Compact Riemann Surfaces Prime Galois Coverings of P 1 by Michael Tsiang B.A., University of California, Santa Cruz, 2004 A THESIS SUBMITTED IN PARTIAL FULFILMENT OF T H E REQUIREMENTS FOR T H E D E G R E E OF Master of Science The Faculty of Graduate Studies (Mathematics) The University of British Columbia June, 2007 © Michael Tsiang 2007 Abstract The uniqueness of the hyperelliptic involution is well known in the theory of Riemann surfaces. More precisely, we know that if X is a hyperelliptic compact Riemann surface, there is a unique automorphism r of order 2 such that X/(T) = P 1 . We wish to generalize the situation slightly. We say X is a prime Galois covering o / F 1 if there exists an automorphism r of (odd) prime order p such that X/{T) = P 1 . This leads us to ask the question: When is this automorphism r unique? We begin by building the necessary background to understand prime Galois coverings of P 1 . We then prove a theorem due to Gonzalez-Diez that answers our question about uniqueness. The proof given here follows his proof (given in [G-D]) quite closely, though we elaborate and modify certain details to make it more self contained. ii Table of Contents Abstract ii Table of Contents iii Acknowledgements iv 1 Background 1 1.1 Compact Riemann Surfaces 1 1.1.1 Meromorphic Functions 3 1.1.2 Meromorphic Differentials 4 1.2 Projective Curves 5 1.3 Normalization 8 1.4 Divisors 9 1.4.1 Special Divisors 12 1.5 The Riemann-Hurwitz Formula 13 1.6 Group Actions 14 1.6.1 Galois Coverings 15 2 Prime Galois Coverings of P 1 17 2.1 Examples 17 2.2 Preliminaries 19 2.2.1 A Special Case 21 3 The Main Theorem 24 3.1 Two Lemmas 24 3.2 The Crucial Step 25 3.3 Conclusion 27 Bibliography 28 iii Acknowledgements I would like to thank my advisor Zinovy Reichstein for his guidance and patience throughout my time at UBC. I also acknowledge the help of all of the staff in the department office, especially Lee Yupitun for being all knowing. I feel exceptionally fortunate and grateful to have the unending support of my parents, my brother David, and my sister Angela. Moreover, I thank my fellow graduate students for their friendship and help during my grad-uate career. In particular, I thank Christy Cameron, Warren Code, Chris Coulter, Alex Duncan, Fiorina Halasan, Patrick Ingram, Vishaal Kapoor, Simon Rose, Patrick Walls, Erick Wong, and Katya Yurasovskaya. Lastly, I give a special thanks to Amy Goldlist. This paper, as well as most of my graduate work, would not have been possible without her friendship, support, and copy-editing skills. iv Chapter 1 B a c k g r o u n d Compact Riemann surfaces are of interest in different areas of mathemat-ics. There are equivalences which connect the complex analytic objects to objects arising in pure algebra and in algebraic geometry. The main purpose of this paper is to analyze the group actions of com-pact Riemann surfaces which admit a special type of automorphism. Namely, we will consider p-sheeted coverings of the Riemann sphere, where the cov-ering map is induced by an automorphism of prime order p. We will give references rather than proofs for most of the background results which are standard in the literature. The first two sections build the parallel languages of compact Riemann surfaces and projective curves. The third section then links the two together via a process called normalization. Given this link, we are able to apply techniques from either category to prove interesting results in both. Using standard notation, we let C denote the complex plane, and let P 1 = C U {00} denote the Riemann sphere. 1.1 Compact Riemann Surfaces Definition. A compact Riemann surface is a compact connected Hausdorff topological space X together with an open covering {Ui} of X and a family of mappings zi : Ui —> C such that (i) each Zi : Ui —» C is a homeomorphism of Ui onto an open subset Zi(Ui) C C; and (ii) if Ui f l Uj 7^ 0 , then the function Zj o z~l : Zi(Ui n Uj) -> Zj(Ui n Uj) 1 Chapter 1. Background is biholomorphic. We call such a (Ui,Zi) a local (holomorphic) coordinate, and {(Ui,Zi)} a holomorphic coordinate covering. By abuse of notation, we will often call z a local coordinate instead of the pair (U, z) when it is unnecessary to specify the set U. From the classification of two-dimensional (real) manifolds, any compact orientable connected two-dimensional manifold is homeomorphic to a sphere with handles. The number of handles, g, is called the genus, a topological invariant. Thus every compact Riemann surface, being a compact orientable connected one-dimensional complex manifold, has an associated genus. Def in i t i on . Let X and Y be compact Riemann surfaces with {(Ui,Zi)} and {{Va,wa)} be their respective holomorphic coordinate coverings. A holomorphic mapping f : X - + Y is a family of continuous mappings fi-Ui^ Y, such that (i) ft = fj on Ui n Uj for Ut H Uj ^ 0; and (ii) wa o fi o z~x is a holomorphic function on f~l(Va) H Ui whenever f-\va)nul^0. The degree of / , denoted deg/, is the number of points in the inverse image of any point in Y, counting multiplicity. A holomorphic mapping / : X —> Y of degree d is sometimes called a degree d covering map, and X is called a d-sheeted covering (or cover) of Y. The following proposition makes clear why this terminology makes sense. P r o p o s i t i o n 1. [M] Let X and Y be compact Riemann surfaces. If f : X —+ Y is a nonconstant holomorphic mapping, then f is surjective. Def in i t i on . An isomorphism between compact Riemann surfaces X and Y is a holomorphic mapping / : X —> Y which is bijective and whose inverse f~l : Y —-> X is holomorphic. If there exists an isomorphism between X 2 Chapter 1. Background and Y, we say that X and Y are isomorphic, and we write X = F , Definition. An isomorphism / : X —> X is called an automorphism of X . The set of automorphisms of X forms a group denoted by Aut(X). The general compact Riemann surface admits only the trivial automor-phism. However, when Aut(X) is nontrivial, we obtain a (nontrivial) group action on X; we will elaborate on group actions in later sections. A result we will apply implicitly is that Aut(X) is a finite group for every compact Riemann surface X of genus g > 2. This can be proven by considering a finite set of special points on X called Weierstrass points which every automorphism must permute. The famous bound |Aut(X)| < 84(g — 1) is Hurwitz's Theorem. A detailed proof can be found in [Fa]. 