+00 496E+00 498E+00 948E+00  1 1 1 1 1  239E+00 239E+00 239E+00 238E+00 236E+00  1 2 2 2 2  938E- 08 098E- 08 236E- 08 250E- 08 165E- 08  3 3 5 1 3  096E -02 122E -02 357E -02 263E -01 245E -01  61 63 65 67 69  71 73 75 77 79  4 6 8 9 9  365E+01 310E+01 610E+01 750E+01 950E+01  9 1 1 1 1  937E+02 008E+03 014E+03 016E+03 016E+03  -3 -3 -3 -3 -3  586E -03 665E -03 705E -03 716E -03 718E -03  7 7 8 8 8  261E+03 938E+03 565E+03 830E+03 874E+03  9 9 8 8 8  914E+15 192E+15 571E+15 327E+15 287E+15  7 1 4 5 6  158E+01 993E+02 413E+02 940E+02 228E+02  9 1 1 1 1  938E+03 007E+04 014E+04 015E+04 015E+04  7 1 1 1 1  046E+00 008E+01 367E+01 544E+01 575E+01  1 1 1 1 1  230E+00 214E+00 185E+00 167E+O0 163E+00  2 1 1 1 1  025E854E687E613E600E-  8 2 5 7 8  602E -01 353E+00 617E+00 954E+00 416E+00  71 73 75 77 79  08 08 08 08 08  1 3 5 7 9  Model 04310191  SS 3000./ r  TauRM.  1.00/  -3.00  RE res  TE r  HE res  DE res  EPS rms  g_eff/g  Chi_H/RM  Hconv/H  Lum/L *  0 .OOOE+00 9 .529E- 13 1 .869E- 12 2 .387E- 12 2 .686E- 12  1 .267E- 08 1 .039E- 06 2 . 231E- 07 1 .026E- 10 3 .321E- 11  9 .997E 9 .997E 9 .997E 9 .997E 9 .997E  -01 -01 -01 -01 -01  5 . 288E + 01 4 .320E+01 3 .520E+01 2 .862E+01 2 . 323E + 01  0 .OOOE+00 0 .OOOE+OO 0 .OOOE+00 0 .OOOE+00 0 .OOOE+OO  9 .868E-01 9 .868E-01 9 .868E-01 9 .868E-01 9 .868E-01  1 3 5 7 9  10 10 10 09 09  2 .816E2 .932E3 .146E3 .430E3 .499E-  12 12 12 12 12  1 . 505E- 1 1 5 . 164E-11 7 .664E- 11 9 .829E- 11 1 . 230E- 10  9 .997E 9 .997E 9 .997E 9 .997E 9 .997E  -01 -01 -01 -01 -01  1 .883E+01 1 . 528E + 01 1 . 243E + 01 1 .015E+01 8 .340E+00  0..OOOE+00 0 .OOOE+00 0 .OOOE+00 0 .OOOE+OO 0 .OOOE+00  9 .868E-01 9 .868E-01 9 .868E-01 9 .868E-01 9 .868E-01  1 1 13 15 17 19  09 09 09 09 09  4 .555E4 .401E4 .657E 4 .996E6. 413E-  12 12 12 12 12  1 . 209E1 .638E1 .753E1 .873E1 .611E-  10 10 10 10 10  9 .998E 9 .998E 9 .998E 9 .998E 9 .998E  -01 -01 -01 -01 -01  6 .905E+00 5 .771E+00 4 .877E+00 4 . 173E + 00 3 . 623E+00  0 .OOOE+OO 0 .OOOE+00 0 .OOOE+OO 0 .OOOE+00 0. OOOE+00  9 .868E-01 9 .868E-0-1 9 .867E-01 9 .867E-01 9 .867E-01  21 23 25 27 29  . 12 - 1 896E09 , 09 12 - 1.832E09 12 - 1.736E, 12 2,, 742E- 09 . 09 12 1.914E-  7 .493E8..228E9..800E3. 033E4 ,347E.  12 12 12 11 1 1  1.816E. 1 .755E1.663E. 2 .627E1 .834E-  10 10 10 10 10  9 .998E -01 9 .998E -01 9 998E -01 9 .998E -01 9 .999E -01  3 . 194E + 00 2 . 862E + 00 2 .614E+00 2 .442E+00 2 . 359E + 00  0. OOOE+00 0, OOOE+00 0..OOOE+00 0. OOOE+00 9 •668E-15 .  9 .867E-01 9 .867E-01 9 .867E-01 9 .866E-01 9 .866E-01  31 33 35 37 39  1 .372E- 09 4 .080E- 09 9 .436E- 09 2 . 329E- 08 1.071E. 08  9 .999E 9 .999E 9 .999E 9..999E 9..999E  -01 -01 -01 -01 -01  2 . 390E + 00 2 .451E+00 2 .235E+00 2 .010E+00 1 .634E+O0  5..081E-16 0..OOOE+OO 0. OOOE+00 0. OOOE+OO 0. OOOE+00  9 .865E-01 9 .863E-01 9 .865E-01 9 .889E-01 1 .001E+00  41 43 45 47 49  10 11 11 12 12  9..445E- 09 7., 704E- 10 9..060E- 10 4.. 362E- 1 1 1 996E- 10  9. 999E -01 9 .998E -01 9..997E -01 9 .995E -01 9..994E -01  1 . 223E + 00 1 .068E+00 1 .015E+00 1 .015E+00 1.004E+00 .  0. OOOE+00 9. 943E-07 4 . 275E-04 7 ,, 379E-03 6 . 437E-03  1 .010E+00 1.011E+00 . 1.010E+00 , 1 .008E+00 1 .007E+00  51 53 55 57 59  3. 038E- 09 3.. 333E-1 . 234E- 09 -4 . 399E. - 3 . 511E- 09 - 1 616E-4. 569E- 10 8. 159E-1 . 671E- 1 1 9 . 322E-  12 13 12 13 13  2. 910E- 10 T . 182E-10 3.. 363E- 10 4. 376E- 11 1 603E. 12  9 .995E -01 9. 995E -01 9 . 992E -01 9. 981E -01 9. 950E -01  9 .896E-01 9,.955E-01 9 996E-01 1 001E+00 . 1 002E+00 .  4 . 165E-04 1 217E-04 . 1 822E-02 . 7 . 210E-01 5. 842E+00  1 .008E+00 1 .007E+00 1 005E+00 . 1 004E+00 . 1 003E+00 ,  61 63 65 67 69  - 11 . 13E-12 6 .762E-1 . 117E- 1 1 4. 499E-2. 993E- 12 8. 368E9. 862E- 12 5. 988E- 7 . 580E- 13 - 3 . 536E-  13 13 14 14 15  1 248E. 1 071E. 2. 867E9. 446E7. 287E-  9. 868E -01 9 . 639E -01 9 . 139E-01 8.. 782E -01 8.. 7 11E-01  1 002E+00 . 1 002E+00 . 1 OOOE+00 . 1 OOOE+00 . 9. 999E-01  3. 1 8. 1 1  1•002E+00 . 1 001E+00 . 1 OOOE+00 . 1 OOOE+OO . 1 OOOE+00 .  71 73 75 77 79  1 3 5 7 9  1 .000E-05 1 .585E-05 2 .512E-05 3 .981E-05 6 .310E-05  9 . 149E+12 9 . 146E+12 9 . 142E+12 9 .139E+12 9 . 135E+12  1 . 190E 1 .054E 2 . 262E 1 .997E 1 .849E  08 06 07 13 13  -7 .202E- 1 .137E-4 .106E- 1.910E-6 .508E-  13 4 .979E- 08 10 2 . 191E-09 13 1 .876E- 09 13 1 .071E- 09 14 .3 .467E- 10  11 13 15 17 19  1 .OOOE-04 1 .585E-04 2 .512E-04 3 .981E-04 6 .310E-04  9 . 132E+12 9 . 128E+12 9 .125E+12 9 .121E+12 9 .118E+12  2 . 127E 1 . 316E 4 . 322E 8 .657E 3 .021E  13 13 13 14 13  8 .557E- 1 .349E-8 .065E- 1.803E-3 .460E-  16 14 14 13 13  -1 .571E-5 .391E-8 .001E- 1.026E-1 .284E-  21 23 25 27 29  1 .000E-03 1 .585E-03 2 .512E-03 3 .981E-03 6 .310E-03  9 .114E+12 9 .110E+12 9 .107E+12 9 .103E+12 9 .098E+12  6 . 702E 1 .860E 1 .487E 2 .201E 3..625E  14 13 12 13 13  -5 -8 -1 -1 -1  .919E.480E.144E.538E.962E-  13 13 12 12 12  -1 .262E-1 .710E-1 .830E- 1 .956E- 1.682E.  31 33 35 37 39  1.000E-02 . 1 .585E-02 2 .512E-02 3 .981E-02 6 .310E-02  9..094E+12 9..089E+12 9 .083E+12 9 .077E+12 9 .070E+12  1.687E . 5..943E 8.. 780E 6 . 132E 3 .075E  12 13 13 12 12  -2 -2 -3 -4 -5  . 386E.909E.417E.014E.544E-  41 43 45 47 49  1 .000E-01 1 .585E-01 2 .512E-01 3..981E-01 6..310E-01  9 .062E+12 9 .051E+12 9 .038E+12 9 .025E+12 . 9..016E+12  1 .005E 5 .744E 6 .597E 1 082E . 5,.476E  1 1- 1 .070E11 -2 .754E11 -4 .922E10 -3. .277E11 -3 .501E-  10 11 1.432E. 08 1,304E, 11 -4 .259E- 08 -2. 679E- 10 11 -9, 851E- 08 -8. , 136E-10 11 2. 431E- 07 -3. 276E- 10 . 07 7 .028E- 10 13 1.118E-  51 53 55 57 59  1.OOOE+00 . 1.468E+00 . 2..154E+00 3..162E+00 4..642E+00  9..010E+12 9. 006E+12 9 . 004E+12 9..001E+12 8. 999E+12  1 491E . 3.. 896E 6.. 470E 2,, 746E 3. 003E  11 5..210E12 1.281E. 13 - 1.066E. 13 1.662E. 13 3..219E-  12 -9 . 861E- 08 -1 . 532E12 8. 043E- 09 -2. 587E14 9. 459E- 09 1 010E. 13 -4. 553E- 10 -4. 455E13 2. 084E- 09 6 .734E-  61 63 65 67 69  6..813E+00 1.000E+01 . 1 445E+01 . 2. 089E+01 3. 020E+01  8. 8. 8. 8. 8.  996E+12 990E+12 985E+12 982E+12 979E+12  4,. 301E 3,. 243E 1 884E . 9. 631E 4. 922E  14 14 14 15 15  14 14 14 14 15  71 73 75 77 79  4..365E+01 6.,310E+01 8. 610E+01 9..750E+01 9. 950E+01  8. 8. 8. 8. 8.  978E+12 978E+12 977E+12 977E+12 977E+12  2. 462E 1 048E . 5. 159E 1 044E . 6. 119E  15 2. 095E- 15 15 6. 331E- 16 16 0. 000E+00 15 - 1 381E. 15 15 1 240E. 14  3..301E4..313E2.,329E1.151E. 5.. 380E-  Page 230  Convergence checks  CN, T i O , H20 (hm)  i  13 12 13 13 14  150E+01 774E+02 . 294E+02 . 579E+03 760E+03 .  to CO o  Model 04310191  SS 3000./  1.00/  -3.00  CN, T i O , H20 (hm)  TauRM  Cp/Cv  1 3 5 7 9  1 .000E-05 1 .585E-05 2 .512E-05 3 .981E-05 6 .310E-05  1.205E+00 1.201E+00 1.199E+00 1.198E+00 1.198E+00  1 .606E+10 1.518E+10 1.451E+10 1.401E+10 1.364E+10  2.680E+00 2.852E+00 3.005E+00 3.131E+00 3.228E+00  1 1 13 15 17 19  1 .000E-04 1 . 585E-04 2 . 512E-04 3 .981E-04 6 .310E-04  1.199E+00 1.200E+00 1.200E+00 1.201E+00 1.202E+00  1.339E+10 1.322E+10 1.312E+10 1.308E+10 1.308E+10  21 23 25 27 29  1 .000E-03 1 . 585E-03 2 .512E-03 3 .981E-03 6 . 310E-03  1.203E+00 1.204E+00 1.204E+00 1.205E+00 1.206E+00  31 33 35 37 39  1 .OOOE-02 1 . 585E-02 2 .512E-02 3..981E-02 6 . 310E-02  41 43 45 47 49  Page 231  Vconv  Hconv  3 . 274E-05 4 .935E-05 7 .542E-05 1 . 166E-04 1 .817E-04  0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO  0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 0 OOOE+OO  0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO  1 3 5 7 9  5.452E-02 5.413E-02 5.400E-02 5.406E-02 5.429E-02  0 .OOOE+OO 2 .848E-04 0 .OOOE+OO 4 .470E-04 0 .OOOE+OO 7 .013E-04 0 .OOOE+OO 1 .097E-03 0 .OOOE+OO ' 1 . 