Elliptic Curves, Modular Forms & Fermat’s Last Theorem Proceedings of a conference held in the Institute of Mathematics of the Chinese University of Hong Kong Edited by John Coates University of Cambridge S.T. Yau Harvard Universtiy & Director of the Institute of Mathematics of the Chinese University of Hong Kong Second Edition International PressEditorial Board Barry Mazur Harvard University 1 Oxford Street Cambridge, MA 02138 USA. Copyright © 1997 by International Press International Press 96 Mt Auburn Str P.O. Box 2872 Cambridge, MA 02238-2872 All rights reserved. No part of this work can be reproduced in any form, electronic or mechanical, recording, or by any information storage and data retrieval system, without specific authorization from the publisher. Reproduction for classroom or personal use will, in most cases, be granted without charge. Elliptic Curves, Modular Forms & Fermat's Last Theorem, 2nd Edition Edited by J. Coates and S.-T. Yau ISBN 1-57146-049-7 Printed on acid free paper in the United States of AmericaTable of Contents Foreword by John Coates and Shing-Tung Yau .........scscscseseerseeees 1 Fermat’s Last Theorem H. DARMON, F. DIAMOND, R. TAYLOR ...+.eceseeeeeeeeeeeeeteteeeneenennene 2 Characteristic distributions for non-torsion A-modules IE PSAL IST ER ete teeette eee eects eee testes ete ee teeters 141 On the symmetric square of a modular elliptic curve J. COATES, A. SYDENHAM... 00. cc ccceceeeeeeeeeeeeeeeeeeeeeeeeneeeeaeeees 152 ‘The refined conjecture of Serre FRED DIAMOND ........2c0cceceseeeesceeseeceseeseceeseeeeeseessaeeeeeae 172 Wiles minus epsilon implies Fermat NOAM ELKIES.........0.2:0eceeceeseeer ence eeeeeseeeeceeneeeweserereeeeeee 187 Geometric Galois representations J.-M. FONTAINE, B. MAZUR. ....000.0cccseecesceeeeeseessesseenseeseenees 190 On elliptic curves GERHARD FREY Complete intersections and Gorenstein rings H.W. LenstRA, JR. Homologie des courbes modulaires L. MEREL.... 0. ee esse eeece nee sec ence eneeeeeeeseeereeteeeeepeteeeeeeeeene 258 Irreducible Galois representation KENNETH A. RIBET.... Mod p representations of elliptic curves K. Ruin, A, SILVERBERG A review of non-Archimedean elliptic functions JOHN TATE On Galois representations associated to Hilbert modular forms IT RICHARD TAYLOR.Conference on Elliptic Curves and Modular Forms Hong Kong, December 18-21, 1993 Copyright ©1997 International Press Foreword A conference, on the general theme of “Elliptic curves and modular forms”, was held in the Mathematics Department and the Institute of Mathematical Sciences of The Chinese University of Hong Kong from December 18-21, 1993. ‘The impetus for organizing the conference arose from Andrew Wiles’ deep and spectacular work on the celebrated conjecture that every elliptic curve over Q is modular, although only some of the lectures at the conference were specif- ically related to this theme. At the time of the conference, the difficulties in the last hurdle in Wiles’ work (the proof of the conjectural upper bound for the order of the Selmer group attached to the symmetric square of a modular form) had still not been overcome. However, the optimism shared by all at the conference that is was only a matter of time until the proof would become complete has happily been borne out by subsequent events. It is now history that Wiles himself, assisted by R. Taylor, found a beautiful proof of the desired upper bound. As a result, we now know today the remarkable fact that every semi-stable elliptic curve over Q is modular. Not only is this result revolution- ary in its own right for the study of the arithmetic of these elliptic curves, but it has the added bonus that it provides at last a proof of Fermat’s last theorem, thanks to the earlier work of Frey, Ribet and others. We are grateful to H. Darmon, F. Diamond and R. Taylor for their kind permission to publish their beautiful survey article on Wiles’ work as the first article in this second edition. During the conference itself, lectures were given by John Coates, Noam Elkies, Matthias Flach, Jean-Marc Fontaine, Gerhard Frey, Dick Gross, Victor Koly- vagin, Ken Ribet, Karl Rubin, Jean-Pierre Serre, John Tate, Richard Taylor, and Don Zagier. The present short volume is a mixture of the texts of some of these lectures, together with a number of recent articles related to the gen- eral theme of the conference. Finally, the organizers of the conference wish to express their warmest thanks to Professor Charles Kao, Vice-Chancellor, and to Professor S.Y. Cheng, Chairman of the Mathematics Department, and Dr. K.W. Leung, of The Chinese University of Hong Kong for their support and as- sistance throughout the preparation and running of the conference. Finally, and most importantly, the organizers wish to express their gratitude to The Chinese University of Hong Kong, the Ho Sin Hang Education Endowment Fund, the Lee Hysan Foundation Limited, the Pei Hua Education Foundation Limited, and the Sing Ho Yu Foundation for their generous financial support for the conference. John Coates, Cambridge University Shing-Tung Yau, Director Institute of Mathematical Sciences The Chinese University of Hong KongFermat’s Last Theorem Henrt DARMON Department of Mathematics, McGill University, Montreal, QC Canada H3A 2K6 FRED DIAMOND Departament of Mathematics, MIT, 77 Massachusetts Avenue, Camridge, MA 02139, USA RICHARD TAYLOR Departament of Mathematics, Harvard University, 1 Oxford Street, Cambridge, MA 02138, USA The authors would like to give special thanks to N. Boston, K. Buzzard, and B. Conrad for providing so much valuable feedback on earlier versions of this paper. They are also grateful to A. Agboola, M. Bertolini, B. Edixhoven, J. Fearnley, B.H. Gross, R. Gross, L, Guo, F. Jarvis, H. Kisilevsky, E. Liverance, J. Manoharmayum, K. Ribet, D. Rohrlich, M. Rosen, R. Schoof, J.-P. Serre, C. Skinner, D. Thakur, J. Tilouine, J. Tunnell, A. Van der Poorten, and L. Washington for their helpful comments. Darmon thanks the members of CICMA and of the Quebec-Vermont Num- ber Theory Seminar for many stimulating conversations on the topics of this paper, particularly in the Spring of 1995. For the same reason Diamond is grateful to the participants in an informal seminar at Columbia University in 1993-94, and Taylor thanks those attending the Oxford Number Theory Semi- nar in the Fall of 1995. Parts of this paper were written while the authors held positions at other institutions: Darmon at Princeton, Diamond at Cambridge and the Institute for Advanced Study, and Taylor at Cambridge and Oxford. During some of the period, Diamond enjoyed the hospitality of Princeton University, and Taylor that of MIT. The writing of this paper was also supported by research grants from NSERC (Darmon), EPSRC # GR/J9461 and NSF # DMS 9304580 (Di- amond) and by an advanced fellowship from EPSRC (Taylor). This article owes everything to the ideas of Wiles, and the arguments pre- sented here are fundamentally his [W3], though they include both the work [TW] and several simplifications to the original arguments, most notably that of Faltings. In the hope of increasing clarity, we have not always stated theo- rems in the greatest known generality, concentrating instead on what is needed for the proof of the Shimura-Taniyama conjecture for semi-stable elliptic curves. This article can serve as an introduction to the fundamental papers [W3] and [TW], which the reader is encouraged to consult for a different, and often more in-depth, perspective on the topics considered. Another useful more advanced reference is the article [Di2] which strengthens the methods of [W3] and [TW] to prove that every elliptic curve that is semistable at 3 and 5 is modular.For the reprinting of this paper the only changes we have made are to correct a few typos. We have not updated the material to take account of recent developments in the subject, most notably: « F, Diamond, The Taylor- Wiles construction and multiplicity one, Invent. Math. 128 (1997), 379-391 gives a simplification to the arguments which we would have incorporated into this exposition if we were writing it today. Some progress has been made on conjecture 3.13 for small primes. See N. Shepherd-Barron and R. Taylor, Mod 2 and mod 5 icosahedral repre- sentations, J. Amer. Math. Soc., 10 (1997), 283-298. The Shimura-Taniyama conjecture is now known for elliptic curves with conductor not divisible by 27. See B. Conrad, F. Diamond, R. Tay- lor, Modularity of certain potentially crystalline Galois representations, in preparation.Contents 1 Elliptic curves and modular forms 1.1 Elliptic curves . 