1.1.1 Meromorphic Functions Definition. A holomorphic mapping / : X —» P 1 is called a meromorphic function on X, or simply a function on X. The field of meromorphic func-tions on X, also called the function field of X, is denoted by C(X). Let / : X —> Y be a degree d covering map. For g € C(Y) we define /*(<?)= S ° / by the following commutative diagram: p i Since g is a meromorphic function on Y, then the diagram shows that g o f is a meromorphic function on X. We now have a map f':C(Y) —* C(X) 9 '—> f*{g) = g°f, and /* is called the pullback of / . We can easily check that /* is a ring homomorphism and is in fact a C-algebra homomorphism. Since / is surjective, then / induces an injective pullback /*. This shows that C(X) is a finite algebraic field extension over C(Y) of degree d. 3 Chapter 1. Background The field of meromorphic functions on P 1 is isomorphic to the function field in one variable x over C. In other words, C(PX) S C ( i ) . Thus, if / : X —> P 1 is a meromorphic function on X of degree d, then C(X) is a finite algebraic field extension over C(x) of degree d. As every finite extension over a field of characteristic zero is separable, then there is an algebraic element y such that C(X) = C(x,y), and the minimal polynomial of y over C(x) has degree d. Definition. Let X be a compact Riemann surface, P e l , and / a nonzero meromorphic function on X. By choosing a local coordinate z in a neigh-bourhood of P such that z(P) — 0, we have that in a neighbourhood of P, f = zvh(z), v G Z, where is a holomorphic function satisfying /i(0) ^ 0. For any local coordinate z such that z(P) — 0, the value of v is the same and uniquely determined by / . This v is called the order or multiplicity of / at P, de-noted Vp(f). When vp(f) > 0, p is called a zero of / and vp(f) is the order of the zero P. When vp(f) < 0, P is called a po/e of / and |fp(/)| is the order of the pole P. The ring Ox,p — {/ 6 : vp{f) > 0} is an example of a discrete valuation ring of C(X) . Using discrete valuation rings, we can construct a categorical equivalence between finite extensions of fields of transcendence degree one over C and compact Riemann surfaces. We will not explicitly need this equivalence here, and so the reader is referred to [Fu]. 1.1.2 M e r o m o r p h i c D i f f e r e n t i a l s Definition. A meromorphic (respectively, holomorphic) differential on an open set V C C is an expression u> of the form ijj — f{z)dz where / is a meromorphic (resp. holomorphic) function on V. We say that OJ is a meromorphic (resp. holomorphic) differential in the coordinate z. A meromorphic (resp. holomorphic) differential UJ on a Riemann surface X is a family of meromorphic (resp. holomorphic) differentials {WJ} such that 4 Chapter 1. Background (i) if {(U~i,Zi)} is a holomorphic coordinate covering of X, then = fi(zi)dzi, where fi is a meromorphic (resp. holomorphic) function on Zi(U~i) C C; and (ii) if Zi = tpij(zj) is the coordinate transformation on Ui fl Uj ( / 0), then the local representation of the differential changes by fi(<Pij(zj))d<Pij(zj) = fj(zj)dzj, which is the chain rule. The set of meromorphic differentials on X is denoted by Q1(X). Definition. Let X be a compact Riemann surface with {(Ui, Zi)} a holomorphic coordinate covering of X and u> meromorphic differential on X. If P e Ui n Uj, then vpUi) = "P { ^ ^ ) ) d ^ ~ ) = ^ ( / i ) -This means we can define vP(w) = vp(U), P £ Ui. If vp{w) > 0, then P is called a zero of u>. If vp(uj) < 0, then P is called a pole of u>. 1.2 Projective Curves Definition. Let V be a finite dimensional vector space over C. Consider the equivalence relation ~ on V\{0}:. u ~ v if and only if there exists a A G C* with u = Xv. The projective space associated to V is defined by p(v) := y\{o}/ ~ • The dimension of P{V) is defined by dimP(y) := dimV - 1. Geometrically, two vectors are equivalent (under ~) if and only if they span the same line in V. Thus the projective space associated to V is the set of all lines through the origin (one-dimensional subspaces) in V. 5 Chapter 1. Background Examples, (a) The projective line is defined to be P(C 2 ), the projective space associated to C 2 . We denote the equivalence class map by n : C 2\{(0,0)} -^P(C 2 ) . We write (x:y) :=TT((x,y))eF(C2), and call (x : y) the homogeneous coordinates of the point Tr((x,y)). By defining the map P(C 2) —> P ^ C u i o o } (x : y) i—> < y { oo if y = 0 we see that P(C 2) is equivalent to C U {oo}, which justifies the notation P 1 . The point (1 : 0) is called the point at infinity. (b) The projective plane P 2 is defined to be the projective space associated to C 3 . That is, P 2 = P(C 3). As in the previous example, (x : y : z) are the homogeneous coordinates of the point it((x,y,z)) & P 2 . The set Loo = {(x:y.z)eF2:z = 0} is called the line at infinity. Definition. A projective plane curve X C P 2 is defined as X = {(x:y.z):F{x,y,z) = 0}, for some nonconstant homogeneous polynomial F in three variables over C. The degree of X is the degree of the defining homogeneous polynomial. If we restrict to C 2 = P 2 \ L 0 O , then X satisfies the affine equation f(x,y) = 0, where f(x,y) = F(x,y, 1). 