709E-03  0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 0 OOOE+OO 0 OOOE+OO  0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 0..OOOE+OO 0 OOOE+OO  0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 0 OOOE+OO 0. OOOE+OO  1 1 13 15 17 19  3 .779E -02 3 .974E -02 4 . 181E-02 4 . 400E -02 4 .673E -02  5.467E-02 5.519E-02 5.586E-02 5.670E-02 5.774E-02  0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO  2 .649E-03 4 .083E-O3 6 . 254E-03 9 .522E-03 1 . 440E-02  0. OOOE+OO 0 OOOE+OO 0 .OOOE+OO 0..OOOE+OO 0 .OOOE+OO  0..OOOE+OO 0..OOOE+OO 0 .OOOE+OO 0. OOOE+OO 0. OOOE+OO  0. OOOE+OO 0..OOOE+OO 0. OOOE+OO 0, OOOE+OO 0. OOOE+OO  21 23 25 27 29  3.160E+00 3.061E+00 2.928E+00 2.748E+00 2.504E+00  4 .996E -02 5 . 381E-02 5 .871E -02 6.. 484E -02 7 .233E . -02  5.906E-02 6.077E-02 6.308E-02 6.638E-02 7.151E-02  0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 0..OOOE+OO 7.. 233E-02  2 . 160E-02 3 . 213E-02 4 . 724E-02 6..830E-02 9. . 563E-02  0 OOOE+OO 0..OOOE+OO 0. OOOE+OO 0. OOOE+OO 1. 077E+01  0. 0. 0. 0. 3.  OOOE+OO OOOE+OO OOOE+OO OOOE+OO 487E-06  0. 0. 0. 0. 9.  OOOE+OO OOOE+OO OOOE+OO OOOE+OO 668E-15  31 33 35 37 39  1.688E+10 1.814E+10 1.954E+10 2.098E+10 2.246E+10  2.185E+00 1.824E+00 1.532E+00 1.322E+00 1.172E+00  8. 078E -02 8, 664E -02 9. . 229E -02 -01 1. 203E . 1. 747E . -01  8.041E-02 9.687E-02 1.230E-01 1.616E-01 2.185E-01  8. 078E-02 0..OOOE+OO 0. OOOE+OO 0. OOOE+OO 0. OOOE+OO  1. .287E-01 1.733E-01 . 2,, 753E-01 5. 487E-01 1.413E+00 .  3. 595E+00 0. OOOE+OO 0. OOOE+OO 0. OOOE+OO 0. OOOE+OO  1. 836E-07 0. OOOE+OO 0. OOOE+OO 0. OOOE+OO 0. OOOE+OO  5. 0. 0. 0. 0.  081E-16 OOOE+OO OOOE+OO OOOE+OO OOOE+OO  41 43 45 47 49  1.426E+00 1.500E+00 1.570E+00 1.622E+00 1.649E+00  2.375E+10 2.484E+10 2.612E+10 2.772E+10 2.969E+10  1.094E+00 1.056E+00 1.031E+00 1.017E+00 1.010E+00  2.,442E -01 3.,643E -01 5. 809E -01 8., 178E-01 7 . 887E -01  2.750E-01 3.174E-01 3.537E-01 3.786E-01 3.913E-01  0. OOOE+OO 3..641E-01 5.. 737E-01 7.. 706E-01 7. 477E-01  3.. 527E + 00 0. OOOE+OO 7 .076E+00 3. 242E+03 1 421E+01 . 2. 489E+04 2..527E+01 6 .529E+04 3,. 173E + 01 6 .284E+04  0. OOOE+OO 3. 637E+02 1 565E+05 . 2. 702E+06 2. 359E+06  0. OOOE+OO 9. 943E-07 4. 275E-04 7 . 379E-03 6. 437E-03  51 53 55 57 59  6 . 813E+00 1 OOOE+01 . 1 445E+01 . 2. 089E+01 3. 020E+01  1.658E+00 1.654E+00 1.631E+00 1.575E+00 1.473E+00  3.195E+10 3.455E+10 3.742E+10 4.070E+10 4.444E+10  1.004E+00 1.002E+00 1.004E+00 1.011E+00 1.031E+00  5. 694E -01 4. 957E -01 7. 192E -01 1. .330E+00 2. 551E+00  3.958E-01 3.943E-01 3.850E-01 3.611E-01 3. 115E-01  5. 633E-01 4. 934E-01 6. 618E-01 7. 490E-01 5. 957E-01  3. 067E+01 3. 620E+01 7 ,. 173E+01 1 851E+02 . 4 . 994E+02  2. 1 8. 2. 5.  502E+04 616E+04 . 333E+04 775E+05 374E+05  1 527E+05 . 4. 467E+04 6 .695E+06 2. 652E+08 2. 150E+09  4. 165E-04 1 217E-04 . 1 822E-02 . 7 . 210E-01 5. 842E+00  61 63 65 67 69  4.. 365E + 01 6 . 310E + 01 8. 610E+01 9 . 750E+01 9 .950E+01  1.347E+00 1.250E+00 1.210E+00 1.205E+00 1.205E+00  4.876E+10 5.400E+10 5.972E+10 6.256E+10 6.306E+10  1.086E+00 1.220E+00 1.452E+00 1.590E+00 1.616E+00  4. 929E+00 9. 841E+0O 1 791E+01 . 2. 299E+01 2. 400E+01  2. 379E-01 1 .653E-01 1 .220E-01 1 . 100E-01 1 .083E-01  3. 865E-01 2. 262E-01 1 471E-01 . 1 274E-01 . 1 247E-01 .  1 359E+03 . 3. 769E+03 9 .053E+03 1 ..284E+04 1 359E+04 .  8. 806E+05 1 429E+06 . 2. 228E+06 2. 708E+06 2. 801E+06  1 159E+10 . 3. 150E+01 6 . 531E+10 1 774E+02 . 3. 054E+11 8. 294E+02 5. 813E+1 1 1 .579E+03 6 .481E+1 1 1 760E+03 .  71 73 75 77 79  DELad  DELbub  1 . 316E -02 1 .553E -02 1 .808E -02 2 . 113E-02 2 .433E -02  6 . 370E-02 6.054E-02 5.814E-02 5.641E-02 5.525E-02  0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO  3.298E+00 3.344E+00 3.372E+00 3.386E+00 3.388E+00  2 . 731E-02 2 .992E -02 3 . 212E -02 3 .410E -02 3 .597E -02  1.312E+10 1.319E+10 1.331E+10 1.346E+10 1.365E+10  3.380E+00 3.361E+00 3.332E+00 3.291E+00 3.235E+00  1.206E+00 1.206E+00 1.205E+00 1.203E+00 1.201E+00  1.391E+10 1.423E+10 1.465E+10 1.519E+10 1;591E+10  1 .000E-01 1.585E-01 . 2,, 512E-01 3. 981E-01 6 . 310E-01  1.199E+00 1.204E+00 1.224E+00 1.265E+00 1.340E+00  51 53 55 57 59  1 OOOE+OO . 1. 468E . + 00 2..154E+00 3.. 162E + 00 4. 642E+00  61 63 65 67 69 71 73 75 77 79  H_P  Qconv  DELrad  Convective q u a n t i t i e s TauRMb  Hconv/H  to  Model 05310191 TauRM  SS 3000./ 2 . 0 0 / Col Mass  -3.00  r/R*  Summary of p h y s i c a l  CN, T i O , H20 (hm)  - 1  T  n  Pe  P_gas  P_rad  Page 232  quantities Mu  ChiRM  Rho  1 3 5 7 9  1 .OOOE-05 1 .585E-05 2 .512E-05 3 .981E-05 6 .3 TOE-05  1.711E-02 , 2 .292E-02 3 .049E-02 4..040E-02 5 .343E-02  1 . 209E -03 1 . 171E-03 1 . 135E-03 1 .099E -03 1 .063E -03  2 .056E+03 2 .065E+03 2 .076E+03 2 .087E+03 2 .099E+03  6..015E+12 8 .018E+12 1 .061E+13 1.399E+13 . 1.840E+13 .  3..281E-06 4..335E-06 5. 698E-06 7.. 473E-06 9.. 793E-06  1 .707E+00 2 . 286E+00 3 .041E+00 4 .031E+00 5 . 331E + 00  1 .067E-01 1 .067E-01 1 .068E-01 1 .068E-01 1 .069E-01  1 .448E+00 1 .473E+00 1 .497E+00 1 .520E+00 1 .541E+00  1.447E. 1.961E. 2 .638E3 .530E4..707E-  1 1 1 1 1  1 1 1 1 1  9 .086E -04 1 . 105E-03 1 . 340E -03 1 .619E -03 1 .949E -03  1 3 5 7 9  1 1 13 15 17 19  1 .000E-04 1 .585E-04 2 .512E-04 3 .981E-04 6 .310E-04  7 .061E-02 9 . 335E-02 1. .236E-01 1 .640E-01 2 .185E-01  1 .028E -03 9 .934E -04 9 .585E -04 9 .234E -04 8 .878E -04  2. 2. 2. 2.. 2..  2 .416E+13 3 .173E+13 4.. 170E+13 5..491E+13 7..252E+13  1. ,284E-05 1 689E-05 . 2..228E-05 2. 953E-05 3. 937E-05  7 .045E+00 9 . 314E + 00 1 .233E+01 1 .637E+01 2 .180E+01  1 .070E-01 1 .071E-01 1 .072E-01 1 .074E-01 1 .076E-01  1 .561E+00 1 . 579E+00 1 . 596E+00 1 .610E+00 1 .623E+00  6,.262E8 .319E.1 .105E1 468E. 1 .955E-  1 1 1 1 10 10 10  2 . 339E -03 2 . 795E -03 3 . 323E -03 3 .926E -03 4 .604E -03  1 1 13 15 17 19  21 23 25 27 29  1 .000E-03 1 .585E-03 2 . 512E-03 3 .981E-03 6 .310E-03  2,.924E-01 3..936E-01 5..333E-01 7 .283E-01 . 1.003E+00 .  8 .517E -04 8 . 146E-04 7 . 766E -04 7 . 373E -04 6 .965E -04  2.. 198E+03 2 . 220E+03 2 .244E+03 2 . 270E+03 2 . 300E+03  9..617E+13 1.282E+14 . 1.719E+14 . 2,.320E+14 3..152E+14  5,, 286E-05 7 .. 159E-05 9,. 793E-05 1.356E-04 . 1 903E-04 .  2 .918E+01 3 .928E+01 5 . 324E + 01 7 .27OE+01 1 .001E+02  1 .080E-01 1 .084E-01 1 .091E-01 1 .099E-01 1 . 111E-01  1 .635E+00 1 .644E+00 1 .651E+00 1 .656E+00 1 . 656E+00  2..610E3 .499E4..712E6 .377E. 8. 671E-  10 10 10 10 10  5.. 355E -03 6 . 174E-03 7 .051E -03 -03 7 .977E . 8 .935E -03  21 23 25 27 29  31 33 35 37 39  1 .OOOE-02 1 . 585E-02 2 .512E-02 3 .981E-02 6 .310E-02  1. .393E+00 1.955E+00 . 2.. 773E+00 3..988E+00 5.. 843E+00  6 .539E -04 6 .092E -04 5..618E -04 5 . 106E-04 4 . 542E -04  2,. 334E+03 2.. 374E+03 2..422E+03 2 .479E+03 2.. 549E+03  4.. 316E+14 5..955E+14 8.. 284E+14 1, .164E+15 1.658E+15 ,  2., 714E-04 3, 945E-04 5, 871E-04 8. 989E-04 •1 428E-03 .  1 . 391E+02 1 .952E+02 2 .769E+02 3 .983E+02 5 .836E+02  1 . 128E-01 1 .152E-01 1 .186E-01 1 . 236E-01 1 . 310E-01  1 .652E+00 1 .642E+00 1 .622E+00 1 .593E+00 1 .552E+00  09 1,184E, 09 1.623E. 2..232E- 09 3. 078E- 09 4.. 273E- 09  9 .907E -03 1 .086E -02 1. .172E-02 1. .238E -02 1 . 259E -02  31 33 35 37 - 39  41 43 45 47 49  1.OO0E-O1 . 1.585E-01 . 2..512E-01 3..981E-01 6..310E-01  8.. 822E+00 3..896E -04 1.390E+01 . 3.. 130E-04 2.,321E+01 2.. 191E-04 4,.143E+01 1.028E . -04 7 .523E+01 . -2, .906E -05  2..636E+03 2.. 743E + 03 2. 878E+03 3..052E+03 3,, 269E + 03  2,.422E+15 3. 666E+15 5. 836E+15 9,.827E+15 1 666E+16 .  2..390E-03 4. 282E-03 8.. 372E-03 1 816E-02 . 4..329E-02  8 .812E+02 1 . 388E+03 2.. 319E+03 4 . 140E + 03 7.. 520E+03  1 .418E-01 1 .577E-01 1 .818E-01 2 .193E-01 2 . 792E-01  1 .502E+00 1 .446E+00 1 . 390E + 00 1 .339E+00 1 .298E+00  6 .038E. 8..804E1.347E. 2..186E3..591E-  09 09 08 08 08  1. .212E-02 1 092E . -02 9,. 