1.2 Modular curves and modular forms over 1.3. Hecke operators and Hecke theory 1.4 The L-function associated to a cusp form 1.5 Modular curves and modular forms over Q 1.6 The Hecke algebra... . bee 1.7. The Shimura construction . . 18 The Shimura-Taniyama conjecture . . Galois theory 2.1 Galois representations 2.2 Representations associated to elliptic curves... . . 2.3. Galois cohomology . . . see 2.4 Representations of Ga, . : 2.5 The theory of Fontaine and Laffaille . 2.6 Deformations of representations . 2.7. Deformations of Galois representations 2.8 Special cases ..........2.05 Modular forms and Galois representations 3.1 From modular forms to Galois representations . . . 3.2. From Galois representations to modular forms 3.3. Hecke algebras 3.4 Isomorphism criteria . . 3.5 The main theorem . . . 3.6 Applications... 0.2.0... cece eee ee eee eee Hecke algebras 4.1 Full Hecke algebras . . . 4.2 Reduced Hecke algebras 4.3 Proof of theorem 3.31 4.4 Proof of theorem 3.36 4.5 Homological results . . . Commutative algebra 5.1 Wiles’ numerical criterion... .. « 5.2 Basic properties of 64 and na... . 5.3 Complete intersections and the Gorenstein condition . 5.4 The Congruence ideal for complete intersections . . 5.5 Isomorphism theorems . 5.6 A resolution lemma . 5.7 A criterion for complete intersections 5.8 Proof of Wiles’ numerical criterion . 5.9 A reduction to characteristic ¢ 5.10 J-structures.. 2.0... 2. ee 15 15 20 24 28 29 33 37 43 43 47 50 53, 58 61 64 67 71 71 74 78 82 83, 89 89 93 99 104 lll 115 115 117 119 123 124 126 127 127 128 129Introduction Fermat’s Last Theorem Fermat's Last Theorem states that the equation ay #0 has no integer solutions when n is greater than or equal to 3. Around 1630, Pierre de Fermat claimed that he had found a “truly wonderful” proof of this theorem, but that the margin of his copy of Diophantus’ Arithmetica was too small to contain it: “Cubum autem in duos cubos, aut quadrato quadratum in duos quadrato quadratos, et generaliter nullam in infinitum ultra qua- dratum potestatem in duos ejusdem nominis fas est dividere; cujus rei demonstrationem mirabile sane detexi. Hanc marginis exiguitas non caperet.” Among the many challenges that Fermat left for posterity, this was to prove the most vexing. A tantalizingly simple problem about whole numbers, it stood unsolved for more than 350 years, until in 1994 Andrew Wiles finally laid it to rest. Prehistory: The only case of Fermat’s Last Theorem for which Fermat actu- ally wrote down a proof is for the case n = 4. To do this, Fermat introduced the idea of infinite descent which is still one the main tools in the study of Dio- phantine equations, and was to play a central role in the proof of Fermat’s Last Theorem 350 years later. To prove his Last Theorem for exponent 4, Fermat showed something slightly stronger, namely that the equation x4 +y* = z? has no solutions in relatively prime integers with ryz # 0. Solutions to such an equation correspond to rational points on the elliptic curve v? = u3 —4u. Since every integer n > 3 is divisible either by an odd prime or by 4, the result of Fermat allowed one to reduce the study of Fermat’s equation to the case where n= Cis an odd prime. In 1753, Leonhard Buler wrote down a proof of Fermat’s Last Theorem for the exponent £ = 3, by performing what in modern language we would call a 3-descent on the curve 25 + y? = 1 which is also an elliptic curve. Euler’s argument (which seems to have contained a gap) is explained in [Edw], ch. 2, and (Dicl], p. 545. It took mathematicians almost 100 years after Euler’s achievement to han- dle the case £ = 5; this was settled, more or less simultaneously, by Gustav Peter Lejeune Dirichlet [Dir] and Adrien Marie Legendre (Leg] in 1825. Their elementary arguments are quite involved. (Cf. [Edw], sec. 3.3.) In 1839, Fermat's equation for exponent 7 also yielded to elementary meth- ods, through the heroic efforts of Gabriel Lamé. Lamé’s proof was even more intricate than the proof for exponent 5, and suggested that to go further, new theoretical insights would be needed. The work of Sophie Germain: Around 1820, in a letter to Gauss, Sophie Germain proved that if é is a prime and q = 2¢+ 1 is also prime, then Fermat’s equation z¢ + y¢ = z with exponent @ has no solutions (z, y,z) with zyz #06 Introduction (mod £). Germain’s theorem was the first really general proposition on Fer- mat’s Last Theorem, unlike the previous results which considered the Fermat equation one exponent at a time. The case where the solution (2, y, z) to z° + y¢ = z! satisfies zyz # 0 (mod £) was called the first case of Fermat’s Last Theorem, and the case where £ divides xyz, the second case. It was realized at that time that the first case was generally easier to handle: Germain’s theorem was extended, using similar ideas, to cases where k€+ 1 is prime and k is small, and this led to a proof that there were no first case solutions to Fermat’s equation with prime exponents £ < 100, which in 1830 represented a significant advance. The division between first and second case remained fundamental in much of the later work on the subject. In 1977, Terjanian [Te] proved that if the equation 2 + y = 2 has a solution (z, y, z), then 2€ divides either z or y, i.e., “the first case of Fermat’s Last Theorem is true for even exponents”. His simple and elegant proof used only techniques that were available to Germain and her contemporaries. The work of Kummer: The work of Ernst Eduard Kummer marked the beginning of a new era in the study of Fermat’s Last Theorem. For the first time, sophisticated concepts of algebraic number theory and the theory of L- functions were brought to bear on a question that had until then been addressed only with elementary methods. While he fell short of providing a complete solution, Kummer made substantial progress. He showed how Fermat’s Last Theorem is intimately tied to deep questions on class numbers of cyclotomic fields which are still an active subject of research. Kummer’s approach relied on the factorization (et u)(c+ Co) (e+ G79) of Fermat’s equation over the ring Z(C_] generated by the £th roots of unity. One observes that the greatest common divisor of any two factors in the product on the left divides the element (1 ~ ¢;), which is an element of norm £. Since the product of these numbers is a perfect é-th power, one is tempted to conclude that (x+y),... ,(c+¢{~'y) are each ¢-th powers in the ring Z[¢¢ up to units in this ring, and up to powers of (1 — (). Such an inference would be valid if one were to replace Z[(;] by Z, and is a direct consequence of unique factorization of integers into products of primes. We say that a ring R has property UF if every