6 Chapter 1. Background A projective plane curve is irreducible if its defining homogeneous polyno-mial is irreducible. Definition. Let X be a projective plane curve. A rational function f : X —> P 1 on X is a quotient of homogeneous polynomials of the same degree: h(x,y,z) with degg = degh, and h is not identically zero on X. The field of rational functions, also called the function field of X, is denoted by C(X) . As the notation suggests, it transpires that the function field of a projective plane curve and that of a compact Riemann surface are identical. Definition. Let X be a projective plane curve defined by the homoge-neous polynomial F(x,y,z). We say a point P G X is singular (or P is a singularity of X) if A smooth point on X is one that is not singular. If every point on X is smooth), then X is a smooth or nonsingular curve. Any projective curve has at most finitely many singular points (see [Hu]). Definition. Let X be a projective plane curve defined by the homoge-neous polynomial F(x,y, z). We call a point P G X a double point of X if all of the first partials of F vanish at P but not all of the second partials. At these points, X has two tangent lines at P, either intersecting or one tangent counted with multiplicity. If the tangent lines are distinct, then P is called an ordinary double point of X. We can similarly define triple points and, even more generally, fc-tuple points, but we will not need them for our purposes. Ordinary double points are the simplest and most "well behaved" singularities. Let X and Y be projective plane curves of degrees d and e, respectively. The intersection number of X and Y, denoted X -Y, is the number of points in X fl Y, counting multiplicity. If X and Y do not share a common com-ponent - that is the defining homogeneous polynomials of X and Y do not 7 Chapter 1. Background have a common irreducible factor - then X • Y — de. This is known as Bezout's Theorem (see [Hu]). For a more general definition of intersection number, see [Ha, Chapter V]. 1.3 Normalization There is a startling and beautiful connection between projective plane curves and Riemann surfaces. An important result is that every smooth projective plane curve is a compact Riemann surface. In this section, we will see that every irreducible projective plane curve has a compact Riemann surface as-sociated to it, and thus we establish a categorical equivalence. An extensive treatment of normalization can be found in [Gr]. . Definition. Suppose X is an irreducible projective plane curve, and S its set of singular points. If there exists a compact Riemann surface X' and a holomorphic mapping o:X' —> P 2 , such that (i) o(X') = X; (ii) cr _ 1(5) is a finite set; and (iii) o : X'\o-l(S) -* X\S is injective, then we call (X', o) the normalization of X. Note that if X is a smooth projective plane curve, then X itself along with the identity mapping is a normalization of X. If (X',o) is a normalization of X, we often (by abuse of terminology) say that X' is a normalization of X, leaving the map o implicit. Theorem 1. [Gr] For any irreducible projective plane curve X C P 2 , there exists a normalization of X. Moreover, the normalization is unique: if (X',o) and (X,d) are normalizations of X, then there exists an isomor-phism T:X' —> X 8 Chapter 1. Background such that the diagram X'- 1 "X X commutes. Theorem 2. [Gr] Any compact Riemann surface X' can be obtained through the normalization of a certain projective plane curve X with at most ordinary double points. That is, there exists a holomorphic mapping o : X' —* P 2 such that cr(X') is a projective curve possessing at most ordinary double points. With these two theorems working in the background, we will let X be a compact Riemann surface or, equivalently, a smooth projective curve for the remainder of the paper. Given this we can always use a homogeneous polynomial to define our compact Riemann surfaces. Nevertheless, for ease of computation and notation, we often write the algebraic equation in affine coordinates, even if the curve defined this way is singular. The curve we are actually interested in is often the normalization of the curve we actually write down! We simply verify (when necessary) that the singular points in the affine model do not affect our computations after we normalize the curve. Two projective plane curves X and Y are birationally isomorphic if their respective normalizations X' and Y' are isomorphic. This implies that bira-tionally isomorphic curves have isomorphic function fields (C(X) = C(Y)), and there is a unique nonsingular curve (in the sense that normalizations are unique), with an isomorphic function field. It is this nonsingular model with which we work. 1.4 Divisors Divisors are a way of encoding information about zeros and poles of func-tions and differentials. The beautiful and powerful Riemann-Roch Theorem is a statement which allows us to determine how many linearly independent functions (or differentials) there are on a compact Riemann surface with prescribed zeros and poles. 9 Chapter 1. Background Definition. A divisor D of X is a formal finite sum n D = miP\ H 1- mnPn = ^ rriiPi, i=l where m* G Z and Pi G X , i = 1,..., n. The degree of I? is defined as n deg-D = y^mj. i=i The support of D is the set Supp(D) of points in X which have nonzero coefficients in D. The set of all divisors of X forms an abelian group Div(X) under the obvious addition of divisors. There is a partial ordering on Div(X). For divisors D\ = J2npP a n d D2 = "Y^mpP, we write D\ > D2 if and only if np > mp for all P G X. A divisor D is effective if D > 0. Definition. Let / be a nonzero meromorphic function on X. The divi-sor of / , denoted (/), is the divisor of the form (/) = E Pex Divisors of this form are called principal divisors. Definition. Two divisors A and D are said to be linearly equivalent if their difference is a principal divisor, that is, A- D = (f) for some / G C(X). This equivalence relation is denoted by A ~ D. Since every nonzero meromorphic function has the same number of zeros as poles, we have that every principal divisor has degree 0. If two divisors are linearly equivalent, then they have the same degree. On P 1 , the converse is also true. That is, every divisor on P 1 of degree 0 is a principal divisor. 