108E-03 7 .. 322E -03 6 .874E -03  41 43 45 47 49  51 53 55 57 59  1.OOOE+00 , 1.468E+00 , 2..154E+00 3 . 162E+00 4,.642E+00  1.203E+02 , 1.545E+02 . 1 836E+02 . 2. 075E+02 2..283E+02  -1 . 433E -04 -2, .092E -04 -2. .575E -04 -2. .938E -04 -3, . 240E -04  3,.517E+03 3 707E+03 3. 904E+03 4.. 136E + 03 4, 419E+03  2. 477E+16 3. 019E+16 3. 406E+16 3. 635E+16 3.. 743E+16  1 048E-01 . 1 974E-01 , 3. 569E-01 6 .421E-01 1.127E+00 .  1 203E+04 . 1 .545E+04 1.836E+04 . 2 .075E+04 2 . 283E + 04  3 .699E-01 4 .622E-01 5 .767E-01 7 .335E-01 9 .606E-01  1 .269E+00 1 .257E+00 1 .249E+00 1 .245E+00 1 .242E+00  5..220E- 08 08 6 .302E. 7 .067E- 08 7 .512E. 08 7.. 717E- 08  1 045E . -02 1. 778E . -02 3.. 128E-02. 5.. 586E -02 8.. 553E -02  51 53 55 57 59  61 63 65 67 69  6,.813E+00 1.OOOE+01 . 1.445E+01 . 2..089E+01 3..020E+01  2..517E+02 2. 863E+02 3..306E+02 3. 689E+02 3. 937E+02  -3, , 574E ^04 4., 760E + 03 -4. 050E -04 ' 5,. 156E + 03 -4. 624E -04 5,. 592E+03 -5. 098E -04 6. 084E+03 -5. . 405E -04 6 . 635E+03  3. 831E+16 4. 023E+16 4.. 283E+16 4.,393E+16 4..299E+16  1 834E+00 . 2. 873E+00 5. 690E+00 1 570E+01 . 4 . 803E+01  2 . 518E+04 2..864E+04 3 . 307E+04 3..690E+04 3 .938E+04  1 .294E+00 1 .785E+O0 2 .473E+00 3 .469E+00 4 .909E+00  1 .240E+00 1 .239E+00 1 . 239E + 00 1 . 239E + 00 1 .238E+00  7. 890E- 08 8..280E- 08 8. 812E- 08 9. 035E- 08 8. 834E- 08  9 . 481E -02 9 .078E . -02 1. 205E . -01 2. 457E -01 5. 862E -01  61 63 65 67 69  71 73 75 77 79  4.. 365E + 01 6 ,310E+01 , 8. 610E+01 9 , 750E+01 9 .950E+01 .  4. 084E+02 4.,169E+02 4. 211E+02 4..223E+02 4..225E+02  -5. . 594E -04 -5. , 711E-04 -5. . 774E -04 -5. . 793E -04 -5. . 795E -04  4. 083E+16 3..811E+16 3. 568E+16 3. 471E+16 3. 455E+16  1 437E+02 . 4. 034E+02 9. 073E+02 1 233E+03 . 1 296E+03 .  4..085E+04 4.. 170E + 04 4..212E+04 4.. 224E + 04 4., 226E+04  6 .992E+00 1 .OOOE+01 1 . 357E+01 1 .534E+01 1 .564E+01  1 . 235E+00 1 .227E+00 1 .212E+00 1 .203E+00 1 .201E+00  8. 371E- 08 7 . 766E- 08 7 . 183E- 08 6 .933E- 08 6. 891E- 08  1. 452E . + 00 3. 658E+00 8. 077E+00 1 ..111E + 01 1 ..170E + 01  71 73 75 77 79  1 12E+03 127E+03 142E+03 159E+03 178E+03  7 ,. 247E+03 7 .924E+03 8., 550E+03 8,.8T5E+03 8. 859E+03  Model 05310191 TauRM  SS 3000./ 2.00/ r  -3.00  TE res  RE res  HE res  EPS rms  g_eff/g  0. OOOE+OO 3. 658E- 13 4,, 382E- 13 4. 256E- 13 3., 700E- 13  3. 062E- 07 2. 436E- 07 1. ,163E-10 8. 673E- 11 1.584E. 11  1.OOOE+00 1.OOOE+00 1.OOOE+00 1.OOOE+00 1.OOOE+OO  2. 807E- 13 2..255E- 13 2..661E- 13 3. 634E- 13 4,. 179E-13  1 542E. 2.. 260E1 729E, 4. 655E1 229E,  DE res  1 3 5 7 9  1 000E-05 . 1 585E-05 . 2..512E-05 3..981E-05 6.. 310E-05  9 .021E+12 9 .021E+12 9 .021E+12 9 .020E+12 9 .020E+12  3.. 105E-07 2. 470E- 07 2.. 140E-13 2. 703E- 13 1 434E. 13  . 08 -4. , 102E-12 1.150E- 3 . 087E- 12 2. 068E- 10 , 09 -2. , 189E-12 1.214E-1 .. 304E- 12 9. 055E- 10 . 10 -5. . 165E-13 - 1.654E-  1 1 13 15 17 19  1.000E-04 . 1.585E-04 . 2..512E-04 3..981E-04 6..310E-04  9 .020E+12 9 .019E+12 9 .019E+12 9 .019E+12 9 .019E+12  3.. 691E- 13 2. 967E- 13 3. 631E- 13 1. .133E-13 1 893E. 13  - 1 281E. - 1 512E. - 3 . 562E-5. 759E-9 . 316E-  13 13 13 13 13  - 1,609E, -2. ,360E- 1.805E. -4. .860E1.283E,  21 23 25 27 29  1.000E-03 . 1 .585E-03 2..512E-03 3 .981E-03 6 .310E-03  9 .018E+12 9 .018E+12 9 .018E+12 9 .017E+12 9 .017E+12  1. .339E1.720E. 2.. 715E2, 428E2.. 788E-  13 13 13 13 13  -1 ..387E- 1 875E. -2, , 330E-2. . 774E-3. 040E-  12 12 12 12 12  -3. .297E- 10 -8. .778E- 10 - 1.198E, 09 - 1.451E. 09 - 1.727E. 09  31 33 35 37 39  1.000E-02 . 1 .585E-02 2..512E-02 3 .981E-02 6 .310E-02  9 .016E+12 9 .016E+12 9 016E+12 9 .015E+12 9 .015E+12  3. 034E3. 015E3..162E4..731E1•015E.  13 13 13 13 12  . 09 4,.474E- 14 -3. 014E- 12 - 1.960E, 09 -1 . 044E- 13 -2, 667E- 12 - 1.955E-2. . 125E-12 - 1.435E. 09 - 1.259E. 13 -1 ..617E- 12 4..807E- 10 3 .718E- 13 -1 . 459E- 12 6..470E- 09 2 .654E- 12  41 43 45 47 49  1 .000E-01 1 .585E-01 2.. 512E-01 3 .981E-01 6 .310E-01  9 .014E+12 9 .013E+12 9 .013E+12 9 .01 1E+12 9 .010E+12  4..414E1 892E. 7. 982E2., 439E3. 987E-  12 11 11 10 10  -2, •296E-6 . 473E-3, 331E-1 ..367E-2. 008E-  51 53 55 57 59  1 .OOOE+00 1 .468E+00 2 .154E+00 3 . 162E + 00 4 .642E+00  9 .009E+12 9 .009E+12 9 .008E+12 9 .008E+12 9 .008E+12  2,.183E1 687E, 1.484E. 2,.885E4. 660E-  1 .894E- 10 10 -3. 593E- 11 9 .046E- 07 . 07 -9 .289E- 12 11 -7 ..231E- 13 - 1.498E1 .985E- 12 12 2..352E- 13 -5 .094E- 09 12 3..245E- 12 - 1 .320E- 08 -1 .407E- 12 12 5..644E- 12 -2 .948E- 09 -2 .454E- 12  61 63 65 67 69  6 .813E+00 1 .OOOE+01 1 .445E+01 2 .089E+01 3 .020E+01  9 .007E+12 9 .007E+12 9 .006E+12 9 .006E+12 9 .006E+12  7..892E1.077.E, 1.629E. 6,, 438E3..093E-  13 13 13 14 14  4,.315E1.126E. 2 .533E9.. 564E4..832E-  13 1 .300E- 08 3 .951E13 2 .182E- 09 -4 .414E13 -6 .905E- 09 -4 .415E14 - 1.693E. 09. -4 .608E14 -6 .982E- 1 1 5 .604E-  71 73 75 77 79  4 .365E+01 6 .310E+01 8 .610E+01 9 .750E+01 9 .950E+01  9 .O05E+12 9 .005E+12 9 .005E+12 9 .005E+12 9 .005E+12  1. .461E6..861E3..765E3..350E2. 093E-  14 15 15 15 15  2..458E- 14 6 .539E- 1 1 7 .685E. 1 1 6 .241E1. .182E-14 1.991E6..243E- 15 -3 .169E- 12 1 .242E6. 583E- 15 -4 .903E- 12 -2 .625E7 .322E. 16 4 . 103E-13 1 .325E-  09 09 09 10 10  Page 233  Convergence checks  CN, T i O , H20 (hm)  4,,078E3,. 984E3. 849E3 . 256E2,.071E-  08 7 .791E12 1.608E. 1.982E. 12 3..412E- 08 1 1 3 .288E- 08 3 .047E10 -2 .320E- 07 - 1 .025E10 -2. .107E- 07 -2 .192E-  13 13 13 13 13  10 10 10 11 11  3..158E- 1 1 8. 408E- 11 1. .148E-10 1,390E. 10 10 1 654E.  Chi_H/RM  Hconv/H  Lum/L'  0..OOOE+00 0 .OOOE+00 0 .OOOE+00 0 .OOOE+00 0 .OOOE+00  9 .889E-01 9.889E-01 9.889E-01 9.889E-01 9.889E-01  1 3 5 7 9  1 . 352E + 01 1.OOOE+00 . 1 .OOOE+OO 1.121E+01 1.OOOE+00 9 . 322E + 00 1.OOOE+00 7.. 782E+00 1.OOOE+00 6..543E+00  0 .OOOE+00 0 .OOOE+OO 0 .OOOE+00 0 .OOOE+00 0 .OOOE+00  9.889E-01 9.889E-01 9 .889E-01 9 .888E-01 9.888E-01  11 13 15 17 19  1.OOOE+00 5 . 549E+00 1.OOOE+00 4 .752E+00 1.OOOE+00 4..111E+00 1 .OOOE + 00 3 . 593E + 00 1 .OOOE + 00 3 .178E+00  0 .OOOE+00 0 .OOOE+00 0 .OOOE+OO 0 .OOOE+00 0 .OOOE+00  9.888E-01 9.888E-01 9.888E-01 9.888E-01 9.888E-01  21 23 25 27 29  3 . 456E+01 2..864E+01 2 . 376E+01 1 .970E+01 1.632E+01 .  10 10 10 11 10  1.OOOE+00 1 .OOOE + 00 1.OOOE+00 1 .OOOE+00 1.OOOE+00  2 .846E+00 2 .578E+00 2 .358E+00 2 .173E+00 2 .011E+00  0 .OOOE+OO 0 .OOOE+00 4 .196E-09 2 .843E-07 1 .145E-06  9.888E-01 9.888E-01 9.887E-01 9.887E-01 9.887E-01  31 33 35 37 39  1. .540E- 09 3..269E- 09 3.,150E- 09 2..222E- 08 2 ,019E- 08  1.OOOE+00 1 .OOOE + 00 1.OOOE+OO 1.OOOE+00 1.OOOE+00  1 .869E+00 1 .753E+00 1 .692E+00 1 .681E+00 1 .685E+00  1 .926E-06 2 .195E-06 1 .932E-06 1 .224E-06 2 .702E-06  9.886E-01 9.886E-01 9.885E-01 9.885E-01 9.903E-01  41 43 45 47 49  8 .665E- 08 08 1.435E. 4 .879E- 10 1 .264E- 09 2 .824E- 10  1.OOOE+00 1.OOOE+00 1.OOOE+00 9.999E-01 9.999E-01  1 .437E+00 1 .145E+00 1 .024E+00 9 .889E-01 9 .936E-01  1 .480E-05 1 .680E-04 8 .838E-03 1 .203E-01 3 .453E-01  1 .005E + 00 1 .010E + 00 1.009E+00 1.008E+00 1.006E+00  51 53 55 57 59  12 14 13 15 14  1.245E. 09 2 .090E- 10 6 .614E- 10 1 621E- 10 6 687E- 12  9.999E-01 9 .999E-01 9 .998E-01 9.996E-01 9.991E-01  9 .930E-01 9 .961E-01 9 .997E-01 1 .001E+00 1 .001E+00  2 . 309E-01 6 . 763E-02 2 .296E-01 2 . 401E + 00 1 . 285E + 01  1.006E+00 1 .005E+00 1.004E+00 1.003E+00 1.002E+00  61 63 65 67 69  14 14 14 15 16  6..263E1 .907E3 .036E4 696E3 .935E-  9.978E-01 9.944E-01 9.876E-01 9.830E-01 9.821E-01  1 .002E+00 1 .001E+00 1 .OOOE+00 1 .OOOE+00 9 .999E-01  5 .278E+01 2 . 185E + 02 7 .637E+02 1 .284E+03 1 . 