non-zero element of R is uniquely a product of primes, up to units. Mathematicians such as Lamé made attempts at proving Fermat's Last Theo- rem based on the mistaken assumption that the rings Z[¢/] had property UF. Legend even has it that Kummer fell into this trap, although this story now has been discredited; see for example [Edw], sec. 4.1. In fact, property UF is far from being satisfied in general: one now knows that the rings Z[¢/] have property UF only for £ < 23 (cf. [Wa], ch. 1). It turns out that the full force of property UF is not really needed in the applications to Fermat’s Last Theorem. Say that a ring R has property UF‘ if the following inference is valid: ab = 2, and ged(a,b) = 1 = a and b are éth powers up to units of R. If a ring R has property UF, then it also has property UF, but the converse need not be true. Kummer showed that Fermat’s last theorem was true forIntroduction 7 exponent ¢ if Z[¢¢] satisfied the property UF; (cf. [Wa]). The proof is far from trivial, because of difficulties arising from the units in Z[¢,] as well as from the possible failure of property UF. (A number of Kummer’s contemporaries, such as Cauchy and Lamé, seem to have overlooked both of these difficulties in their attempts to prove Fermat’s Last Theorem.) Kummer then launched a systematic study of the property UF for the rings Z[;]. He showed that even if Z[¢/ failed to have unique factorization, it still possessed unique factorization into prime ideals. He defined the ideal class group as the quotient of the group of fractional ideals by its subgroup consisting of principal ideals, and was able to establish the finiteness of this class group. The order of the class group of Z[¢,], denoted hz, could be taken as a measure of the failure of the ring Z[¢¢] to satisfy UF. It was rather straightforward to show that if € did not divide he, then Z[Cd satisfied the property UF. In this case, one called @ a regular prime. Kummer thus showed that Fermat’s last theorem is true for exponent ¢ if ¢ is a regular prime. He did not stop here. For it remained to give an efficient means of computing hg, or at least an efficient way of checking when ¢ divides hy. The class number hy can be factorized as a product he = hy hz, where hj is the class number of the real subfield Q((¢)*, and hig is defined as hy/hj. Essentially because of the units in Q((¢)*, the factor hj is somewhat difficult to compute, while, because the units in Q((p)+ generate the group of units in Q(¢¢) up to finite index; the term h; can be expressed in a simple closed form. Kummer showed that if £ divides hj, then ¢ divides hy . Hence, ¢ divides he if and only if £ divides h;. This allowed one to avoid the difficulties inherent in the calculation of hj. Kummer then gave an elegant formula for hy by considering the Bernoulli numbers By, which are rational numbers defined by the formula He produced an explicit formula for the class number h7, and concluded that if £ does not divide the numerator of Bz;, for 1 < i < (é—3)/2, then @ is regular, and conversely. The conceptual explanation for Kummer's formula for hz lies in the work of Dirichlet on the analytic class number formula, where it is shown that hy can be expressed as a product of special values of certain (abelian) L-series L(s,x) = > x(n)n=* n=l “ associated to odd Dirichlet characters. Such special values in turn can be expressed in terms of certain generalized Bernoulli numbers By,y, which are related to the Bernoulli numbers B; via congruences mod é. (For more details, see [Wa].) These considerations led Kummer to initiate a deep study relating congru- ence properties of special values of Z-functions and of class numbers, which was to emerge as a central concern of modern algebraic number theory, and was to reappear ~ in a surprisingly different guise — at the heart of Wiles’ strategy for proving the Shimura-Taniyama conjecture.8 Introduction Later developments: Kummer's work had multiple ramifications, and led to a very active line of enquiry pursued by many people. His formulae relating Bernoulli numbers to class numbers of cyclotomic fields were refined by Kenneth Ribet [Ri], Barry Mazur and Andrew Wiles [MW], using new methods from the theory of modular curves which also play a central role in Wiles’ more recent, work. (Later Francisco Thaine [Th] reproved some of the results of Mazur and Wiles using techniques inspired directly from a reading of Kummer.) In a development more directly related to Fermat’s Last Theorem, Wieferich proved that if €? does not divide 2‘~! — 1, then the first case of Fermat’s Last Theorem is true for exponent é. (Cf. [Ri], lecture VIII.) ‘There were many other refinements of similar criteria for Fermat’s Last theorem to be true. Computer calculations based on these criteria led to a verification that Fermat’s Last theorem is true for all odd prime exponents less than four million [BCEM], and that the first case is true for all £ < 8.858 - 10?° [Su]. The condition that ¢ is a regular prime seems to hold heuristically for about. 61% of the primes. (See the discussion on p. 63, and also p. 108, of [Wa], for example.) In spite of the convincing numerical evidence, it is still not known if there are infinitely many regular primes. Ironically, it is not too difficult to show that there are infinitely many irregular primes. (Cf. [Wa].) Thus the methods introduced by Kummer, after leading to very strong results in the direction of Fermat’s Last theorem, seemed to become mired in difficulties, and ultimately fell short of solving Fermat’s conundrum’. Faltings’ proof of the Mordell conjecture: In 1985, Gerd Faltings [Fa] proved the very general statement (which had previously been conjectured by Mordell) that any equation in two variables corresponding to a curve of genus strictly greater than one had (at most) finitely many rational solutions. In the context of Fermat’s Last Theorem, this led to the proof that for each exponent n > 3, the Fermat equation 2" +y" = 2” has at most finitely many integer solutions (up to the obvious rescaling). Andrew Granville [Gra] and Roger Heath-Brown [HB] remarked that Faltings’ result implies Fermat’s Last Theorem for a set of exponents of density one. However, Fermat’s Last Theorem was still not known to be true for an infinite set of prime exponents. In fact, the theorem of Faltings seemed ill- equipped for dealing with the finer questions raised by Fermat in his margin, namely of finding a complete list of rational points on all of the Fermat curves zx" +y" = 1 simultaneously, and showing that there are no solutions on these curves when n > 3 except the obvious ones. Mazur’s work on Diophantine properties of modular curves: Although it was not realized at the time, the chain of ideas that. was to lead to a proof of Fermat’s Last theorem had already been set in motion by Barry Mazur in the mid seventies. The modular curves Xo(é) and X(¢) introduced in sec- tion 1.2 and 1.5 give rise to another naturally occurring infinite family of Dio- phantine equations. These equations have certain systematic rational solutions corresponding to the cusps that are defined over Q, and are analogous to the ‘However, W. McCallum has recently introduced a technique, based on the method of Chabauty and Coleman, which suggests new directions for approaching Fermat’s Last Theo- rem via the cyclotomic theory. An application of McCallum’s method to showing the second case of Fermat’s Last Theorem for regular primes is explained in [Mc].Introduction 9 so-called “trivial solutions” of Fermat’s equation. Replacing Fermat curves by modular curves, one could ask for a complete list of all the rational