10 Chapter 1. Background Definition. Let w be a nonzero meromorphic differential on X. The divisor of UJ, denoted (UJ), is the divisor of the form (UJ) = £ vp(w)P. Pex Divisors of this form are called canonical divisors. All canonical divisors on a compact Riemann surface X axe linearly equivalent, so we are often able to consider the canonical divisor class K := (UJ) € Div(X) / ~ without specifying the differential UJ. Abusing notation, we often say K is a canonical divisor, though it is in fact an equivalence class. Definition. For a divisor D on X, define L(D) = {0^f€ C(X) : (/) > -D} U {0}. This is a vector space over C. We denote the dimension by 1(D) := dimL(D). Definition. The complete linear series of D is defined as |D| = {E e Div(X) : E ~ D and E > 0}. It is not difficult to see that there is a natural bijection between the complete linear series |£)| and the projective space P(L(D)). Definition. For a divisor D on X, define fl(D) = {uje Q.\X) : (UJ) >D}U {0}. This is also a vector space over C, and we denote the dimension by i(D) := dim0(D). Proposition 2. [Fu] Let K be a canonical divisor on X. Then for any divisor D on X, i(D) = £(K - D). 11 Chapter 1. Background Theorem (Riemann-Roch). [Fu] If X is a compact Riemann surface of genus g and D is a divisor of degree d on X, then £(D) - £(K - D) = d+ 1 -g. Corollary 1. If X is a compact Riemann surface of genus g, then there are g linearly independent holomorphic differentials on X. That is, if K is a canonical divisor, then £(K) = g. Corollary 2. Let K be any canonical divisor on a compact Riemann surface of genus g. Then degK — 2g — 2. 1.4.1 S p e c i a l D i v i s o r s As an application of the Riemann-Roch Theorem, we consider special divi-sors and prove Clifford's Theorem. Definition. An effective divisor D on X is a special divisor if 1(D) > 1 and £(K - D) > 1. Remark 1. Note that if D is an effective divisor of degree d < g — 1, then the Riemann-Roch Theorem gives 1(D) - £(K - D) = d + 1 - g < (g - 1) + 1 - g = 0, so 1 < (.(D) < £(K — D). This shows that any effective divisor of degree d < g — 1 is a special divisor. In particular, when d = g — 1, we have £(D) = l(K - D). Proposition 3. An effective divisor D on X is a special divisor if and only if there exists an effective divisor E such that D + E is a canonical divisor on X. The divisor E is called a complementary divisor of D. Proof. Let D be a special divisor on X. Consider the complete linear series \K - D\ = {E e Div(X) : E ~ K - D and E > 0}. Since \K — D\ is in one-to-one correspondence with P(L(K — D)), then dim |K - D\ = £(K - D) - 1 > 0. In particular, \K — D\ is nonempty, and hence there exists E € Div(X) with E > 0 such that E ~ K - D or, equivalently, D + E ~ K. Conversely, if there exists an effective divisor E such that D 4- E = K is canonical, then £(K - D) = £(E) > 1. • 12 Chapter 1. Background Theorem (Clifford). If D is a special divisor on a curve of genus g > 2, then £(D) < i d e g D + 1. Proof. Let K be a canonical divisor, which we can choose to be effective (since g > 2). We may assume £(D — P) ^ €(Z?) for every P G X , since otherwise we can work with D — P and get a better estimate. By Proposi-tion 3, there is an effective divisor E such that D + E = K, so in particular E = K — D > 0. Since D and K — D are effective, then L(0) is a subspace of L(K — D), and L(D) is a subspace of L(K). Choose h G L(D) such that ft. ^ L(D — P) for any P € S u p p ^ — £>). Consider the linear transformation T : L(K - D)/L(0) = L(K)/L(D) 7 = A-By our choice of h, we see that T is injective, so dim(L(K - D)/L(0)) < dim(L(K)/L(D)), or equivalently, i(K - D) - 1 < g - £(D). Thus £(D)+£(K -D)<g+l. By the Riemann-Roch Theorem, £(D) - £(K - D) = degD + 1 - g. Adding the two expressions gives 2(.(D) < degD + 2, which yields the result. • 1.5 The Riemann-Hurwitz Formula The main tool from the background is the Riemann-Hurwitz Formula. We will apply this repeatedly in our proofs. Definition. Let / : X —* Y be a nonconstant holomorphic mapping with P G X. If z is a local coordinate in a neighbourhood of f(P) such that 13 Chapter 1. Background z(f(P)) = 0, then g := zo f : X —> P 1 is a meromorphic function on X. We define the multiplicity of / at P, denoted vp(f), as This is independent of the choice of local coordinate. When Y = P 1 , this is equivalent to the notation defined earlier for mero-morphic functions. Definition. Let / : X —> Y be a nonconstant holomorphic mapping. A point P e l i s called a ramification point of / if vp(f) > 1. A point Q 6 Y is called a branch point of / if it is the image of a ramification point of / . We say / is unramified if / has no ramification points. For any nonconstant holomorphic mapping / , the set of ramification points is finite. Consider Note that for P € X such that vp(f) = 1, the coefficient of P in R is zero. Thus we can consider R as a sum over only the ramification points of / , which implies that is a finite sum. Hence R S Div(X). We call R the ramification divisor of f. Theorem (Riemann-Hurwitz Formula). [Gr] Let X be a compact Rie-mann surface of genus gx, and Y be a compact Riemann surface of genus gy. Let f : X —* Y be a nonconstant holomorphic mapping with deg/ = n, and let R denote the ramification divisor of f. Then 2gx-2 = (2gY-2)n + degR. 1.6 Group Actions Definiton. Let i i be a finite group and X a compact Riemann surface. An action of H on X is a map H x X —* X, which we write as (h, P) H-> hP, such that (i) (hxh2)P = hi(h2P) for huh2 £ H and P e X, and (ii) IP = P for P e X, where 1 € i i is the identity element. 