402E + 03  1.001E+00 1.001E+00 1.OOOE+00 1.OOOE+00 1.OOOE+00  71 73 75 77 79  12 11 1 1 10 10  1 878E, 1 872E. 1 .374E4..604E6,.197E-  12 12 13 13 14  ^ co  Model 05310191  SS 3000./ 2.00/  -3.00  Page 234  Convective q u a n t i t i e s  CN, T i O , H20 (hm)  Vconv  Hconv  2 .643E-05 4 . 242E-05 6 . 764E-05 1 .074E-04 1 .700E-04  0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO  0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO  0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO  1 3 5 7 9  0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO  2 .682E-04 4 .222E-04 6 .627E-04 1 .037E-03 1 .617E-03  0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO  0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO  0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO  11 13 15 17 19  5.. 237E -02 5.. 294E -02 5., 357E -02 5.. 428E -02 5.. 507E -02  0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO  2 .514E-03 3 .897E-03 6 .027E-03 9 . 306E-03 1 .435E-02  0 .OOOE+OO 0..OOOE+OO 0. OOOE+OO 0 OOOE+OO 0. OOOE+OO  0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 0..OOOE+OO  0 .OOOE+OO 0..OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO  21 23 25 27 29  4 . 725E -02 5 . 304E -02 6 .027E -02 6. 864E -02 7 .. 709E -02  5. 597E -02 5.. 700E -02 5. 825E -02 5. 984E -02 6. 200E -02  0 .OOOE+OO 0 .OOOE+OO 6 .027E-02 6 860E-02 7 702E-02  2 .210E-02 3 . 396E-02 5 200E-02 7. 898E-02 1. .177E-01  0. OOOE+OO 0. OOOE+OO 5.. 858E+02 2.. 165E+03 3. 124E+03  0,.OOOE+OO 0,.OOOE+OO 1,.532E+00 1. 038E+02 4..179E+02  0 .OOOE+OO 0 .OOOE+OO 4,.196E-09 2..843E-07 1. 145E-06 .  31 33 35 37 39  .3.241E + 00 2.964E+00 2.579E+00 2.127E+00 1.685E+00  8. 441E -02 9. 079E -02 9 ,. 705E -02 1 061E . -01 1. .328E -01  6. 512E -02 6 .992E -02 7 . 779E -02 9 . 165E -02 1 179E . -01  8. 433E-02 9. 072E-02 9. 699E-02 1 061E-01 . 1 .327E-01  1, 710E-01 . 2, 428E-01 3.. 381E-01 4. 852E-01 8. 271E-01  3. 3. 2. 2. 2.  7. 031E+02 8. 016E+02 7. 059E+02 4. 473E+02 9. 876E+02  1 926E-06 . 2..195E-06 1 932E-06 . 1 224E-06 . 2: 702E-06  41 43 45 47 49  2.303E+09 2.451E+09 2.597E+09 2.761E+09 2.957E+09  1.348E+00 1.199E+00 1.109E+00 1.055E+00 1.027E+00  1 862E . -01 2. 510E -01 3. 821E -01 5. 9 11E -01 7. 621E -01  -01 1 667E . 2. 168E -01 2. 727E -01 3. 271E -01 3. 663E -01  1.861E-01 . 2. 499E-01 3..650E-01 4. 866E-01 5. 477E-01  2. 010E+00 4..395E+00 9. 184E+00 1 854E+01 . 3. 123E+01  4. 429E+03 1 020E+04 . 3. 972E+04 9. 874E+04 1 444E+05 .  5. 6. 3. 4. 1  410E+03 143E+04 231E+06 399E+07 263E+08 .  1 1 8. 1 3.  480E-05 . 680E-04 . 838E-03 203E-01 . 453E-01  51 53 55 57 59  1.639E+00 1.653E+00 1.646E+00 1.618E+00 1.553E+00  3.189E+09 3.456E+09 3.749E+09 4.081E+09 4.453E+09  1.013E+00 1.005E+00 1.004E+00 1.006E+00 1.016E+00  7. 045E -01 5. 686E -01 6. 360E -01 1 037E+0O . 1 868E+0O .  3. 863E -01 3. 932E -01 3. 912E -01 3. 795E -01 3. 508E -01  5. 472E-01 5. 054E-01 5. 099E-01 5. 043E-01 4. 434E-01  3. 817E+01 4 . 157E+01 6 . 367E+01 1 449E+02 . 3. 690E+02  1 276E+05 . 8. 389E+04 1 233E+05 . 2. 647E+05 4. 543E+05  8. 2. 8. 8. 4.  445E+07 474E+07 399E+07 782E+08 700E+09  2. 309E-01 6 .763E-02 2. 296E-01 2. 401E+00 1 285E+01 .  61 63 65 67 69  1.448E+00 1.333E+00 1.261E+00 1.242E+00 1.239E+00  4.875E+09 5.364E+09 5.859E+09 6.088E+09 6.128E+09  1.042E+00 1.107E+00 1.224E+00 1.296E+00 1.309E+00  3. 374E+00 6. 095E+00 9. 897E+00 1 201E+01 . 1 241E+01 .  2. 974E -01 2. 266E -01 -01 1 707E . 1 523E . -01 1 496E . -01  3. 536E-01 2. 548E-01 1 842E-01 . 1 621E-01 . 1 589E-01 .  9. 479E+02 2. 439E+03 5. 438E+03 7 .500E+03 7 . 904E+03  7 .011E + 05 1 054E+06 . 1 494E+06 . 1 731E+06 . 1 775E+06 .  5. 278E+01 1 931E+10 . 7. 995E+10 2. 185E+02 2. 794E+1 1 7 .637E+02 4. 699E+11 1 284E+03 . 5. 128E+1 1 1 402E+03 .  71 73 75 77 79  TauRM  Cp/Cv  DELad  DELbub  1 3 5 7 9  1 .000E-05 1 .585E-05 2 .512E-05 3 .981E-05 6 .310E-05  1 .207E+00 1.210E+00 1.212E+00 1.213E+00 1.214E+00  1.257E+09 1.223E+09 1.196E+09 1.175E+09 1.158E+09  3.600E+00 3.715E+00 3.798E+00 3.855E+00 3.889E+00  1 .485E -02 1 .676E -02 1 . 830E -02 1 .989E -02 2 . 159E-02  5 . 237E -02 5 . 152E-02 5 .099E -02 5 .070E -02 5 .060E -02  0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO  11 13 15 17 19  1 .OOOE-04 1 .585E-04 2 .512E-04 3 .981E-04 6 .310E-04  1.215E+00 1.216E+00 1.217E+00 1.217E+00 1.218E+00  1.144E+09 1.135E+09 1.128E+09 1.124E+09 1.123E+09  3.904E+00 3.905E+00 3.894E+00 3.874E+00 3.849E+00  2 . 339E -02 2 . 525E -02 2 . 708E -02 2 . 884E -02 3 .059E -02  5 .064E -02 5..081E -02 5.. 108E-02 5.. 143E-02 5.. 187E-02  21 23 25 27 29  1 .OOOE-03 1 .585E-03 2 .512E-03 3 .981E-03 6 .310E-O3  1.218E+00 1.219E+00 1.220E+00 1.221E+00 1.222E+00  1.124E+09 1.127E+09 1.134E+09 1.143E+09 1.157E+09  3.819E+00 3.787E+0O 3.752E+00 3.718E+00 3.683E+0O  3 . 231E-02 3 .428E -02 3 .657E -02 3 .933E -02 4 . 280E -02  31 33 35 37 39  1 .000E-02 1 .585E-02 2 .512E-02 3 .981E-02 6 .310E-02  1.224E+00 1.227E+00 1.230E+00 1.233E+00 1.235E+00  1.177E+09 1.205E+09 1.243E+09 1.296E+09 1.367E+09  3.649E+00 3.615E+00 3.575E+00 3.517E+00 3.416E+00  41 43 45 47 49  1 ;000E-01 1.585E-01 . 2 .512E-01 3 .981E-01 6..310E-01  1.235E+00 1.233E+00 T.227E+00 1.225E+00 1.236E+00  1.461E+09 1.578E+09 1.722E+09 1.895E+09 2.094E+09  51 53 55 57 59  1.OOOE+OO . 1.468E+00 . 2..154E+00 3.. 162E + 00 4..642E+00  1.282E+00 1.345E+00 1.428E+00 1.522E+00 1.598E+00  61 63 65 67 69  6. 813E+00 1 OOOE+01 . 1 445E+01 . 2. 089E+01 3. 020E+01  71 73 75 77 79  4..365E+01 6.,310E+01 8. 610E+01 9. 750E+01 9. 950E+01  H_P  Qconv  DELrad  TauRMb  365E+03 175E+03 736E+03 1 12E + 03 535E+03  Hconv/H  Mode"! 06310191  SS 3500./  TauRM  1.00/  Col Mass  -3.00  r/R*  Summary of p h y s i c a l  CN, T i O , H20 (hm)  - 1  T  Pe  n  quantities  P_rad  P_gas  Page 235 Rho  Mu  ChiRM  1 3 5 7 9  1 1 2 3 6  000E-05 585E-05 512E-05 981E-05 310E-05  5 7 1 1 1  1 968E -03 772E-01 1 196E -03 668E-01 024E+00' 4 061E -04 369E+00 -3 958E -04 833E+00 - 1 206E -03  2 2 2 2 2  670E+03 687E+03 708E+03 733E+03 760E+03  1 2 2 3 4  559E+13 059E+13 728E+13 620E+13 802E+13  3 5 6 9 1  984E-05 227E-05 952E-05 330E-05 261E-04  5 7 1 1 1  748E+00 638E+00 020E+01 366E+01 830E+01  2 2 2 2 2  031E-01 035E-01 042E-01 050E-01 061E-01  1 1 1 1 1  242E+00 243E+00 243E+00 243E+00 244E+00  3 4 5 7 9  217E - 11 249E - 11 631E - 11 474E - 11 918E -11  2 3 3 4 5  841E316E895E594E438E-  1 1 13 15 17 19  1 1 2 3 6  000E-04 585E-04 512E-04 981E-04 310E-04  2 3 4 5 7  452E+00 276E+00 366E+O0 800E+00 672E+00  -2 -2 -3 -4 -5  023E -03 843E -03 664E -03 485E -03 301E -03  2 2 2 2 2  790E+03 821E+03 854E+03 889E+03 924E+03  6 8 1 1 1  364E+13 418E+13 111E+14 460E+14 910E+14  1 2 3 4 5  712E-04 333E-04 182E-04 340E-04 909E-04  2 3 4 5 7  451E+01 279E+01 376E+01 821E+01 711E+01  2 2 2 2 2  075E-01 091E-01 110E-01 133E-01 160E-01  1 1 1 1 1  244E+00 244E+00 245E+00 245E+00 245E+O0  1 1 2 3 3  315E - 10 739E - 10 295E - 10 017E - 10 949E - 10  6 7 9 1 1  463E- 05 718E- 05 270E- 05 121E- 04 366E- 04  1 1 13 15 17 19  21 23 25 27 29  1 1 2 3 6  000E-03 585E-03 512E-03 981E-03 310E-03  1 1 1 2 2  010E+01 321E+01 719E+01 223E+01 856E+01  -6 -6 -7 -8 -9  1 1 1E -03 913E -03 706E -03 489E -03 261E -03  2 2 3 3 3  960E+03 996E+03 034E+03 071E+03 109E+03  2 3 4 5 6  487E+14 220E+14 145E+14 301E+14 738E+14  8 1 1 1 2  024E-04 085E-03 463E-03 962E-03 622E-03  1 1 1 2 2  016E+02 332E+02 736E+02 248E+02 892E+02  2 192E-01  2 271E-01 2 322E-01 2 383E-01  1 1 1 1 1  245E+00 245E+00 246E+00 246E+00 246E+00  5 6 8 1 1  143E - 10 660E - 10 573E - 10 097E -09 394E -09  1 2 2 3 4  677E- 04 075E- 04 587E- 04 249E- 04 108E- 04  21 23 25 27 29  31 33 35 37 39  1 1 2 3 6  000E-02 585E-02 512E-02 981E-02 310E-02  3 4 5 7 9  647E+01 628E+01 835E+01 303E+01 067E+01  -1 -1 -1 - 1 -1  002E -02 077E -02 151E -02 223E -02 294E -02  3 3 3 3 3  149E+03 190E+03 233E+03 280E+03 332E+03  8 1 1 1 2  509E+14 067E+15 329E+15 642E+15 010E+15  3 4 6 8 1  496E-03 656E-03 208E-03 309E-03 120E-02  3 4 5 7 9  699E+02 701E+02 934E+02 438E+02 247E+02  2 2 2 2 3  457E-01 549E-01 665E-01 815E-01 009E-01  1 1 1 1 1  246E+00 246E+00 246E+00 246E+00 246E+00  1 2 2 3 4  761E -09 5 209E -09 6 751E -09 8 1 398E -09 158E -09 . 