points on the curves Xo(é) and X;(é). This problem is perhaps even more compelling than Fermat's Last Theorem: rational points on modular curves correspond to objects with natural geometric and arithmetic interest, namely, elliptic curves with cyclic subgroups or points of order é. In [Mazl] and [Maz2], B. Mazur gave essentially a complete answer to the analogue of Fermat’s Last Theorem for modular curves. More precisely, he showed that if ¢ # 2,3,5 and 7, (i.e., X;(é) has genus > 0) then the curve Xj(é) has no rational points other than the “trivial” ones, namely cusps. He proved analogous results for the curves Xo(é) in [Maz2], which implied, in particular, that an elliptic curve over Q with square-free conductor has no rational cyclic subgroup of order £ over Q if ¢ is a prime which is strictly greater than 7. This result appeared a full ten years before Faltings’ proof of the Mordell conjecture. Frey’s strategy: In 1986, Gerhard Frey had the insight that these construc- tions might provide a precise link between Fermat’s Last Theorem and deep questions in the theory of elliptic curves, most notably the Shimura Taniyama conjecture. Given a solution af + b¢ = c to the Fermat equation of prime degree £, we may assume without loss of generality that a¢ = —1 (mod 4) and that 6¢ = 0 (mod 32). Frey considered (following Hellegouarch, [He], p. 262; cf. also Kubert-Lang [KL], ch. 8, §2) the elliptic curve E:y =2(¢-a')(r +4). This curve is semistable, i.e., it has square-free conductor. Let E[é] denote the group of points of order é on E defined over some (fixed) algebraic closure Q of Q, and let L denote the smallest number field over which these points are defined. This extension appears as a natural generalization of the cyclotomic fields Q(¢z) studied by Kummer. What singles out the field L for special attention is that it has very little ramification: using Tate’s analytic description of E at the primes dividing abe, it could be shown that L was ramified only at 2 and é, and that the ramification of L at these two primes was rather restricted. (See theorem 2.15 of section 2.2 for a precise statement.) Moreover, the results of Mazur on the curve Xo(é) could be used to show that L is large, in the following precise sense. The space E(é] is a vector space of dimension 2 over the finite field Fy with ¢ elements, and the absolute Galois group Ga = Gal (Q/Q) acts Fy-linearly on E{é). Choosing an F;-basis for E{é], the action is described by a representation Bee : Gal(L/Q) 4 GLa(Fe)- Mazur’s results in (Mazl] and (Maz2] imply that gy is irreducible if @ > 7 (using the fact that E is semi-stable). In fact, combined with earlier results of Serre [Se6], Mazur’s results imply that for @ > 7, the representation pg, is surjective, so that Gal (L/Q) is actually isomorphic to GL2(Fy) in this case. Serre’s conjectures: In (Se7], Jean-Pierre Serre made a careful study of mod £ Galois representations f : Gg —+ GL2(Fe) (and, more generally, of repre- sentations into GL2(k), where k is any finite field). He was able to make very precise conjectures (see section 3.2) relating these representations to modular forms mod ¢. In the context of the representations fz, that occur in Frey’s construction, Serre’s conjecture predicted that they arose from modular forms10 Introduction (mod €) of weight two and level two. Such modular forms, which correspond to differentials on the modular curve Xo(2), do not exist because Xo(2) has genus 0. Thus Serre’s conjecture implied Fermat’s Last Theorem. The link between fields with Galois groups contained in GL2(F,) and modular forms mod £ still appears to be very deep, and Serre’s conjecture remains a tantalizing open problem. Ribet’s work: lowering the level: The conjecture of Shimura and Taniya- ma (cf. section 1.8) provides a direct link between elliptic curves and modular forms. It predicts that the representation fg,¢ obtained from the ¢-division points of the Frey curve arises from a modular form of weight 2, albeit a form whose level is quite large. (It is the product of all the primes dividing abc, where a‘ + b¢ = c! is the putative solution to Fermat’s equation.) Ribet (R5] proved that, if this were the case, then fg,¢ would also be associated with a modular form mod £ of weight 2 and level 2, in the way predicted by Serre’s conjecture. This deep result allowed him to reduce Fermat’s Last Theorem to the Shimura-Taniyama conjecture. Wiles’ work: proof of the Shimura-Taniyama conjecture: In [W3] Wiles proves the Shimura-Taniyama conjecture for semi-stable elliptic curves, provid- ing the final missing step and proving Fermat’s Last Theorem. After more than 350 years, the saga of Fermat's Last theorem has come to a spectacular end. The relation between Wiles’ work and Fermat's Last Theorem has been very well documented (see, for example, [R8], and the references contained therein). Hence this article will focus primarily on the breakthrough of Wiles {W3] and Taylor-Wiles [TW] which leads to the proof of the Shimura-Taniyama conjecture for semi-stable elliptic curves. From elliptic curves to é-adic representations: Wiles’ opening gambit for proving the Shimura-Taniyama conjecture is to view it as part of the more gen- eral problem of relating two-dimensional Galois representations and modular forms. The Shimura-Taniyama conjecture states that if E is an elliptic curve over Q, then E is modular. One of several equivalent definitions of modularity is that for some integer N there is an eigenform f = )> ang” of weight two on To(N) such that #E(Fp) =p+1—ap for all but finitely primes p. (By an eigenform, here we mean a cusp form which is a normalized eigenform for the Hecke operators; see section 1 for definitions.) This conjecture acquires a more Galois theoretic flavour when one considers the two dimensional é-adic representation PEt: Ga — GL2(Zo) obtained from the action of Gg on the ¢-adic Tate module of E: TE = lim B{t"|(Q). An £-adic representation p of Gg is said to arise from an eigen- form f = Dang" with integer coefficients ap if tr (p(Frob p)) = ap, for all but finitely many primes p at which p is unramified. Here Frob, is a Frobenius element at p (see section 2), and its image under p is a well-defined conjugacy class.Introduction ll A direct computation shows that #E(F,) = p+ 1- tr(pz,¢(Frobp)) for all primes p at which pg, is unramified, so that E is modular (in the sense defined above) if and only if for some @, pg, arises from an eigenform. In fact the Shimura-Taniyama conjecture can be generalized to a conjecture that every ¢-adic representation, satisfying suitable local conditions, arises from a modular form. Such a conjecture was proposed by Fontaine and Mazur [FM]. Galois groups and modular forms Viewed in this way, the Shimura-Taniyama conjecture becomes part of a much larger picture: the emerging, partly conjectural and partly proven correspon- dence between certain modular forms and two dimensional representations of Gg. This correspondence, which encompasses the Serre conjectures, the Fontaine-Mazur conjecture, and the Langlands program for GL2, represents a first step toward a higher dimensional, non-abelian generalization of class field theory. Two-dimensional representations of Gg: In the first part of this century, class field theory gave a complete description of G3?, the maximal (continu- ous) abelian quotient of Gg. In fact the Kronecker-Weber theorem asserts that GY = T], Z}, and one obtains a complete description of all one-dimensional representations of Gg. In the second half of this century much attention has focused on attempts to understand the whole group Gg, or more precisely to describe all