14 Chapter 1. Background The orbit of a point P G X is the set Orb(P) = {hP:he H}. The stabilizer of P is the subgroup Stab(P) = {h e H : hP = P}. The kernel of the action of on X is the subgroup K = {h G H : hP = P for all P G X } . If K is trivial, then the action is called faithful. By fixing h £ H, the map which sends P to hP is a bijection. If for every h G H, this bijection is a holomorphic mapping from X to itself, then H is a group of automorphisms acting on X. The quotient space X/H is the set of orbits. We have nice structure when H is a group of automorphisms of X, as the following results show. Theorem 3. [M] Let a finite group H < Aut(X) act faithfully on a compact Riemann surface X. Then X/H is a compact Riemann surface. In addi-tion, the map IT which sends P —> Orb(P) is a holomorphic mapping from X —* X/H of degree \H\, and vp(ir) — |Stab(P)| for any point P G X. Proposition 4. [M] Let a finite group H < Aut(X) act faithfully on X. Then Stab(P) is cyclic for every P G X. 1.6.1 G a l o i s C o v e r i n g s We finally arrive at defining our,main objects of study, prime Galois cover-ings. Though we define the coverings over general compact Riemann sur-faces, we will restrict to the case where X is a prime Galois covering of P 1 throughout the rest of the paper. Definition. A compact Riemann surface X is a Galois covering of Y if there exists an automorphism r G Aut(X) such that Xj(r) = Y. If r has order d in Aut(X), then the Galois covering is d-sheeted. When d is prime, then we say X is a prime Galois covering of Y (and we use p instead of d to denote prime). 15 Chapter 1. Background The terminology "d-sheeted Galois covering" makes sense. Indeed, the natural projection ix : X —> X/(r) is a holomorphic mapping of degree d. Furthermore, if X is a Galois covering of Y, then the function field C(X) is a Galois extension of C(Y). We see immediately that the ramification points of n are precisely the fixed points of T . A natural question which arises when studying Galois coverings is whether the automorphism r unique. That is, if there exists o £ Aut(X) with |cr| = | T | such that X/(cr) = X/(T), is a = r? In this section, we will see that when | T | = 2 and X/{r) = P 1 , then our question is answered in the affirmative for every such X. It is this special case that we use as motivation for studying prime Galois coverings of P 1 for general (odd) primes. Definition. A compact Riemann surface X of genus g > 2 is hyperelliptic if it is a 2-sheeted Galois covering of P 1 . Thus there exists an automorphism r € Aut(X) of order 2 such that X/(T) = P 1 . The automorphism r is called the hyperelliptic involution. A quick application of the Riemann-Hurwitz Formula shows that the hyperelliptic involution has 2g + 2 fixed points. Moreover, the Riemann-Hurwitz Formula also implies that any involution on X which has 2g + 2 fixed points must yield a quotient of P 1 . The following theorem says there actually are no other such involutions. Theorem 4. [Fa] The hyperelliptic involution T on a (hyperelliptic) compact Riemann surface X of genus g > 2 is the unique involution on X with 2g + 2 fixed points. Through suitable change of coordinates, any hyperelliptic compact Rie-mann surface can be described by (the normalization of) the affine equation y2 = f(x), where the hyperelliptic involution sends the point (x, y) to (x, —y). More generally, a prime Galois covering of P 1 admits the affine equation yp = f(x), and the automorphism r £ Aut(X) of prime order p sends (a;, y) to (x, (py) for a primitive p-th root of unity ( p. The natural projection X —> X/(T) sends (x, y) to x. 16 Chapter 2 Prime Galois Coverings of P The appeal of hyperelliptic compact Riemann surfaces is that the groups of automorphisms are (in theory) easy to compute for these surfaces. Indeed, if X is a hyperelliptic compact Riemann surface and r is the hyperelliptic involution, then we have a central extension 1 — • (r) —> A u t p f ) —>H —> 1, where H = Aut(X) / (T) is a finite group of automorphisms acting on P 1 . The finite automorphism groups which act on P 1 have been completely classified (see [Sh]): they are cyclic, dihedral, A4, S 4 , and A$. In a similar way, we wish to classify all groups which act as automor-phisms on prime Galois coverings of P 1 . Ideally, we hope for a unique auto-morphism r of order p such that X/(T) = P 1 . However, we will see examples of curves which have two distinct automorphisms of prime order which quo-tient to yield P 1 . Fortunately our automorphism r is unique once the genus of X is sufficiently large. Moreover, when two distinct automorphisms exist, we can classify the curve up to isomorphism. 2.1 Examples Example 1. Let Cp be the Fermat curve defined by the equation yp = xp - 1. The groups generated by Ti(x,y) = (x,(py) and T2(x,y) — ((px,y) are distinct, but we see that T\ arid T2 are conjugate via the automorphism o~(x,y) = (y,x). Note there is another automorphism of order p given by T(X, y) — (£px, (Py)-The group (r) is also conjugate to (TI ) and ( T 2 ) , since ip o r o <p_1 = T ^ - 1 , where <p(x,y) = (1/x, —y/x). Example 2. Define the curves Dp by the equation yp = (xp - l)(xp - Xp)p~\ 17 Chapter 2. Prime Galois Coverings of P 1 where A £ C, A p / 1. As in the previous example, let rj (x, y) = (x, (py) and T2(x,y) = ((px,y). The automorphism 'xp - \ p xp-x{\ - \ p ) \ y conjugates fa) and fa) since cr o T\ O c r - 1 = T j - 1 -Example 3. Let X be the curve defined by the equation yp = (xp- l)(xp-Xp)k, for A p ^ l,fc < p — 1. In this case, T\(x,y) = (x,(py) has 2p fixed points, which are given by 0), (CpA, 0), i = Meanwhile, r2(x,y) = (( px, y) has p fixed points, given by (0, CpA), i = 1,... ,p, which implies n and T2 are not conjugate. Note that the affine equation for X has a singularity at oo. To show that oo is not fixed by T2, we consider the natural projection