1  231E- 04 710E- 04 688E- 04 138E- 03 513E- 03  31 33 35 37 39  41 43 45 47 49  1 1 2 3 6  000E-01 585E-01 512E-01 981E-01 310E-01  1 1 1 1 2  1 15E+02 - 1 363E -02 355E+02 - 1 429E -02 623E+02 - 1 491E -02 904E+02 - 1 548E -02 175E+02 - 1 597E -02  3 3 3 3 3  391E+03 461E+03 547E+03 660E+03 811E+03  2 2 3 3 4  432E+15 901E+15 393E+15 862E+15 240E+15  1 2 3 4 7  528E-02 123E-02 049E-02 593E-02 281E-02  1 1 1 1 2  139E+03 386E+03 662E+03 951E+03 231E+03  3 3 4 4 5  261E-01 591E-01 024E-01 609E-01 441E-01  1 1 1 1 1  245E+00 245E+00 244E+00 243E+00 242E+00  5 5 7 7 8  029E -09 995E -09 008E -09 969E -09 741E -09  2 2 4 6 1  050E- 03 856E- 03 164E- 03 546E- 03 120E- 02  41 43 45 47 49  51 53 55 57 59  1 1 2 3 4  OOOE+00 468E+00 154E+00 162E+00 642E+00  2 2 2 3 4  424E+02 640E+02 940E+02 442E+02 200E+02  - 1 638E -02 - 1 673E -02 - 1 720E -02 - 1 792E -02 - 1 890E -02  4 4 4 4 5  019E+03 247E+03 513E+03 809E+03 154E+03  4 4 4 5 6  484E+15 624E+15 849E+15 332E+15 082E+15  1 1 2 3 6  199E-01 802E-01 553E-01 623E-01 475E-01  2 2 3 3 4  488E+03 71 1E+03 021E+03 541E+03 328E+03  6 8 1 1 1  691E-01 245E-01 042E+00 343E+00 779E+00  1 1 1 1 1  241E+00 240E+00 239E+00 239E+00 239E+00  9 9 9 1 1  236E -09 519E -09 979E -09 097E -08 251E -08  1 2 2 1 2  888E- 02 334E- 02 179E- 02 898E- 02 135E- 02  51 53 55 57 59  61 63 65 67 69  6 1 1 2 3  813E+00 OOOE+01 445E+01 089E+01 020E+01  4 5 5 6 6  985E+02 532E+02 842E+02 016E+02 110E+02  - 1 980E -02 -2 039E -02 -2 073E -02 -2 092E -02 -2 104E -02  5 6 6 7 7  564E+03 041E+03 562E+03 149E+03 802E+03  6 6 6 6 5  695E+15 849E+15 660E+15 295E+15 856E+15  1 5 1 4 1  665E+00 181E+00 586E+01 681E+01 296E+02  5 5 6 6 6  143E+03 712E+03 034E+03 213E+03 308E+03  2 3 4 6 9  422E+00 370E+00 698E+00 621E+00 401E+00  1 1 1 1 1  239E+00 238E+00 236E+00 230E+00 214E+00  1 1 1 1 1  377E -08 408E -08 367E -08 285E -08 180E -08  3 9 2 6 1  829E- 02 150E- 02 310E- 01 086E- 01 674E+0O  61 63 65 67 69  71 73 75 77 79  4 6 8 9 9  365E+01 310E+01 610E+01 750E+01 950E+01  6 6 6 6 6  158E+02 182E+02 194E+02 198E+02 198E+02  -2 -2 -2 -2 -2  8 9 1 1 1  526E+03 327E+03 007E+04 038E+04 043E+04  5 4 4 4 4  398E+15 950E+15 590E+15 452E+15 430E+15  3 7 1 1 1  298E+02 514E+02 333E+03 610E+03 656E+03  6 6 6 6 6  354E+03 374E+03 380E+03 380E+03 380E+03  1 1 2 2 3  342E+01 923E+01 61 1E + 01 952E+01 012E+01  1 1 9 9 9  175E+00 093E+00 803E-01 266E-01 177E-01  1 8 7 6 6  053E -08 985E -09 472E -09 850E -09 750E -09  4 1 2 3 3  767E+00 328E+01 806E+01 567E+01 694E+01  71 73 75 77 79  11'OE -02 114E -02 116E -02 1 17E -02 117E -02  2 228E-01  05 05 05 05 05  1 3 5 7 9  to CO  Model 06310191  SS 3500./  TauRM  r  1.00/  -3.00  TE res  RE res  HE res  DE res  1 3 5 7 9  1 .000E-05 1 .585E-05 2 .512E-05 3 .981E-05 6 . 310E-05  6 .633E+12 6 .628E+12 6 .623E+12 6 .617E+12 6 .612E+12  6 .072E- 08 - 1 .743E- 09 2 .588E- 06 0 .OOOE+OO 7 .153E- 06 -3 .526E- 07 -6 .563E- 09 -3 .666E- 10 2 .818E- 10 -2 .394E- 10 8 .581E- 08 -1 . 201E- 10 1 .351E- 10 -6 .470E- 11 9 .153E- 08 7 . 852E- 12 5 .935E- 1 1- 1 .738E- 11 6 .323E- 08 5 . 107E-11  11 13 15 17 19  1 .OOOE-04 1 .585E-04 2 .512E-04 3 .981E-04 6 .310E-04  6 .607E+12 6 .601E+12 6 .596E+12 6 .590E+12 6 .585E+12  2 .046E8 .620E3 .OOOE1 .307E5 .803E-  1 1-5 .710E12 -2 .270E12 - 1.435E12 -9 .386E13 -5 .207E-  21 23 25 27 29  1 .000E-03 6 .580E+12 1 .585E-03 6 .574E+12 2 .512E-03 6..569E+12 3 .981E-03 6 .564E+12 , 6 . 310E-03 .6..559E+12  2 . 361E8 .536E2 .607E1 .157E1.378E.  13 14 14 14 14  -2 -9 -4 -2 -2  31 33 35 37 39  1 .000E-02 1 .585E-02 2 .512E-02 3 .981E-02 6 . 310E-02  6..554E+12 6 ,549E+12 . 6,. 544E+12 6,.539E+12 6,. 534E+12  1 .543E2..046E2..693E3..322E3 .378E-  14 14 14 14 14  -3 .666E- 14 -6 .838E- 14 - 1. 120E-13 - 1 .610E- 13 -2 .046E- 13  41 43 45 47 49  1 .000E-01 1 .585E-01 2 .512E-01 3 .981E-01 6 . 310E-01  6 . 530E+12 6..525E+12 6 . 521E+12 6 . 518E+12 6. 514E+12  2..586E1.953E. 2. 543E3..876E5..765E-  14 14 14 14 14  -2. .294E-2 .197E- 1.728E, - 1 088E. -5. 483E-  13 13 13 13 14  51 53 55 57 59  1.OOOE+OO . 1. .468E + 00 2,.154E+00 3.. 162E + 00 4., 642E+00  6 6 6 6 6  . 512E+12 ,. 509E+12 .506E+12 .501E+12 .495E+12  6. 054E3. 576E5.. 705E4. 576E2. 462E-  14 14 14 14 14  -3. -4. -4. -3. -2.  61 63 65 67 69  6 .813E+00 . 1.OOOE+01 . 1 445E+01 . 2. 089E+01 3. 020E+01  6. 489E+12 6 .485E+12 6. 483E+12 6 . 481E+12 6. 481E+12  1 551E. 1 257E. 1 036E. 8. 787E7 .602E-  71 73 75 77 79  4.. 365E+01 6.. 310E+01 8. 610E+01 9.. 750E+01 9. 950E+01  6 .480E+12 6. 480E+12 6. 480E+12 6. 480E+12 6. 480E+12  6. 728E6. 026E5. 460E4. 814E3. 328E-  .420E.682E.095E.090E.086E-  12 3 .074E- 08 12 1 .089E- 08 12 1 .460E- 09 13 - 1.631E- 09 13 -2 .424E- 09  4 .089E2 .310E1 .084E5 .696E3 .259E-  13 14 14 14 14  1 .827E- 12 9 .529E- 13 4.. 171E-13 1.633E, 13 4..759E- 14  -2 . 335E- 09 -1 .922E- 09 - 1.425E- 09 -9 .919E- 10 -6 .630E- 10  Page 236  Convergence checks  CN, T i O , H20 (hm)  1 1 11 1 1 12 12  Ch i_H/RM  Hconv/H  Lum/L*  EPS rms  g_eff/g  2 . 551E- 07 7 .054E- 06 8 .224E- 09 8 .768E- 09 6 .057E- 09  9 .998E -01 9 .998E -01 9 .998E -01 9 .998E -01 9 .998E -01  2 .452E+02 2 .287E+02 2 .007E+02 1 .682E+02 1 .355E+02  0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO  1 .008E+00 1 .008E+00 1 .008E+00 1 .008E+00 1 .008E+00  1 3 5 7 9  2 .944E- 09 1 .043E- 09 1 . 399E- 10 1 .562E- 10 2 .322E- 10  9 .998E -01 9 .998E -01 9 .998E -01 9 .999E -01 9 .999E -01  1 .061E+02 8 .112E+01 6 .104E+01 4 .543E+01 3 .359E+01  0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO  1 .008E+00 1 .008E+00 1 .008E+00 1 .008E+00 1 .008E+00  1 1 13 15 17 19  2 .237E1 .841E1 .365E9 .501E6 .351E-  10 10 10 1 1 1 1  9 .999E -01 9 .999E -01 9 .999E -01 9 .999E -01 9 .999E -01  2 .478E+01 1 .831E+01 1 . 362E+01 1 .024E+01 7 . 820E+00  0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO  1 .008E+00 1 .008E+00 1 .007E+00 1 .007E+00 1 .007E+00  21 23 25 27 29  -3, .298E- 1 444E. -2. .604E-3, 736E-4. .553E-  14 13 13 13 13  4..243E2..971E1.998E. 1 . 206E4 .998E-  11 1 1 11 1 1 12  9 .999E -01 9 . 999E -01 9.. 999E -01 9 .999E -01 9 999E -01  6 .071E+00 4 .804E+00 3 .864E+00 3 .146E+00 2 .579E+00  0 .OOOE+OO 0..OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 0. OOOE+OO  1 .007E+00 1 .007E+00 1 .007E+00 1 .007E+00 1 .008E+00  31 33 35 37 39  11 -5. 027E11 -6..148E1 1 - 1.121E. 1 1-2. 401E10 - 1 063E,  13 13 12 12 12  1.289E. 5..003E4..755E6. 933E6..375E-  12 12 12 12 1 1  9 .999E -01 9 .999E -01 9 998E -01 9, 998E -01 9 996E -01  2 .117E+00 1 .741E+00 1 .456E+00 1 . 268E + 00 1 .164E+00  0. OOOE+OO 0..OOOE+OO 0. OOOE+OO 0. OOOE+OO 4,, 280E- 10  1 .008E+00 1 .008E+00 1 .008E+00 1 .008E+00 1 .007E+00  41 43 45 47 49  880E793E717E516E059E-  14 1 468E. 09 7. 848E14 1.969E, 11 1 771E. 14 -1 .,826E- 09 - 1 028E. 14 -2. .754E- 09 - 1 274E. 14 2..408E- 09 - 3 . 826E-  12 12 11 11 12  1 1 1 2. 2.  406E. 894E. 749E. 637E307E-  10 12 10 10 10  9 994E -01 9 .993E -01 9 .994E -01 9 .995E -01 -01 9 ,994E ,  1. .133E + 00 1.106E+00 , 1 031E+00 . 9 .923E-01 9 .904E-01  1 023E-06 . 2. 669E-06 1.109E-07 . 0. OOOE+OO 0. OOOE+OO  1 .006E+00 1 .006E+00 1 .006E+00 1 .006E+00 1 .005E+00  51 53 55 57 59  14 14 14 15 15  - 1 593E. - 1 219E. - 9 . 377E-7. 662E-6. .497E-  14 2.,854E- 09 1 836E. 14 2.,848E- 10 9.,046E15 -6. 