its representations. Although there has been a fair degree of suc- cess in using modular forms to construct representations of Gg, less is known about how exhaustive these constructions are. The major results in the lat- ter direction along these lines are the work of Langlands [L12] and the recent work of Wiles ([W3] completed by [TW]). Both concern two-dimensional rep- resentations of Gg and give significant evidence that these representations are parametrised (in a very precise sense) by certain modular forms. The purpose of this article is to describe both the proven and conjectural parts of this the- ory, give a fairly detailed exposition of Wiles’ recent contribution and explain the application to Fermat’s Last theorem. To make this description somewhat more precise let us distinguish three types of representation. Artin representations and the Langlands-Tunnell theorem: Contin- uous representations p : Gg + GLa(C) are called (two-dimensional) Artin representations. Such representations necessarily have finite image, and are therefore semi-simple. We restrict our attention to those which are irreducible. They are conjectured to be in bijection (in a precise way) with certain newforms (a special class of eigenforms). Those p which are odd (ie. the determinant of complex conjugation is -1), should correspond to weight 1 holomorphic new- forms. Those which are even should correspond to certain non-holomorphic (Maass) newforms. Two partial but deep results are known. (a) (Deligne-Serre) If f is a holomorphic weight one newform then the cor- responding Artin representation can be constructed ({DS]). (b) (Langlands-Tunnell) If p is a two dimensional Artin representation with soluble image then the corresponding modular form exists ((L12] and {Tu)).12 Introduction The proof of the latter result is analytic in nature, invoking the trace formula and the theory of L-functions. t-adic representations and the Fontaine-Mazur conjecture: By an ¢- adic representation we shall mean any continuous representation p : Gg -> GL,(K) which is unramified outside a finite set of primes and where K is a finite extension of Q, (generalizing slightly the notion of ¢-adic representation that was introduced before). Given a holomorphic newform f one can attach to f a system of £-adic representations, following Eichler, Shimura, Deligne and Serre. These ¢-adic representations are called modular. The Fontaine-Mazur conjecture (see [FM]) predicts if p is an odd, irreducible, é-adic representation whose restriction to the decomposition group at £ is well enough behaved, then pis modular. (The restriction on the behaviour of the representation on the decomposition group at £ is essential in this conjecture; it is not true that all odd, irreducible two dimensional é-adic representation are modular.) Before Wiles’ work almost nothing was known about this conjecture, except that cer- tain very special cases could be deduced from the work of Hecke, Langlands and Tunnel. Mod £ representations and Serre’s conjecture: A mod ¢ representation is a continuous representation p : Gg —+ GL2(f,). For example if E/Q is an elliptic curve then the action of Gg on the ¢-division points of E gives rise to a mod £ representation fg, which is just the reduction modulo é of pg,¢. One can use the work of Eichler, Shimura, Deligne and Serre to associate to each mod £ eigenform a mod £ representation of Gg. The mod £ representations which arise in this way are called modular. Serre has conjectured [Se7] that every odd (absolutely) irreducible mod é representation is modular and should arise from a mod £ eigenform with certain very specific properties. This conjecture can be thought of as having two parts. The first asserts that every odd irreducible mod ¢ representation is modular. About this very little is known. It is known for p : Gg > GL2(F2) by work of Hecke. It is also known for f : Gg + GL2(Fs). This latter result is an application of the Langlands-Tunnell theorem using the two accidents that there is a section to the homomorphism GL2(Z[V—2]) -* GLo(F3) and that GL,(Fs) is soluble. Partial results for p : Gg -> GL2(Fs) follow from Wiles’ work. Given a mod £ representation arising from a mod ¢ eigenform, the second part of Serre’s conjecture predicts the minimal weight and level for that mod £eigenform. Here the situation is much better. There has been a lot of work over the last decade (including ideas from Mazur, Ribet, Carayol and Gross) and the problem is nearly completely resolved (see [Dil]). As was pointed out earlier, Ribet’s contribution [R5] implies that, if one can show that the Galois representation fg,¢ arising from the (semi-stable) Frey curve attached to a solution of Fermat's equation with exponent ¢ is modular, then one can show that this representation does not exist—because it would be modular of weight two and level two— and hence one can deduce Fermat’s Last Theorem. However we have seen that to show fz, is modular it suffices to show that for some 9, the éo-adic representation pg,¢, is modular. In particular it suffices to verify that either pg,3 or pgs is modular. Hence the Shimura-Taniyama conjecture can be reduced to (part of) the Fontaine-Mazur conjecture for £ = 3Introduction 13 and 5. We have seen that for these primes part of Serre’s conjecture is known, so it turns out it suffices to prove results of the form “Serre’s conjecture for € implies the Fontaine-Mazur conjecture for ¢”. This is the direction of Wiles’ work, although nothing quite this general has been proven yet. Deformation theory: Thus the problem Wiles faces is to show that if p is an odd ¢-adic representation which has irreducible modular reduction p and which is sufficiently well behaved when restricted to the decomposition group at ¢, then p is modular. In fact he only proves a weakened version of such a result, but one which is sufficient to conclude that all semistable elliptic curves are modular. Wiles approaches the problem by putting it in a more general setting. On the one hand he considers lifts of f to representations over complete noetherian local Zealgebras R. For each finite set of primes 5, one can consider lifts of type ©; these are lifts which are well-behaved on a decomposition group at @, and whose ramification at primes not in ¥ is rather restricted. In particular, such a lift is unramified outside DU S where S is the set of ramified primes of f. A method of Mazur (see [Maz3]) can then be used to show that if p is absolutely irreducible, then there is a representation pt : Gq + Gla(Rr) which is universal in the following sense. If p: Gg -+ GL2(R) is a lift of p of type , then there is a unique local homomorphism Ry —+ R such that p is equivalent to the pushforward of p3""”. Thus the equivalence classes of type 5 lifts to GL2(R) can be identified with Hom(Rz, R). The local ring Rr is called the universal deformation ring for representations of type 5. On the other hand Wiles constructs a candidate for a universal modular lifting of type ped : Gg — GLa(Tr). The ring Tx is constructed from the algebra of Hecke operators acting on a certain space of modular forms. The universal property of Rr gives a map Ry + Tz. The problem thus becomes: to show that this map is an isomorphism?. In fact, it can be shown to be a surjection without great dif- ficulty, and the real challenge is to prove injectivity, i.e., to show, in essence, that Rg is not larger than Ty. By an ingenious piece of commutative algebra, Wiles found a numerical criterion for this map to be an isomorphism, and for the ring Tx to be a local complete intersection. This numerical criterion seems to be very close to a special case of the Bloch-Kato conjecture [BK]. Wiles further showed (by com- bining arguments from Galois cohomology and from the theory of congruences between modular forms) that this numerical criterion was satisfied if the min- imal version Ty of this Hecke algebra (obtained by taking 5 = 0, ie., allowing the least possible amount of ramification in the deformations) was a complete intersection. Finally in [TW] it was proved that Tg is a complete intersection. 