X —> X/(T) = P 1 that sends (x,y) to x. By taking the parameter x = j, where t is a local parameter near 0, we see that X has p points oo;, i — 1,..., p, above oo G P 1 explicitly given by the p roots t ^ / ( l -t)p(l - \ptp)k y ~ 7T+fc ' i — 1,... ,p. These points are not fixed by the action of T2. Thus we are able to work with the normalization of X and our analysis remains the same. By construction X/fa) = P 1 , so the Riemann-Hurwitz Formula tells us that X/fa) ^ P 1 since the number of fixed points of T2 is different from that of T\. Remark 2. Suppose X has an algebraic equation yP = ( x - a 1 ) d l ( x - a 2 ) d 2 - - - ( x - a „ ) d " , where 1 < di < p — 1. Consider the parameter ^ = y s ( x - a 1 ) t l ( x - a 2 ) t 2 - - - ( x - a n ) t " for some choice of s,U ^ 0 (mod p), i — 1,... ,n. One easily checks that C(x, z) = C(x,y) and so z is a valid change of variables. Then ZP = y S P { x _ a i ) p t l { x _ a 2 ) P t 2 ^ , { x _ a n ) P t n = (x - a!) s d l • • • (x - an)sd-(x - a i ) p t l • • • (x - an)pt» = ( x - a i ) s d l + p t l • • • ( x - a „ ) s d " + p i " . 18 Chapter 2. Prime Galois Coverings of P In particular, since gcd(di,p) = 1, we can choose s and t\ such that sd\ + pt\ — 1. Once we fix s, we then choose the U such that 1 < sdi + Ph < p — 1, for i = 2,..., n. This computation shows that we can always assume one branch point has multiplicity 1. Moreover, we see that if di = 0 (mod p), then n n n sdi + PU = ^] sdi — s E / di = 0 (mod p). 2.2 Pre l iminar ies The argument used to prove the following lemma is a slight generalization of the proof of [P, Lemma 5]. Lemma 1. Let X,X\,X2 be smooth curves of genus g,g\,g2 respectively. If o\: X —> X\ and 02 • X —> X2 are (possibly ramified) covers of the same prime degree p and g > p2 + (gi + g2 — 2)p + 2 then there is an isomorphism a: X\ —> X2 such that the diagram X X\ a > X2 commutes. Note that here the covers o\ and 02 are not assumed to be Galois. Proof. Let Y be the image of the map X ^ X l x X 2 . Denote the degree of X —> Y by s and the degree of the projection Y —> Xi by ri. Then sr\ = sr2 = p, so either (i) s = p and r\ = r2 = 1 or (ii) s = 1 and r\ = r2 — p. 19 Chapter 2. Prime Galois Coverings of P 1 If (i) holds then Y is the graph of an isomorphism a between X\ and X2, which yields the desired result. It thus remains to rule out (ii), that is, to show that s ^ 1. Assume the contrary. Then Y is birationally isomorphic to X (but possibly singular). Let H = Xi x {pt} and V = {pt} x X2. (Here H stands for "horizontal" and V for "vertical".) Then Y • H = r2 = p, Y • V = r\ = p . An inequality of Castelnuovo and Severi now tells us that X -X < 2nr2 = 2p 2; see [Ha, Exercise V.1.9]. By the adjunction formula [Ha] we now see that 2 5 - 2 < X-(X+K) = X-(X+(29l-2)V+{2g2-2)H) < 2p2 + (2gi+2g2-2)p contradicting our assumption on g. • Theorem 5. Let X be a curve of genus g and H\, H2 be subgroups of Aut(X) of prime order p. Denote the genus ofX/Hi by gi. If 9>p2 + (9i+92- 2)p + 2 then Hi — H2. Proof. Take Xi = X/Hi and Oi: X —> Xi to be the natural projection. By Lemma 1 there is an isomorphism a between Xi and X2 such that aoi = a2. Let x € X in general position and Xi = o~i(x) € Xi for i = 1,2. Then a(xi) — x2 and o-2l(x2) = o71(a~1x2) = o7l(xi). But in our situation, o~l(xi) is precisely Hi • x^ In other words, we have Hi • x = H2 • x for every x G X. Equivalently, H\ = H2, as claimed. • Corollary 3. Let X be a curve of genus g > p 2 — 2p + 2. Then there is at most one subgroup H of order p in Aut(X) such that X/H = P 1 . In particular, if such a subgroup H exists, it is normal in Aut(X). 20 Chapter 2. Prime Galois Coverings of P 1 2.2.1 A Special Case We briefly consider the specific case where X is a compact Riemann surface of genus 4 that is a 3-sheeted covering of P 1 . Proposition 5. Let X be a compact Riemann surface of genus 4. Suppose there exist functions on X of degree 3. Then one of the following conditions holds. Either (i) There exists a function f of degree 3 on X such that (/) = A — D, where A, D are effective and D + D ~ K, and any other function of degree 3 on X is a Mobius transformation of f; or (ii) There is no such function, and there are exactly two functions of degree 3 on X which are not Mobius transformations of each other. Proof. [Fa] We first note that 3 = g — 1, so Remark 3 implies that every divisor D of degree 3 is a special divisor. By Clifford's Theorem, 1(D) < I < 3. (i): Suppose there exists a function / of degree 3 such that (/) = A — D with A, D effective and D + D ~ K. By definition, / e L(D), so 1(D) > 1. From the preceding comment, we have 1 < £(D) < 3, so 1(D) = 2. Assume that there is an effective divisor D\ / D with deg D\ = 3 and £(D\) — 2. (If no such divisor exists, then the only functions of degree 3 on X are constant multiplies of / and (i) is proved.) Let f\ £ L(D\)\C. If Di is linearly equivalent to D, then there exists h £ C(X) such that (h) + D\ = D. Then L(D) = (1,/) and L(D\) — (h,fh) = (l,/i). Thus there are constants a,b,c,d € C such that 1 = afh + bh fl — cfh + dh so that af + b cf + d Now assume f\ is not a Mobius transformation of / , which means D\ is not linearly equivalent to D. This implies l , / , / i , / / i are four linearly independent functions in L(D-\-D\). So £(D+Di) > 4, and by the Riemann-Roch Theorem, e(K-D-Di) =l(D + Di)-deg(D + Di)-l + g = l(D + Di)-§-l + A > 1. 