529E- 1 1 9. 749E15 -3. 444E- 1 1- 1 703E. 15 -7 . 031E- 12 - 1 305E.  12 13 15 13 13  2. 2. 6. 3. 6.  734E728E254E299E737E-  10 11 12 12 13  9. 989E -01 9 .974E -01 9 .934E -01 9. 827E -01 9 . 524E -01  9. 967E-01 1 OOOE+OO . 1 001E+00 . 1 002E+00 . 1 002E+00 .  4. 935E-05 2. 594E-02 8. 657E-01 7 . 542E+00 5. 017E+01  1 .003E+00 1 .002E+00 1 .001E+00 9 .997E-01 9 .986E-01  61 63 65 67 69  15 15 15 15 15  -5. -5. -4. -4. -4.  15 15 15 15 15  6. 391E- 14 8. 645E -01 1 1 . 18E-13 6 . 229E -01 8. 713E- 14 2. 037E -01 5. 021E- 15 - 1 223E . -02 3. 314E- 15 -4. 797E -02  1 001E+00 . 1 OOOE+OO . 1 OOOE+OO . 9 .999E-01 9 .996E-01  3. 813E+02 4. 267E+03 1 .104E+05 0. OOOE+OO 0. OOOE+OO  9 .979E-01 9 .984E-01 9..995E-01 9 .999E-01 1.OOOE+OO .  71 73 75 77 79  681E083E736E643E765E-  -4 .430E-3 .102E-2 .086E- 1.259E-5 .218E1.345E. 5 .223E4 .963E7 .234E6 .655E-  6. 632E-1 . 165E- 9 . 079E-1 . 643E- 1 294E.  10 10 10 10 11  13 -2. 254E- 14 12 3. 120E- 14 13 5. 839E- 15 14 6. 945E- 17 . 17 16 1 792E-  to  CO  Model 06310191  SS 3 5 0 0 . /  TauRM  Cp/Cv  1 3 5 7 9  1 .000E-05 1 .585E-05 2 .512E-05 3 .981E-05 6 .310E-05  1 .472E+00 1 .461E+00 1 .451E+00 1 .442E+00 1 .434E+00  11 13 15 17 19  1 .000E-04 1 .585E-04 2 .512E-04 3 .981E-04 6 .310E-04  21 23 25 27 29  1.00/  -3.00  Convective q u a n t i t i e s  CN, T i O , H20 (hm)  Vconv  Hconv  2 . 715E-05 4 . 170E-05 6 .489E-05 1 .018E-04 1 .606E-O4  0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO  0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO  0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO  1 3 5 7 9  0 . OOOE + OO 0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO  2 .546E-04 4 .051E-04 6 .475E-04 1 .039E-03 1 .672E-03  0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO  0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO  0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 0 :OOOE+OO 0 .OOOE+OO  11 13 15 17 19  01 01 01 01 01  0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 0..OOOE+OO  2 .700E-O3 4 . 370E-03 7 .085E-03 1 . 150E-02 1 .868E-02  0 . OOOE + OO 0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO  0 .OOOE+OO 0..OOOE+OO 0. OOOE+OO 0..OOOE+OO 0 .OOOE+OO  0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO  21 23 25 27 29  01 01 01 01 01  0 .OOOE+OO 0 .OOOE+OO 0..OOOE+OO 0 .OOOE+OO 0 .OOOE+OO  3 .036E-02 4 .941E-02 8 .064E-02 1 .322E-01 2 .181E-01  0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO  0. OOOE+OO 0. OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 0 OOOE+OO  0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO  31 33 35 37 39  9.. 253E-02 1. .171E-01 1 618E-01 . 2. 437E-01 3, 904E-01  2,.934E- 01 3 057E- 01 3 224E- 01 3,.432E- 01 3 .649E- 01  0..OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 0".OOOE+OO 3.•904E-01  3 .635E-01 6 . 155E-01 1 .075E+00 1 .982E+00 3 . 874E + 00  0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 3 .974E+02  0. OOOE+OO 0 OOOE+OO 0 OOOE+OO 0..OOOE+OO 2..993E-01  0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 4 .280E- 10  41 43 45 47 49  1 .014E+00 1 .010E+00 1 .005E+00 1 .002E+00 1 .002E+00  5. 919E-01 6. 315E-01 4. 821E-01 3. 587E-01 3. 729E-01  3. 825E- 01 3. 916E- 01 3. 957E- 01 3. 966E- 01 3. 933E- 01  5. 916E-01 6 309E-01 4.•820E-01 0. OOOE+OO 0. OOOE+OO  7 .275E+00 5 . 301E + 03 7 . 162E+02 9.. 792E + 00 7 .. 281E + 03 1. 869E+03 1 018E+01 . 2..492E+03 7. 777E+01 1 038E+01 . 0 OOOE+OO 0. OOOE+OO 1:423E+01 . 0..OOOE+OO 0. OOOE+OO  1. 023E-06 2. 669E-06 1.109E-07 . 0. OOOE+OO 0. OOOE+OO  51 53 55 57 59  1 .004E+00 1 .012E+00 1 .O34E+00 1 .091E+00 1 .232E+00  5. 963E-01 1 149E+00 . 2. 208E+00 4. 311E+00 8. 839E+00  3. 828E- 01 3. 565E- 01 3. 053E- 01 2..312E- 01 1 595E. 01  5. 939E-01 1 015E+00 . 1 064E+00 . 6 .846E-01 3. 493E-01  3, 029E+01 8. 030E+01 2., 140E+02 5., 806E+O2 1 621E+03 .  1 693E+04 . 1. 325E . + 05 4. 090E+05 7 .828E+05 1. 338E . + 06  3. 477E+04 1 830E+07 . 6. 112E+08 5. 327E+09 3. 544E+10  4. 935E-05 2.. 594E-02 8. 657E-01 7 .. 542E + 00 5. 017E+01  61 63 65 67 69  1 .540E+00 1 995E+01 . 2 .115E+00 5. 819E+01 2 .735E+00 3. 482E+02 2 .927E+00 -6 . 564E+03 . 2..950E+00 - 1 199E+03  1 1 . 19E-01 8. 819E- 02 8. 097E- 02 8. 057E- 02 8. 064E- 02  1 792E-01 . 1 147E-01 . 1 046E-01 . 0. OOOE+OO 0. OOOE+OO  4. 654E+03 2..428E+06 1. .302E + 04 5. 288E+06 . 2.. 755E + 04 1 616E+07 3.. 505E+04 0. OOOE+OO 3. 630E+04 0. OOOE+OO  2. 694E+11 3. 015E+12 7 . 799E+13 0. OOOE+OO 0. OOOE+OO  3. 813E+02 4. 267E+03 1. .104E + 05 0. OOOE+OO 0. OOOE+OO  71 73 75 77 79  Qconv  DELrad  1 .857E+10 1 .850E+10 1 .849E+10 1 .853E+10 1 .861E+10  1 .038E+00 1 .044E+00 1 .049E+00 1 .054E+00 1 .059E+00  1 .971E-02 2 .519E-02 2 .903E-02 3 .238E-02 3 .527E-02  2 .881E2 .868E2 .848E2 .824E2 .801E-  01 01 01 01 01  0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO  1 .428E+00 1 .423E+00 1 .420E+00 1 .417E+00 1 .414E+00  1 .872E+10 1 .886E+10 1 .902E+10 1 .919E+10 1 .937E+10  1 .063E+00 1 .066E+00 1 .069E+00 1 .072E+00 1 .075E+00  3 . 764E-02 3 .961E-02 4 . 121E-02 4 .258E-02 4 . 376E-02  2 .780E2 .761E2 .745E2 .731E2 .720E-  01 01 01 01 01  1 .000E-03 1 .585E-03 2 .512E-03 3 .981E-03 6 .310E-03  1 .413E+00 1 .411E+00 1 .411E+00 1 .410E+00 1 .411E+00  1 .956E+10 1 .976E+10 1 .996E+10 2 .017E+10 2 .038E+10  1 .077E+00 1 .080E+00 1 .082E+00 1 .083E+00 1 .085E+00  4 .485E-02 4,. 593E-02 4..705E-02 4,.832E-02 5,.009E-02  2 .710E2 .702E2 .696E2 .693E2 .693E-  31 33 35 37 39  1 .000E-02 1 .585E-02 2 .512E-02 3 .981E-02 6 .310E-02  1 .412E+00 1 .415E+00 1 .420E+00 1 .427E+00 1 .438E+00  2 .060E+10 2 .084E+10 2 .109E+10 2 .136E+10 2 .168E+10  1 .085E+00 1 .085E+00 1 .084E+00 1 .081E+00 1 .076E+00  5. 251E-02 5 .588E-02 6 083E-02 6.. 779E-02 7 .771E-02  2 .699E2 .713E2 .739E2 .781E2 .844E-  41 43 45 47 49  1 .000E-01 1 .585E-01 2 .512E-0T 3 .981E-01 6 .310E-01  1 .454E+00 1 .477E+00 1 .508E+00 1 .548E+00 1 .592E+00  2 .204E+10 2 .247E+10 2..301E+10 2..374E+10 2,.472E+10  1 .069E+00 1 .060E+00 1 .048E+00 1 .035E+00 1 .022E+00  51 53 55 57 59  1 OOOE+OO . 1.468E+00 . 2..154E+00 3..162E+00 4. 642E+00  V .630E+00 1 .649E+00 1 .658E+00 1 .660E+00 1 .651E+00  2. 2. 2. 3. 3.  607E+10 755E+10 926E+10 114E+10 331E+10  61 63 65 67 69  6..813E+00 1 OOOE+01 . 1,445E+01 . 2. 089E+01 3. 020E+01  V..626E+00 1 .565E+00 1 .462E+00 1 .337E+00 1 .242E+00  3. 3. 4. 4. 5.  590E+10 896E+10 237E+10 638E+10 130E+10  71 73 75 77 79  4..365E+01 6..310E+01 8. 610E+01 9. 750E+01 9. 950E+01  1 .203E+00 1 .215E+00 1 .254E+00 1 .271E+00 1.273E+00 ,  5. 795E+10 6 . 818E+10 8. 214E+10 8. 965E+10 9. 099E+10  H_P  Page 237  DELad  DELbub  TauRMb  Hconv/H  to co  Model 07310191  SS 3500./ 2 . 