2 Maps of this kind were already considered in [Maz3] and (BMJ, and it is conjectured in [MT] that these maps are isomorphisms in certain cases, though not in exactly the situations considered by Wiles.14 Introduction Outline of the paper Chapter 1 recalls some basic notions from the classical theory of elliptic curves and modular forms, such as modular forms and modular curves over C and Q, Hecke operators and q-expansions, and Eichler-Shimura theory. The Shimura- Taniyama conjecture is stated precisely in section 1.8. Chapter 2 introduces the basic theory of representations of Gg. We describe Mazur’s deformation theory and begin our study of the universal deformation rings using techniques from Galois cohomology and from the theory of finite flat group schemes. We also recall some basic properties of elliptic curves, both to explain Frey’s argument precisely and illustrate the uses of é-adic represen- tations. Chapter 3 explains how to associate Galois representations to modular forms. We then describe what was known and conjectured about associating modular forms to Galois representations before Wiles’ work. After introducing the universal modular lifts of certain mod ¢ representations, we give the proof of Wiles’ main theorems, taking for granted certain results of a more technical nature that are proved in the last two chapters. Chapter 4 explains how to prove the necessary results concerning the struc- ture of Hecke algebras: the generalization by Taylor and Wiles of a result of de Shalit, and the generalization by Wiles of a result of Ribet. Chapter 5 establishes the fundamental results from commutative algebra discovered by Wiles, following modifications of the approach of Wiles and Taylor- Wiles proposed by Faltings and-Lenstra.1 Elliptic curves and modular forms 1.1 Elliptic curves We begin with a brief review of elliptic curves. A general reference for the results discussed in this section is [Sil] and [Si2} ‘Anelliptic curve E over a field F is a proper smooth curve over F of genus one with a distinguished F-rational point. If E/F is an elliptic curve and if w is a non-zero holomorphic differential on E/F then E can be realised in the projective plane by an equation (called a Weierstrass equation) of the form (W) ¥?Z+a.XYZ + a3¥Z? = X* + a.X?Z + ag XZ? + 062? such that the distinguished point is (0 : 1 : 0) (sometimes denoted oo because it corresponds to the “point at infinity” in the affine model obtained by setting Z=1) andw = 77 We also define the following quantities associated to (W): yas tras b= ah +40, b= 2aq tardy be = a2 + dag bg = ajag + 4aza5 — ayasaq + azas -aj Sbababe ~ bbs — 844 — 2705 (03 — 24b4)3/A. i One can check that the equation (W) defines an elliptic curve if and only if A is nonzero. One can also check that such equations define elliptic curves which are isomorphic over F if and only if they give the same quantity j. Thus j only depends on E so we will denote it jg. The quantity A depends only on the pair (E,w) so we shall denote it A(E,w). If u belongs to F* then ul? A(E,uw) = A(E,w)- An elliptic curve E/F has a natural structure of a commutative algebraic group with the distinguished F-rational point as the identity element. An algebraic map between two elliptic curves which sends the distinguished point of one to the distinguished point of the other is automatically a morphism of algebraic groups. A map between elliptic curves which has finite kernel (and hence, is generically surjective) is called an isogeny. Elliptic curves over C: If F = C, then the curve E is isomorphic as a complex analytic manifold to the complex torus C/A, where A is a lattice in C, ie., a discrete Z-submodule of C of rank 2. The group law on E(C) corresponds to the usual addition in C/A. In terms of A, an affine equation for E in A?(C) is given by y? = 42° + gor +93, n=-60 DF wl DO aeA={0} 7 AEA- {0} where In terms of this equation, the map from C/A to E(C) sends z to (x,y) = (p(z), 6'(z)), where g(z) is the Weierstrass -function associated to the lattice A. (CE. [Sil], ch. VL) The inverse map is given by integrating the holomorphic differential w, i.e., sending P € E(©) to the image of J, w in C/A, where 7 is any path on E(C) from oo to P, and A is the lattice of periods J, w, where 7 1516 1. Elliptic curves and modular forms ranges over the integral homology H(B(C),Z). Replacing w by wo changes A to uA, so that A is determined by E only up to homotheties. We scale A so that one of its Z-generatorsis 1, and another, 7, has strictly positive imaginary part. This gives the analytic isomorphism: E(©) ~C/(1,7). The complex number 7 in the complex upper half plane # is well defined, modulo the natural action of SL2(Z) on 1 by Mobius transformations. (Thus the set of isomorphism classes of elliptic curves over C can be identified with the quotient H/SL9(Z).) ‘The map z ++ e?" identifies C/(1, 7) with CX /q®, where q = e?"” is the multiplicative Tate period. The analytic isomorphism E(C) = C*/q* has the virtue of generalizing to the p-adic setting in certain cases, as we will see shortly. Note that |g| < 1. The invariant j can be expressed in terms of g by a convergent power series with integer coefficients: q) + 744 + 196884q+ -+-. (1.1.1) i The following basic facts are a direct consequence of the analytic theory: Proposition 1.1 The subgroup E{n\(C) of points of order n on E(C) is iso- morphic (non-canonically) to Z/nZ x Z/nZ. More generally, if F is any field of characteristic zero, the subgroup E{n|(F) is contained in Z/nZ x Z/nZ. Proof, The analytic theory shows that E(C) is isomorphic as an abstract group to a product of two circle groups, and the first statement follows. The second statement follows from the Lefschetz principle (cf. [Sil], ch. VI, §6). a Proposition 1.2 The endomorphism ring Endc(E) of an elliptic curve over C is isomorphic either to Z or to an order in a quadratic imaginary field. The same is true if one replaces C by any field of characteristic 0. Proof: An endomorphism of E(C) ~ C/A induces multiplication by complex number a on the tangent space. Hence End ¢() is isomorphic to the ring of @ € C satisfying aA C A. Such aring is isomorphic either to Z or to a quadratic imaginary order. The corresponding statement for fields of characteristic 0 follows as in the proof of proposition 1.1. o If End c(E) © Q is a quadratic imaginary field, we say that E has comple multiplication. Remark 1.3 It follows from the arithmetic theory of complex multiplication (cf. [Si2], ch. 1) that any elliptic curve Z with complex multiplication is defined over an abelian extension of the quadratic imaginary field K = Endc(E) @Q If E is defined over Q then K has class number one. There are only finitely many elliptic curves over Q with complex multiplication, up to “twists” (i.e., C-isomorphism).1, Elliptic curves and modular forms 17 Elliptic curves over Qp: Now suppose that £ is an elliptic curve defined over the p-adic field Q,. There is an equation (wm) -¥?Z 4a. XYZ + a3¥Z? = X? + a2X?Z + ag XZ? + 062? for E with the property a; € Z, for all i and |A| is minimal amongst all such equations for B. Although (W™") is not unique, the associated discriminant depends only on E and