21 Chapter 2. Prime Galois Coverings of P 1 Since deg(D + D\) = 6, then D + D\ is a canonical divisor. But D + D is also a canonical divisor, which implies D\ is linearly equivalent to D, which contradicts our assumption. (ii): We now assume that there is no function of degree 3 such that twice the polar divisor of / is canonical. Let / be a function of degree 3 so that (/) = A — D. By Remark 1, D is a special divisor, so Proposition 3 allows us to choose an effective divisor D\ of degree 3 such that D+Di ~ K. Note that D and D\ are not linearly equivalent. As before, Clifford's Theorem gives £(D) = 2, so Remark 1 and Proposition 2 yield l(K - D) = i(D) = 2. This implies there is a divisor D2 of degree 3 such that D2 / D\ and D + D2 ~ K. Choose holomorphic differentials such that (WJ) = D + Di, % = 1,2, and define f\ = u>2/u>\. Since the polar divisor D\ of f\ is not linearly equivalent to D, then we have constructed a function of degree 3 which is not a Mobius transformation of / . Let E denote the polar divisor of a function h of degree 3, where h is not a Mobius transformation of / . Then 1, /, h, fh are four linearly independent functions in L(D + E). Using the same argument as in Case (i), we have that D + E is a canonical divisor. If to is a differential such that (tu) = D + E, then (w) > D, so ui € Q(D). Since i(D) = £(K - D) = 2, then to = ciwi + c2to2 for constants C i , C 2 not both zero. Suppose c2 / 0. Then u)\/(c\U)\ + c2u)2) is a meromorphic function with divisor Remark 3. We revisit Example 3 and consider the case where p = 3 and k = 1. Let x : X —> P 1 denote the projection onto the re-coordinate. The divisor of this function is given by (x) = (0, A) + (0, (3A) + (0, Cf A) - ooj - oo2 - co3 = A - D. The divisor of the differential (l/y2)dx is {D + D1)-(D + E) = D1-E. Arguing as before, h is thus a Mobius transformation of f\. • 3 i=l 22 Chapter 2. Prime Galois Coverings of P 1 We see that since the canonical divisor K has degree 6, then the genus of X is 4. Since D is the polar divisor of a function of degree 3 on X with D + D = K, then by Proposition 5, every function of degree 3 on X is a Mobius transformation of the projection x, and hence the covering of X over P 1 is unique. Remark 4. Proposition 5 also shows that for the curves D 3 (Example 2 with p — 3), (ri) and (T2) are the only subgroups of order 3 which quotient to P 1 . 23 Chapter 3 The Main Theorem W e prove the fol lowing theorem. Theorem 6. [G-D] Let X be a compact Riemann surface of genus g. If A u t ( X ) has two distinct subgroups Hi, H2 of prime order p such that the quo-tient X/Hi is isomorphic to P 1 , then Hi and Hi are conjugate in Aut(X). Moreover, g is either (p — l ) 2 or (p — l)(p — 2 ) / 2 . 3.1 Two Lemmas Lemma 2. Let a,r G Aut(X) have prime order p. Assume that a and T commute and (cr) ^ ( r ) . Then a permutes the fixed point set of r, F i x ( r ) , dividing it into a number of orbits of length p. In particular, the number of points fixed by r is a multiple of p. Proof. If P e F i x ( r ) , then OT(P) = ra(P) impl ies a ( P ) , < 7 2 ( P ) , . . . , ^ - 1 ( P ) e F i x ( r ) . Since the stabi l izer of any point is cyc l ic (Propos i t ion 4), then if P is f ixed by bo th T and a, we have (cr) = ( r ) , cont rad ic t ing our assumpt ion. So c ' ( P ) / ai[P) € F i x ( r ) for 0 < % / j < p - 1. T h u s P,a{P),o2{P),...^^{P) is an orbi t of length p, and hence | F i x ( r ) | is a mul t ip le of p. • Lemma 3. Let X be of arbitrary genus with commuting automorphisms r, o of same prime order p such that X/(r) = P 1 and (a) / ( r ) . Then X is isomorphic to the compact Riemann surface defined by an algebraic equation of the form yp = (xp - l)(xp - Xp)m2 •••(xp- \ P ) m r where 1, A j , . . . , A? € C are all distinct and 1 < rrij < p, for i = 1 , . . . , r. Proof. Since X is a p-sheeted Ga lo is covering of P 1 , X admi ts an algebraic mode l of the form yp = (x-ai)d'{x-a2)d2 •••{x-an)d\ 1 < di < p, 24 Chapter 3. The Main Theorem in which r is expressed as r(x, y) = (x, (py) and (a*, 0), i = 1,..., n are the fixed points of r. Remark 2 allows us to assume d\ = 1, and Lemma 2 implies n = rp for some integer r. Since X/(T) == P 1 (the projection x is the quotient map), then a induces an automorphism o : P 1 —> P 1 of order p. By normalizing x so 0 and oo are the fixed points of 5 and a\ = 1, we can assume that o is the rotation cr(x) = £p • x for some p-th root of unity (p. We can number the fixed points of r so that {(a(fc-i)P+i.O), • • •, (flfcp.O)}, fc = l , . . . r are the c-orbits given in Lemma 2. Then d(fc_1)p+i = • • • = d}.p and a(fc-l)p+2 = ^ ( a ( f c-l ) p+l) = Cp • a ( fc-l )p+l-Thus the algebraic model of X is yp = fl (x - Q f[(x- C;A 2 )^ • • • fl (x - c p A r ) " \ as desired. We also note that a is necessarily of the form er(x, y) — ((px, (py) for some integer 1 < j < p. • 3.2 The Crucial Step Proposition 6. Let X be of arbitrary genus and p > 2 a prime. Then Aut(X) contains a p-group H of order > p 2 generated by automorphisms 7"i, T2 of prime order p with X/fa) = P 1 if and only if X is isomorphic to one of the following curves: Cp:yP = XP~ 1, Dp:yP = (xP - l)(xP - A p ) p _ 1 , for some A 6 C, A p ^ 1. Moreover, in both cases, H is the group of order p2 consisting of automor-phisms of the form r(x, y) = (Cpx, (py) with l<s,d<p. Proof. Since p-groups have nontrivial centre, there exists cr € H of order p commuting with T\ and r 2 . By Lemma 3, X has the form yp = (xp - l)(xp - Af) m 2 • • • (xp - A p ) m r 25 Chapter 3. The Main Theorem with n(x,y) = {x,(Py) and a(x,y) = {(px,(^y). We want to show that T2 has at most 2p fixed points. Suppose otherwise. Then T2 has more than 2p fixed points. Consider x o T2 : X -> P 1 , where x : X —> X/{T\) = P 1 is the projection map. Since a; has degree p, then the Riemann-Hurwitz Formula gives g> ( p - l ) ( 2 p - 2 ) / 2 = p 2 - 2 p + l . By Corollary 3, this means that there is at most one subgroup of Aut(X) of order p such that the quotient space is P 1 . This is a contradiction since ( r i ) 7^ ( T 2)- Thus T2 has at most 2p fixed points. The Riemann-Hurwitz Formula is the same for T\ and T2, so both have the same number of fixed points. By Lemma 2, r\ has either p or 2p fixed points, so X is of one of the following forms: Cp:yp = x p - 1, D^:yp = [xp - l)(xp - A p) f c. At this point, we split into