0 0 / -3.00  TauRM  Col Mass  r/R*  Summary of p h y s i c a l  CN, T i O , H20 (hm) T  -1  P_gas  Pe  n  Page 238  quantities Mu  P_rad  ChiRM  Rho 1 1 1 1 2  283E- 04 472E- 04 704E- 04 977E- 04 288E- 04  1 3 5 7 9  422E- 10 512E- 10 996E- 10 013E- 10 074E- 09  2 3 3 4 4  636E- 04 028E- 04 477E- 04 000E- 04 620E- 04  1 1 13 15 17 19  1 1 2 3 4  442E- 09 934E- 09 585E- 09 437E- 09 537E- 09  5 6 7 8 1  369E- 04 291E- 04 449E- 04 928E- 04 084E- 03  21 23 25 27 29  261E+00 261E+00 260E+00 260E+00 259E+00  5 7 9 1 1  940E- 09 704E- 09 894E- 09 257E- 08 579E- 08  1 1 2 2 3  333E- 03 658E- 03 087E- 03 661E- 03 445E- 03  31 33 35 37 39  1 1 1 1 1  257E+00 255E+00 253E+00 250E+00 247E+00  1 2 2 3 3  957E- 08 388E- 08 863E- 08 354E- 08 813E- 08  4 6 8 1 1  545E- 03 135E- 03 526E- 03 240E- 02 944E- 02  41 43 45 47 49  479E-01 869E-01 888E-01 280E+00 702E+00  1 1 1 1 1  244E+00 242E+00 241E+00 240E+00 239E+00  4 4 4 4 5  176E- 08 381E- 08 529E- 08 726E- 08 098E- 08  3 4 6 6 6  277E- 02 973E- 02 376E- 02 434E- 02 149E- 02  51 53 55 57 59  2 3 4 6 9  320E+00 230E+00 505E+00 349E+00 016E+00  1 1 1 1 1  239E+00 239E+00 238E+00 235E+00 228E+00  5 5 5 5 5  582E- 08 842E- 08 782E- 08 522E- 08 149E- 08  8 1 3 9 2  119E- 02 648E- 01 858E- 01 485E- 01 395E+00  61 63 65 67 69  1 1 2 2 2  287E+01 845E+01 505E+01 832E+01 889E+01  1 1 1 1 1  210E+00 170E+00 106E+00 071E+00 064E+00  4 4 3 3 3  704E- 08 187E- 08 679E- 08 457E- 08 420E- 08  6 1 3 4 4  194E+00 619E+01 519E+01 686E+01 901E+01  71 73 75 77 79  1 3 5 7 9  1 1 2 3 6  000E-05 585E-05 512E-05 981E-05 310E-05  1 2 2 3 4  794E-01 217E-01 800E-01 596E-01 685E-01  1 1 1 1 1  968E -03 913E -03 851E -03 784E -03 713E -03  2 2 2 2 2  605E+03 622E+03 641E+03 664E+03 690E+03  4 6 7 9 1  967E + 099E + 646E + 739E + 257E +  3 3 3 3 4  8 1 1 1 2  798E-05 085E-04 371E-04 765E-04 315E-04  1 2 2 3 4  787E+01 208E+01 788E+01 582E+01 667E+01  2 2 2 2 2  OOOE-01 004E-01 009E-01 016E-01 026E-O1  1 1 1 1 1  252E+00 253E+00 254E+00 255E+00 255E+00  1 032E1 269E1 592E2..029E2 620E-  11 13 15 17 19  1 1 2 3 6  OOOE-04 585E-04 512E-04 981E-04 310E-04  6 8 1 1 2  180E-01 241E-01 108E+00 501E+00 040E+00  1 1 1 1 1  638E -03 559E -03 477E -03 391E -03 304E -03  2 2 2 2 2  719E+03 751E+03 787E+03 825E+03 866E+03  1 2 2 3 5  640E + 162E + 871E + 835E + 140E +  4 4 4 4 4  3 4 5 7 1  085E-04 174E-04 717E-04 908E-04 102E-03  6 8 1 1 2  157E+01 21 1E + 01 105E+02 496E+02 034E+02  2 2 2 2 2  038E-01 053E-01 073E-01 096E-01 124E-01  1 1 1 1 1  256E+00 257E+00 258E+00 258E+00 259E+00  3 4 5 8 1  21 23 25 27 29  1 1 2 3 6  000E-03 585E-03 512E-03 981E-03 310E-03  2 3 '5 6 9  778E+00 780E+00 128E+00 922E+00 277E+00  1 1 1 9 8  215E -03 124E -03 034E -03 432E -04 534E -04  2 2 2 3 3  908E+03 953E+03 998E+03 045E+03 093E+03  6 9 1 1 2  898E + 4 247E + 4 236E + 5 642E + 5 168E + 15  1 2 3 4 5  543E-03 165E-03 038E-03 253E-03 930E-03  2 3 5 6 9  770E+02 769E+02 1 15E + 02 904E+02 256E+02  2 2 2 2 2  159E-01 199E-01 246E-01 302E-01 368E-01  1 1 1 1 1  259E+00 260E+00 260E+00 260E+00 26OE+00  31 33 35 37 39  1 1 2 3 6  000E-02 585E-02 512E-02 981E-02 310E-02  1 1 2 2 3  233E+01 625E+01 120E+01 740E+01 505E+01  7 6 5 5 4  648E -04 777E -04 922E -04 085E -04 267E -04  3 3 3 3 3  141E+03 191E+03 242E+03 297E+03 355E+03  2 3 4 6 7  838E + 15 681E + 5 728E + 5 010E + 5 554E + 5  8 1 1 2 2  226E-03 136E-02 563E-02 150E-02 967E-02  1 1 2 2 3  231E+03 621E+03 1 16E + 03 735E+03 499E+03  2 2 2 2 2  445E-01 538E-01 653E-01 797E-01 982E-01  1 1 1 1 1  41 43 45 47 49.  1 1 2 3 6  000E-01 585E-01 512E-01 981E-01 310E-01  4 5 6 8 9  432E+01 533E+01 808E+01 232E+01 729E+01  3 2 1 1 6  472E -04 704E -04 969E -04 277E -04 459E -05  3 3 3 3 3  421E+03 494E+03 580E+03 686E+03 824E+03  9 1 1 1 1  372E + 5 146E + 6 376E + 6 616E + 16 841E + 6  4 5 8 1 1  124E-02 803E-02 341E-02 244E-01 956E-01  4 5 6 8 9  426E+03 527E+03 801E+03 224E+03 720E+03  3 3 3 4 5  225E-01 547E-01 973E-01 545E-01 336E-01  51 53 55 57 59  1 1 2 3 4  OOOE+00 468E+00 154E+00 162E+00 642E+00  1 1 1 1 1  118E+02 233E+02 352E+02 506E+02 744E+02  9 -3 -7 -1 -1  636E -06 069E -05 097E -05 216E -04 948E -04  4 4 4 4 5  004E+03 201E+03 450E+03 750E+03 097E+03  2 2 2 2 2  021E + 16 124E + 16 199E + 6 296E + 6 477E + 6  3 5 8 1 1  249E-01 109E-01 008E-01 197E+00 804E+00  1 1 1 1 1  118E+04 232E+04 351E+04 506E+04 743E+04  6 7 9 1 1  61 63 65 67 69  6 1 1 2 3  813E+00 OOOE+01 445E+01 089E+01 020E+01  2 2 2 2 2  063E+02 345E+02 523E+02 631E+02 693E+02  -2 -3 -4 -4 -4  848E -04 591E -04 054E -04 340E -04 515E -04  5 5 6 7 7  504E+03 977E+03 494E+03 074E+03 721E+03  2 2 2 2 2  713E+ 6 841E + 6 814E + 16 693E + 6 526E + 6  3 9 2 8 2  547E+00 760E+00 905E+01 601E+01 419E+02  2 2 2 2 2  062E+04 344E+04 523E+04 630E+04 692E+04  71 73 75 77 79  4 6 8 9 9  365E+01 310E+01 610E+01 750E+01 950E+01  2 2 2 2 2  728E+02 748E+02 757E+02 760E+02 761E+02  -4 -4 -4 -4 -4  623E -04 689E -04 726E -04 738E -04 740E -04  8 9 9 1 1  438E+03 230E+03 963E+03 027E+04 032E+04  2 2 2 1 1  341E + 155E + 003E + 944E + 935E+  6 6 6 6 6  6 1 2 3 3  342E+02 527E+03 953E+03 745E+03 887E+03  2 2 2 2 2  727E+04 746E+04 755E+04 757E+04 758E+04  10 10 10 10 10 .  to  CO  Model 07310191 TauRM 1 3 5 7 9  SS 3500./ 2.00/ -3.00 r  CN, T i O , H20 (hm)  TE r e s  RE  r<  HE res  DE res  EPS rms  g_eff/g  Hconv/H  Ch i_H/RM  Lum/L*  07 06 07 10 09  9 9 9 9 9  999E-01 999E-01 999E-01 999E-0.1 999E-01  2 .876E+02 2 .438E+02 2 .025E+02 1 .655E+02 1 . 333E + 02  0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO  1 .007E+00 1 .007E+00 1 .007E+00 1 .007E+00 1 .007E+00  1 3 5 7 9  12 11 11 12 12  4 .150E- 09 2 .688E- 09 8 .026E- 11 2 .492E- 09 3 .268E- 09  9 9 9 9 1  999E-01 999E-01 999E-01 999E-01 OOOE+OO  1 .060E+02 8 .331E+01 6 .468E+01 4 .963E+01 3 . 765E + 01  0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO 0 .OOOE+OO  1 .007E+00 1 .007E+00 1 .007E+00 1 .007E+00 1 .007E+00  11 13 15 17 19  08 7 .545E- 12 09 5 .302E- 12 08 -4 .840E- 12 09 2 .876E- 12 09 -2 . 762E- 13  2 .310E- 09 3 .010E- 10 1 .491E- 09 5 .516E- 10 4 .744E- 10  1 1 1 1 1  OOOE+OO OOOE+OO OOOE+OO OOOE+OO OOOE+OO  2 .827E+01 2 . 104E+01 1 .558E+01 1 . 152E + 01 8 .554E+00  0..OOOE+OO 0..OOOE+OO 0. OOOE+OO 0. OOOE+OO 0..OOOE+OO  1 .007E+00 1 .007E+00 1 .007E+00 1 .006E+00 1 .006E+00  21 23 25 27 29  10 11 11 11 11  1 1 1 1 1  OOOE+OO OOOE+OO OOOE+OO OOOE+OO OOOE+OO  6 .432E+00 4 . 930E + 00 3 . 865E + 00 3 . 103E + 00 2 . 542E + 00  0..OOOE+OO 0. OOOE+OO 0,, OOOE + OO 0. OOOE+OO 0. OOOE+OO  1 .006E+00 1 .006E+00 1 .006E+00 1 .006E+00 1 .006E+00  31 33 35 37 39  1.403E. 10 6,, 346E- 10 1 546E. 10 1 988E. 09 4. 938E- 11  1 1 1 1 9  OOOE+OO OOOE+OO OOOE+OO OOOE+OO 999E-01  2 .113E+00 1.765E+00 . 1.480E+00 . 1.262E+00 . 1.123E+00 .  0. 0. 0. 0. 0.  OOOE+OO OOOE+OO OOOE+OO OOOE+OO OOOE+OO  1 .007E+00 1 .007E+00 1 .007E+00 1 .007E+00 1 .007E+00  41 43 45 47 49  13 - 1 882E. 08 -7 ..043E- 12 1 095E, 14 5. 186E- 09 12 15 2:563E- 10 1. .108E-13 . 15 1 658E. 10 1 502E13 , 14 -2. 737E- 09 - 1,747E12  1 4. 2. 1 2.  9 9 9 9 9  999E-01 1 049E+00 . 999E-01 . 1 031E+00 . 998E-01 1 024E+00 , 998E-01 1 OOOE+OO . 998E-01 9 . 918E-01  3. 193E-05 1. 179E-03 3. 273E-03 7. 554E-04 4. 590E-05  1 .006E+00 1 .006E+00 1 .005E+00 1 .005E+00 1 .005E+00  51 53 55 57 59  -1 . 334E -2. 646E - 1 361E . - 7 . 055E - 5 . 028E  15 2. 048E- 09 1 952E. 09 2. 663E15 1 355E. 15 - 3 . 482E- 1 1-7. 406E16 -4. 028E- 11 -4. 638E. 11 - 3 . 151E16 - 1 557E-  13 13 14 14 14  1 962E. 1 298E. 3. 335E3. 858E1 491E.  10 10 12 12 12  9 9 9 9 9  998E-01 995E-01 989E-01 973E-01 932E-01  9. 965E-01 1 OOOE+OO . 1 001E+00 . 1 002E+00 . 1 002E+00 .  1 2. 2. 1 5.  796E-03 . 175E-01 474E+00 289E+01 . 838E+01  1 .003E+00 1 .002E+00 1.001E+00 . 9 .999E-01 9..992E-01  61 63 65 67 69  - 3 . 561E -2. 620E - 1 626E . -2. 199E - 5 . 512E  16 - 5 . 186E- 13 16 1 390E. 12 . 16 1 592E13 16 -2. 591E- 13 17 -7 . 460E- 16  14 14 15 16 16  4. 971E1 332E. 1 525E. 2. 482E1 227E.  14 13 14 14 16  9 9 9 8 8  824E-01 540E-01 001E-01 670E-01 609E-01  1 001E+00 . 1 001E+00 . 1 OOOE+OO . 9. 999E-01 9. 997E-01  2. 842E+02 1 531E+03 . 6 . 151E+03 1 041E+04 . 1 132E+04 .  9..986E-01 9 . 987E-01 9. 996E-01 9 . 999E-01 1 OOOE+OO .  71 73 75 77 79  6 .633E+12 6 .633E+12 6 .632E+12 6 .632E+12 6 .631E+12  1 .621E- 07 3 .044E 7 .261E- 06 -5 . 803E 1 .479E- 07 5 . 180E 1 .666E- 12 1 .811E 8 .982E- 12 -2 .675E  11 2 .895E- 07 08 3 .476E- 08 1 1 2 .601E- 08 11 -2 .214E- 09 12 -3 . 215E- 08  11 13 15 17 19  1 .000E-04 1 .585E-04* 2 .512E-04 3 .981E-04 6 .310E-04  6 .631E+12 6 .630E+12 6 .630E+12 6 .629E+12 6 .629E+12  9 .228E7 .601E1 .234E2 .874E4 .591E-  12 -4 .333E- 08- -8 .276E12 -2 .806E- 08 - 1.547E12 8 .281E- 10 -1 .371E12 2 .601E- 08 -5 .455E13 3 .411E- 08 3 .318E-  21 23 25 27 29  1 .000E-03 1 .585E-03 2 .512E-03 3 .981E-03 6 .310E-03  6 .628E+12 6 .627E+12 6 .627E+12 6 .626E+12 6 .626E+12  2 . 