is denoted Ag", Moreover the reduction of (W™") modulo the uniformizer p defines a projective curve E, which is independent of the particular minimal equation chosen. If (W) is any equation for E with coefficients in Zp and with discriminant A, then A'p'* divides A. If B is a smooth curve we say that E has good reduction at p. If B has a unique singular point which is a node we say that E has multiplicative reduction at p. Otherwise E has a unique singular point which is a cusp and we say that E has additive reduction at p. If E has good or multiplicative reduction we say that it has semi-stable reduction at p, or simply that E is semi-stable. If (W) defines a smooth curve mod p then E has good reduction at p and (W) is a minimal equation. If A = 0 mod p but 63 4 24b, mod p, then modulo p the equation (W) defines a curve with a node. In this case E has multiplicative reduction at p and (W) is a minimal equation. Curves with good reduction: In that case p does not divide A®!", and the reduction E is an elliptic curve over Fp. If q is any power of p, and F, is the field with q elements, we define the integer N, to be the number of solutions to the equation (W™") in the projec- tive plane P?(F,). Thus N, is the order of the finite group E(F,). We define the integer a, by the formula ay =q+1-N,y. The integers a, are completely determined by ap: more precisely, we have (1 = app! + p'-#)-1 = 1+ app? + apap + apsp- 8 +--+. (1.1.2) We call the expression on the left the (local) L-function associated to E over Q,, and denote it by L(E/Qp,s). Concerning the size of ap we have the following fundamental result of Hasse, whose proof can be found in [Sil], ch. V, §1: Theorem 1.4 [ap| < 2Vp. A further division among curves of good reduction plays a significant role in our later discussion. We say that E has (good) ordinary reduction if p does not divide ap, and that it has supersingular reduction if p divides ap. When E has good reduction at p, we define its local conductor at p to be mp(E) = 0 Curves of multiplicative reduction: Elliptic curves over Q, which have multi- plicative reduction at p can be understood by using the p-adic analytic de- scription discovered by Tate. More precisely, we can formally invert the power18 1. Elliptic curves and modular forms series (1.1.1) expressing j in terms of g, to obtain to a power series for q in j having integer coefficients: 1} + 74457? + 750420j-$ + 87276963254 +--+. a= (1.1.3) If E has multiplicative reduction, then j € Q, is non-integral, and hence the power series (1.1.3) converges, yielding a well-defined value of q in pZy. This is called Tate's p-adic period associated to E over Q,. Note that we have Up(q) = —up(3) = vp(AB™). ‘We say that E has split (resp. non-split) multiplicative reduction at p if the two tangent lines to the node on E(F,) have slopes defined over F, (resp. F,2). Proposition 1.5 (Tate) There is a p-adic analytic isomorphism ®:Y/? EQ), which has the property that o(®(z)) = &(o2%), Vo € Ge,, where 6: Ga, — +1 is © the trivial character, if E has split multiplicative reduction; © the unique unramified quadratic character of Go,, if E has non-split mul- tiplicative reduction. The proof of this proposition is explained in [Si2], ch. V, for example. We define the L-function L(E/Qp, 8) to be _ J (—p-*)-! if B has split reduction, L(E/Q,8) = { (14p~*)-!__ if B has non-split reduction. (11.4) In both cases the conductor mp(B) is defined to be 1. Curves of additive reduction: If E has additive reduction at p, we simply define L(E/Q,8) = 1. (1.1.5) The conductor mp(E) is defined to be 2, if p > 3. When p = 2 or 3, it is determined by a somewhat more complicated recipe, given in [Ta]. Elliptic curves over Q: Let £ be an elliptic curve defined over Q. In partic- ular E may be viewed as a curve over Qy for every p, and we define its (global) conductor by Ne =] p™). ? The curve E is said to be semi-stable if it is semi-stable over all p-adic fields Q,. Note that E is semi-stable if and only if its conductor Ng is square-free. Using the fact that Q has class number 1, one can show that E has a global minimal Weierstrass model (W™") which gives the equation of a minimal Weierstrass model over each Q,. The associated discriminant, denoted A",1. Elliptic curves and modular forms 19 depends only on E. The associated differential, denoted wi*", is called the Néron differential. It is well-defined up to sign. The following, known as the Mordell-Weil theorem, is the fundamental re- sult about the structure of the group of rational points E(Q). (Cf. For example (sil].) Theorem 1.6 The group E(Q) és a finitely generated abelian group. Hence EQ ~Tez’, where T is the (finite) torsion subgroup of E(Q), andr > 0 is the rank of E over Q. Concerning the possible structure of T, there is the following deep result of Mazur, a variant of which also plays a crucial role in the proof of Fermat’s Last Theorem: Theorem 1.7 If E/Q is an elliptic curve, then its torsion subgroup is isomor- phic to one of the following possibilities: Z/nB, 1 3/2. Conjecture 1.9 (Birch-Swinnerton-Dyer) The L-function L(E/Q,s) has an analytic continuation to the entire complez plane,and in particular is ana- lytic at s = 1. Furthermore: ordy=1L(E/Q, s) = r. There is also a more precise form of this conjecture, which expresses the leading coefficient of L(E/Q,s) at s = 1 in terms of certain arithmetic invariants of E/Q. For more details, see [Si], conj. 16.5. As we will explain in more detail in section 1.8, the analytic continuation of L(E/Q, 8) now follows from the work of Wiles and Taylor-Wiles and a strength- ening by Diamond [Di2], for a very large class of elliptic curves over Q, which includes all the semi-stable ones.20 1. Elliptic curves and modular forms Abelian varieties: Elliptic curves admit higher-dimensional analogues, called abelian varieties, which also play a role in our discussion. Analytically, the set of complex points on an abelian variety is isomorphic to a quotient C?/A, where Ais a lattice in C? of rank 2g, satisfying the so-called Riemann period relations. A good introduction to the basic theory of abelian varieties can be found in {CS} and (Wel). 1.2 Modular curves and modular forms over C Modular curves: The group SL2(Z) of two by two integer matrices of deter- minant one acts by fractional linear (Mébius) transformations on the complex upper half plane H = {z€C | Im(z) > 0}, equipped with its standard complex analytic structure. The principal congru- ence group TN) of level N is the subgroup of matrices in $L2(Z) which reduce to the identity matrix modulo the positive integer N. A subgroup I’ of SL2(Z) is called a congruence group ifit contains ['(N) for some N. The level of T is the smallest N for which this is true. The most important examples of congruence groups are: © The group I'(N) consisting of all matrices that reduce modulo N to an upper triangular matrix. © The group P(N) consisting of all matrices that reduce modulo N to a . 1+ ; matrix of the form ( oi) © The principal congruence group I'(N) of level NV consisting of all matrices that reduce modulo N to the identity. Notice the natural inclusions of normal subgroups [(N) C 11 (N) C To(N). ‘The quotient [9(N)/T1 (IN) is canonically isomorphic to (Z/NZ)* via ( a) r dmod W. cd For any subgroup H of (Z/NZ)*, we denote by P'4(N) the group of matrices in Po(N) whose image in T9(N)/T1(N) belongs to H. If Tis a congruence subgroup of SZ2(Z), define Yp to be the quotient of the upper half plane H by the action of I. One equips Yp with the analytic structure coming from the projection map : H —+ Yr. (More precisely, if y = (1), and G, CT’ is the stabilizer of r in IL, then the local ring Oy,,y is identified with the local ring of germs of holomorphic functions at t which are invariant under the action of G;.) This makes Yr into a connected complex analytic manifold of dimension one, i.e., a Riemann surface. If I’ is To(NV) (resp. T\(N), or P(N), we will also denote ¥p by Yo(N) (resp. ¥i(N), or ¥(V)). One compactifies ¥r by adjoining a finite set of cusps which correspond to orbits of P!