two cases, the case where p = 3 and the case where p > 3. We first assume p > 3. We claim that H = (TI,O), meaning T2 € Suppose the contrary. Then T\ and T2 induce on X = X/{c) automorphisms f\ and T2 which again satisfy our hypotheses. Since the genus g of X is strictly less than the genus g of X, this forces X = Cp and X = Dp for some fc G Z and A G C. For the projection X —> X/(o) = X, the Riemann-Hurwitz Formula states 2g - 2 = p(2g - 2) + (p - 1) • (#Fix(<r)). Since p > 3, this contradicts the Riemann-Hurwitz Formulas for Dp —> Dp/{T\) and Cp —> C p / (n) given respectively by 2 5 - 2 = p(-2) + (p - l)2p 2g-2 = p(-2) + (p- l )p . Thus the group H = (T\, T2) consists of the elements r(x,y) = {Cp:i:,Cpy), with 1 <i,j < p. Example 3 illustrates that in order for H to contain two distinct proper subgroups with quotient P 1 , we must have fc = p — 1, so The case p = 3 follows from Remarks 3 and 4. • 26 Chapter 3. The Main Theorem Proof of Theorem 6. By the uniqueness of the hyperelliptic involution, we assume p > 3. Let T\,T2 G Aut(X) be as given. Then each (TV) is contained in some Sylow p-subgroup Hi (i = 1,2). Since Sylow groups are conjugate, there is <p G Aut(X) such that T[ = <p o n o ip~l G H2. Now if (T{) ^ (T2), Proposition 6 implies X is either Cp or Dp and i7 = (T{,T2) is the group given there. From Examples 1 and 2, we have that r{ and r2 are conjugate, which implies that T\ and T 2 are also conjugate. • 3.3 Conclusion In proving Theorem 6, we now know exactly when the automorphism r is not unique. The curves we examined in Examples 1 and 2 turn out to be the only counterexamples in our quest for uniqueness. Moreover, the au-tomorphisms in the counterexamples are conjugate, which is the next best thing to being unique. Much of the beauty in these results comes from the lack of use of high powered techniques. The classical tools we used here, such as the Riemann-Roch Theorem and the Riemann-Hurwitz Formula, allow for interesting re-sults without losing much of the intuition and geometry inherent in the subject. 27 Bibliography [Fa] H. Farkas and I. Kra, Riemann Surfaces, 2nd ed., Springer-Verlag, New York, 1992. [Fu] W. Fulton, Algebraic Curves, 2nd ed., W.A. Benjamin, Inc., Mas-sachusetts, 1974. [G-D] G. Gonzalez-Diez, On prime Galois coverings of the Riemann sphere, Ann. Mat. Pura Appl. (4) 168 (1995), 1-15. [Gr] P.A. Griffiths, Introduction to Algebraic Curves, American Math-ematical Society, 1989. [Ha] R. Hartshorne, Algebraic geometry, Graduate Texts in Mathemat-ics, No. 52., Springer-Verlag, New York-Heidelberg, 1977. [Hu] K. Hulek, Elementary Algebraic Geometry, American Mathemati-cal Society, 2003. [M] R. Miranda, Algebraic Curves and Riemann Surfaces, American Mathematical Society, 1995. [P] B. Poonen, Varieties without extra automorphisms. I. Curves, Math. Res. Lett. 7 (2000), no. 1, 67-76. [Sh] J. Shurman, Geometry of the Quintic, John Wiley & Sons Inc., New York, 1997. 28
- Library Home /
- Search Collections /
- Open Collections /
- Browse Collections /
- UBC Theses and Dissertations /
- Compact Riemann surfaces : prime Galois coverings of...
Open Collections
UBC Theses and Dissertations
Featured Collection
UBC Theses and Dissertations
Compact Riemann surfaces : prime Galois coverings of P¹ Tsiang, Michael 2007
pdf
Notice for Google Chrome users:
If you are having trouble viewing or searching the PDF with Google Chrome, please download it here instead.
If you are having trouble viewing or searching the PDF with Google Chrome, please download it here instead.
Page Metadata
Item Metadata
Title | Compact Riemann surfaces : prime Galois coverings of P¹ |
Creator |
Tsiang, Michael |
Publisher | University of British Columbia |
Date Issued | 2007 |
Description | The uniqueness of the hyperelliptic involution is well known in the theory of Riemann surfaces. More precisely, we know that if X is a hyperelliptic compact Riemann surface, there is a unique automorphism τ of order 2 such that X/〈τ〉 ≅ ℙ¹ . We wish to generalize the situation slightly. We say X is a prime Galois covering of ℙ¹ if there exists an automorphism τ of (odd) prime order p such that X/〈T〉 ≅ ℙ¹. This leads us to ask the question: When is this automorphism τ unique? We begin by building the necessary background to understand prime Galois coverings of ℙ¹. We then prove a theorem due to Gonzlez-Diez that answers our question about uniqueness. The proof given here follows his proof (given in [G-D]) quite closely, though we elaborate and modify certain details to make it more self contained. |
Genre |
Thesis/Dissertation |
Type |
Text |
Language | eng |
Date Available | 2011-03-16 |
Provider | Vancouver : University of British Columbia Library |
Rights | For non-commercial purposes only, such as research, private study and education. Additional conditions apply, see Terms of Use https://open.library.ubc.ca/terms_of_use. |
DOI | 10.14288/1.0080444 |
URI | http://hdl.handle.net/2429/32490 |
Degree |
Master of Science - MSc |
Program |
Mathematics |
Affiliation |
Science, Faculty of Mathematics, Department of |
Degree Grantor | University of British Columbia |
Campus |
UBCV |
Scholarly Level | Graduate |
AggregatedSourceRepository | DSpace |
Download
- Media
- 831-ubc_2007-0623.pdf [ 1.22MB ]
- Metadata
- JSON: 831-1.0080444.json
- JSON-LD: 831-1.0080444-ld.json
- RDF/XML (Pretty): 831-1.0080444-rdf.xml
- RDF/JSON: 831-1.0080444-rdf.json
- Turtle: 831-1.0080444-turtle.txt
- N-Triples: 831-1.0080444-rdf-ntriples.txt
- Original Record: 831-1.0080444-source.json
- Full Text
- 831-1.0080444-fulltext.txt
- Citation
- 831-1.0080444.ris
Full Text
Cite
Citation Scheme:
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}]}"
data-media="{[{embed.selectedMedia}]}"
async >
</script>
</div>
Our image viewer uses the IIIF 2.0 standard.
To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.831.1-0080444/manifest