146E-12 -6 .656E 14 2..412E8 .219E- 14 -8 .086E 15 3 .143E1 .080E- 13 -1 . 429E 14 - 1.557E1 .422E- 13 1 .625E 13 5 .759E2 .808E- 14 5 .574E 14 -4 .953E-  31 33 35 37 39  1 .O00E-O2 1 .585E-02 2 .512E-02 3 .981E-02 6 .310E-02  6 .625E+12 6 .624E+12 6 .624E+12 6 .623E+12 6 .623E+12  3 .607E1 .391E4 .162E8 .851E3 .539E-  14 8.. 538E 14 5 .062E 14 5,.213E 14 -6 , 909E 13 -3. .939E  16 15 16 15 14  -2. .784E- 09 -3 .909E- 13 -3. .660E- 10 8 .531E- 14 -2. .709E- 10 6 .518E- 14 - 1 828E. 10 1 .989E- 13 2,.191E- 10 5 .828E- 13  2 .667E3..506E2 .594E1.751E. 2..098E-  41 43 45 47 49  l'..000E-01 1.585E-01 . 2,.512E-01 3 .981E-01 6,.310E-01  6 .622E+12 6 .622E+12 6 .621E+12 6 .621E+12 6 .620E+12  4,.872E1.801E. 4..314E6..019E3. 641E-  13 9,. 700E 12 3., 722E 12 -1 . 253E 12 -6 . 602E 12 - 1 820E .  14 13 13 13 14  - 1 464E. 09 5 .325E- 13 -6. .626E- 09 -2. .196E- 12 -1 .•613E- 09 -4. .433E- 12 2. 076E- 08 4.. 363E- 12 -5. . 141E-10 7.,217E- 12  51 53 55 57 59  1.OOOE+OO . 1.468E+00 . 2..154E+00 3 .162E+00 4. 642E+00  6 .620E+12 6 .620E+12 6 .620E+12 6,.619E+12 6.. 619E+ 12  1 957E. 13 4,,708E- '14 2. 934E- 14 2. 600E- 14 4,. 706E- 14  2.. 319E -1 . 071E -4. . 395E -2. 375E -2. 181E  61 63 65 67 69  6. 813E+0O 1 OOOE+01 . 1 445E+01 . 2. 089E+01 3. 020E+01  6 . 618E+12 6 .618E+12 6 . 617E+12 6 . 617E+12 6. 617E+12  4. 932E3. 071E1 355E. 7 .026E4. 921E-  15 15 15 16 16  71 73 75 77 79  4..365E+01 6..310E+01 8. 610E+01 9. 750E+01 9. 950E+01  6 .617E+12 6. 617E+12 6. 617E+12 6. 617E+12 6. 617E+12  3. 690E2. 921E2. 348E2. 871E1 003E.  16 16 16 16 16  -4 -9 -9 -4 -9  . 543E . 498E . 548E . 763E .629E  0 .OOOE+OO 7 .316E- 12 1 .491E- 1 1 1 .470E- 11 4 .899E- 12  1 .623E7 .160E1 .458E2 .121E3 .080E-  1 .OOOE-05 1 .585E-05 2 .512E-05 3 981E-05 6 .310E-05  12 12 11 12 12  Page 239  Convergence checks  - 1 930E. - 1 160E. -2. 265E-2. 924E1, 186E-  . 09 802E968E- 10 455E- 11 . 11 588E622E- 10  Model 07310191  SS 3500./ 2.1DO/  CN, T i O , H20 (hm)  Convective  Hconv  3 .724E-05 5 .267E-05 7 .687E-05 1 .144E-04 1 .722E-04  0 .OOOE+OO 0 .OOOE+00 0 .OOOE+00 0 .OOOE+OO 0 .OOOE+00  0 .OOOE+00 0 .OOOE+00 0 .OOOE+00 0..OOOE+OO 0 . OOOE+00  0 .OOOE+00 0 .OOOE+OO 0 .OOOE+00 0 .OOOE+OO 0 .OOOE+00  1 3 5 7 9  0 .OOOE+00 0 .OOOE+OO 0 .OOOE+00 0 .OOOE+00 0 .OOOE+OO  2 .614E-04 4 .001E-04 6 .176E-04 9 .613E-04 1 .509E-03  0 .OOOE+OO 0 .OOOE+00 0 .OOOE+00 0 .OOOE+OO 0 .OOOE+00  0 .OOOE+00 0 .OOOE+00 0 .OOOE+00 0 .OOOE+00 0 .OOOE+00  0 .OOOE+00 0 .OOOE+00 0 .OOOE+OO 0 .OOOE+00 0 .OOOE+00  1 1 13 15 17 19  -01 -01 -01 -01 -01  0..OOOE+00 0 OOOE+00 0. OOOE+00 0 OOOE+00 0 .OOOE+00  2 . 387E-03 3 .805E-03 6 . 111E-03 9 .885E-03 1 .608E-02  0 .OOOE+00 0 .OOOE+OO 0 .OOOE+00 0 OOOE+00 0 .OOOE+OO  0 .OOOE+00 0 .OOOE+00 0 .OOOE+00 0 .OOOE+00 0 .OOOE+00  0 .OOOE+00 0 .OOOE+00 0 .OOOE+00 0 .OOOE+OO 0 .OOOE+00  21 23 25 27 29  5 .562E-02 5 .830E-02 6 .225E-02 6. 807E-02 7 .646E-02  1 . 714E-01 1 . 743E -01 1 . 780E -01 1 .831E -01 1 . 902E -01  0 OOOE+00 0. OOOE+OO 0,.OOOE+00 0 OOOE+OO 0. OOOE+00  2 .628E-02 4 . 307E-02 7 .076E-02 1 . 166E-01 1 . 931E-01  0 .OOOE+00 0..OOOE+OO 0 OOOE+00 0. OOOE+00 0. OOOE+00  0 .OOOE+00 0 .OOOE+00 0 .OOOE+00 0. OOOE+00 0. OOOE+OO  0 .OOOE+00 0 .OOOE+OO 0 .OOOE+00 0 OOOE+OO 0..OOOE+00  31 33 35 37 39  1 . 218E+00 1 .191E+00 1 .158E+00 1 .120E+00 1 .081E+00  8. 829E-02 1 055E-01 . 1.332E-01 . 1 826E-01 . 2.. 702E-01  2 .000E -01 2 . 136E-01 2 . 324E -01 2 .585E -01 2 .933E -01  0. 0. 0. 0. 0.  OOOE+00 OOOE+00 OOOE+00 OOOE+00 OOOE+00  3 .221E-01 5 .428E-01 9 .282E-01 1 .633E+00 3 .024E+00  0..OOOE+00 0..OOOE+OO 0..OOOE+00 0. OOOE+OO 0. OOOE+00  0..OOOE+00 0, OOOE+OO 0, OOOE+00 0. OOOE+00 0. OOOE+00  0,, OOOE + 00 0. OOOE+00 0,, OOOE + OO 0,, OOOE + 00 0,, OOOE+00  41 43 45 47 49  2 .677E+09 2 .812E+09 2 .982E+09 3 .185E+09 3 .419E+09  1 .048E+00 1 .029E+00 1 .017E+00 1 .009E+00 1 .004E+00  4..157E-01 5.. 716E-01 6 ,. 379E-01 5. 440E-01 4. 504E-01  3 . 323E -01 3 .611E -01 3 .810E -01 3 .912E -01 3 .945E -01  4. 5. 6. 5. 4.  147E-01 604E-01 167E-01 366E-01 494E-01  5 .859E+00 9 .803E+00 1. .378E + 01 1.550E+01 . 1 . 715E + 01  9. 527E+03 3.. 210E+04 4 . 542E+04 2. 771E+04 1 065E+04 .  2..162E+04 7 .984E+05 . 2..217E+06 5..116E+05 3.. 109E + 04  3.. 193E-05 1.. 179E-03 3.. 273E-03 7 . 554E-04 4 ., 590E-05  51 53 55 57 59  3 .692E+09 4 .010E+09 4 .360E+09 4 .760E+09 5 .225E+09  1 .003E+00 1 .006E+00 1 .015E+00 1..040E+00 1 .102E+00  5. 244E-01 8. 786E-01 1.589E+00 . 2. 895E+00 5. 305E+00  3 .916E -01 3 . 796E -01 3 .517E -01 2 .995E -01 2 . 287E -01  5. 6. 6. 4. 3.  141E-01 580E-01 154E-01 718E-01 172E-01  2 .677E+01 6 . 179E+01 1 .556E+02 3 .989E+02 1 031E+03 .  3. 500E+04 1 688E+05 . 3. 714E+05 6.. 197E + 05 9 . 591E+05  1.217E+06 . 1 473E+08 . 1 677E+09 . 8. 737E+09 3. 956E+10  1. .796E-03 2.. 175E-01 2. 474E+00 1. .289E + 01 5. 838E+01  61 63 65 67 69  5 .794E+09 6 .557E+09 7 .487E+09 7 .977E+09 8 .065E+09  1.245E+00 .  9. 868E+00 1 867E+01 . 3. 106E+01 3. 766E+01 3. 887E+01  1 625E . -01 1. .18 IE -01 9 . 775E -02 9 . 300E -02 9 .237E . -02  1 997E-01 . 1 319E-01 . 1 037E-01 . 9. 740E-02 9. 658E-02  1 926E+11 . 1 037E+12 . 4. 169E+12 7 .053E+12 7 .673E+12  2. 842E+02 1 .531E+03 6. 151E+03 1 041E+04 . 1 .132E+04  71 73 75 77 79  DELrad  DELad  DELbub  1 .757E+09 1 . 763E+09 1 .771E+09 1 .782E+09 1 .795E+09  1 .191E+00 1 .204E+00 1 .218E+00 1 .232E+00 1 .244E+00  2 .954E-02 3 .081E-02 3 . 264E-02 3 .506E-02 3 .770E-02  1 . 790E -01 1 . 748E -01 1 . 707E -01 '1 .670E -01 1 .642E -01  0 .OOOE+00 0 .OOOE+00 0 .OOOE+OO 0 .OOOE+00 0 .OOOE+00  1.811E+09  1 .255E+00 1 . 263E + 00 1 . 269E+00 1 .274E+00 1 .278E+00  4 .018E-02 4 . 232E-02 4 .414E-02 4 .571E-02 4 . 713E-02  1 .623E -01 1 . 612E -01 1 .608E -01 1 .609E -01 1 .615E -01  1 ..258E+00 1 .261E+00 . 1 ..264E+00 1. .267E + 00 1; 271E+00  1 .926E+09 1 .954E+09 1 .983E+09 2 .013E+09 2 .044E+09  1 .280E+00 1 . 281E+00 1 . 282E+00 1 .281E+00 1. 280E+00  4 .845E-02 4 .971E-02 5 .095E-02 5 .223E-02 5 . 372E-02  1 .625E 1 .637E 1 .652E 1 .670E 1 .690E  1.. 275E+00 1. 279E+00 285E+0O 1. 293E+00 1. 303E+00 1. 1. 316E+00 1. 335E+00 363E+00 1. 403E+00 1. 46OE+00 1. 1. 530E+00 586E+00 1. 627E+00 1. 649E+00 1. 655E+00 1. 1. 647E+0O 1. 618E+00 555E+00 1. 451E+00 1. 335E+00 1. 1. 250E+00 1. 215E+00 1. 223E+00 1. 234E+00 1. 236E+00  2 .075E+09 2 .108E+09 2 .142E+09 2 .178E+09 2 .218E+09  1 .278E+00 1 .273E+00 .266E+00 1.256E+00 1 1 . 240E+00  2 . 264E + 09 2 . 316E + 09 2 .377E+09 2 .453E+09 2 .550E+09  Cp/Cv  3 5 7 9  1 .000E-05 1.585E-05 2 .512E-05 3 .981E-05 6 . 310E-05  1 . 275E+00 1 .269E+00 1 .264E+00 1.259E+00 . 1.256E+00 .  11 13 15 17 19  1 .000E-04 1 .585E-04 2 .512E-04 3 .981E-04 6 .310E-04  1.254E+00 . 1.253E+00 . 1,253E+00 . 1 .254E+00 . 1.256E+00 .  21 23 25 27 29  1 .000E-03 1 .585E-03 2 .512E-03 3 .981E-03 6 .310E-03  31 33 35 37 39  1 .000E-02 1 .585E-02 2 .512E-02 3 .981E-02 6 .310E-02  41 43 45 47 49  1.OOOE-01 . 1.585E-01 . 2 .512E-01 3..981E-01 6..310E-01  51 53 55 57 59  1.OOOE+00 , 1,468E+00 . 2,.154E+00 3..162E+00 4.,642E+00  61 63 65 67 69  6 .813E+00 1 OOOE+01 . 1 445E+01 . 2. 089E+01 3. 020E+01  71 73 75 77 79  4. 365E+01 6. 310E+01 8. 610E+01 9 . 750E+01 9 . 950E+01  Page 240  quantities Vconv  Qconv  TauRM  1  -3.00  H_P  1 .830E+09 1 .851E+09 1 .874E+09 1 .900E+09  1 .540E+00 1.961E+00 . 2..172E+00 2 . 208E + 00  TauRMb  . 2.. 701E + 03 1 495E+06 7 ., 112E + 03 2. 449E+06 3. 816E+06 1 551E+04 . 2. 068E+04 4. 566E+06 2 . 163E + 04 4. 702E+06  Hconv/H  •fto  

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0085580/manifest

Comment

Related Items