(Q) = QU {oo} under [. Call Xp the corresponding compact Riemann surface. (For more details, notably on the definition of the analytic structure on Xr at the cusps, see for example [Kn], p. 311, or (Shi2], ch. 1.) It follows from the definition of this analytic structure that the field Kr of meromorphic functions on Xp is equal to the set of meromorphic functions on # satisfying1. Elliptic curves and modular forms 21 (Transformation property): f(yr) = f(r), for all y €T; « (Behaviour at the cusps): For all y € SL2(Z), the function f(r) has a Puiseux series expansion 7°, anq™/* in fractional powers of q = e277. Riemann’s existence theorem (cf. for example [For], ch. 2) asserts that the analytic structure on Xp comes from an algebraic one, i.e., the field Kr is a finitely generated extension of C of transcendence degree 1. Thus we can, and will, view Xp as a complex algebraic curve over C. If T is To(N) (resp. T1(N), or [(N)), we will also denote Xp by Xo(N) (resp. X1(), or X(N). Examples and exercises: 1 For N = 1, the curve Xo(N) = X(N) = X(N) is a curve of genus 0, and its field of functions is the ring C(j), where j is the classical modular function, G(r) = 7) +744 + 196884q +--+, enir, (CF., for example, [Se4], ch. 7.) 2. Fort € HUP!(Q) = H, define G, to be the stabilizer of tin PSL2(Z), and let e, = #(G,/(G, NT). Show that e, depends only on the I-orbit of 7 in H, and that e, = 1 for all but finitely many 7 in H/T. Using the Riemann-Hurwitz formula (cf. [Ki], sec. 4.3) show that the genus of Xr is given by 1 gf) =1-[PSLo(B): 1] + 3 x (er - 1). ren/T Use this to compute the genus of Xo(p), X1(p), and X(p) for p prime. For details, see [Shi2], sec. 1.6 or [Ogg]. 3. For P =1(2), show that Xp is isomorphic to P!, and that Yp is isomorphic to P! — {0,1, 00}. Show that I'/(+1) is the free group on the two generators n= (§2) a = ($9). 4. Define a homomorphism T(2) —+ Z/nZ x Z/nZ, by sending g: to (1,0) and g to (0,1), and let T denote its kernel. Show that I’ is not in general a congruence subgroup and that the curve Yr == H/T is birationally isomorphic to the Fermat curve of degree n with affine equation 2* + y" = 1. Moduli interpretations: The points in Yr = H/T can be interpreted as elliptic curves over C with some extra “level N” structure. More precisely, IfT =[o(N), then the [orbit of r € 11 corresponds to the complex torus E =C/(1,7) with the distinguished cyclic subgroup of order N generated by 4. Thus, points on Yo(V) parametrize isomorphism classes of pairs (E,C) where B is an elliptic curve over C and C is a cyclic subgroup of E of order N. If = 1,(N), then the [-orbit of 7 corresponds to the complex torus E =C/(1,7) with the distinguished point of order N given by 3. Hence, points on ¥i(N) parametrize isomorphism classes of pairs (E, P) where now P is a point on E of exact order N.22 1. Elliptic curves and modular forms Remark 1.10 One checks that the above rules set up a bijection between points on Yp and elliptic curves with the appropriate structures, and that the projection ¥(N) —+ Yo(N) sending Py(N)z to T'o(N)r becomes the “forget- ful” map sending (B, P) to (E, (P)). Remark 1.11 (This remark will be used in section 1.3 when discussing Hecke operators.) Define an n-isogeny of [structures to be an n-isogeny of the un- derlying elliptic curves which sends one [structure to the other. If p is a prime not dividing N, then there are exactly p + 1 distinct p-isogenies from (C/(r,1), 37), whose images are the pairs: (( 1), x) @=6,...,.7-D, (Chon), 2). If p divides N, then there are only p distinct p-isogenies from (C/(r, 1), #7); since (C/(pr,1), 8) is not a Ty(N)-structure (the point p/N not being of exact order N on the complex torus C/(pr, 1)). Modular forms: Let k be an even positive integer. A modular form of weight k on T’is a holomorphic function f on 2 satisfying: « (Transformation property): f(77) = (er-+@)* f(r), for all y= (25) € Tr. (Behaviour at the cusps): For all.y € PSL2(Z), the function (cr + d)-* f (yr) has a Puiseux series expansion )7>° ang” in fractional pow- ers of g = e?"7. We call > anq"/" the Fourier expansion of f at the cusp 77" (i00). A modular form which satisfies the stronger property that the constant coefficient of its Fourier expansion at each cusp vanishes is called a cusp form. We denote by M,(I) the complex vector space of modular forms of weight k on I, and by S;,(P) the space of cusp forms on I’. (For examples, see [DI], sec. 2.2 and the references therein, especially, [Shi2], ch. 2.) This article is mainly concerned with modular forms of weight 2, and hence we will focus our attention on these from now on. A pleasant feature of the case k = 2 is that the cusp forms in S2(I) admit a direct geometric interpretation as holomorphic differentials on the curve Xp. Lemma 1.12 The map f(r) ++ wy = 2nif(r)dr is an isomorphism between the space S2(P) and the space (Xr) of holomorphic differentials on the curve Xp. Sketch of proof: One checks that the transformation property satisfied by f(r) under I’ causes the expression f(r)dr to be T-invariant, and that the condition of vanishing at the cusps translates into holomorphicity of f(r)dr. (Note, for example, that 2nidr = dq/q, s0 that wy is holomorphic at ico precisely when £(q) vanishes at q = 0.) a As a corollary, we find: Corollary 1.13 The space S(T) is finite-dimensional, and its dimension is equal to the genus 9 of Xr.1. Elliptic curves and modular forms 23 Proof: This follows directly from the Riemann-Roch theorem, cf. [Ki], sec. 6.3. Oo To narrow still further the focus of our interest, we will be mostly concerned with the cases ! = [9(N) and I; (N). A slightly more general framework is sometimes convenient, so we suppose from now on that I’ satisfies Ti(N) CP cI o(N). Such a group [is necessarily of the form I',(N) for some subgroup H of (Z/NZ)*. Because the transformation + ++ + + 1 belongs to I the forms in S2(P) are periodic functions on H of period 1, and hence their Fourier expansions at oo are of the form Slr) = Yang”, g =e? an EC. 50 The Petersson inner product: The spaces S2(P) are also equipped with a natural Hermitian inner product given by H.0)= sea ff wenay= [seater where 7 = 2 + iy. This is called the Petersson inner product. The diamond operators: Suppose now that = y(.V) and let d be an ele- ment of (Z/NZ)*. The map (d) which sends an elliptic curve with -structure (E, P) to the pair (E, dP) gives an automorphism of Yr which extends to Xr. It is called the diamond operator. For r in H and 7 = (2 ’) in To(N), we have (@)(Pr) = Py). Hence (d) acts on S2(L), identified with the holomorphic differentials on Xr, by the rule (a) f(r) = (or + ay*y (E42), er+d. In geometric terms, the diamond operators are the Galois automorphisms of the natural (branched) covering X1(N) — Xo(N) whose Galois group is isomor- phic to P(N)/(41y(N)) = (Z/NZ)*/(+1). Given an even Dirichlet character x:(Z/NZ)* —+ C¥, say that f is a modular form of level NV and character x if it belongs to the x-eigenspace in S(I'y (NN)) under this action. Let Sp(N, x) denote the space of all such forms, Thus a function f in S(N, x) is a cusp form on I’; (N) which satisfies the stronger transformation property: ar+b\ _ 2 ab f (4) = x(d)(er + d)?f(r), for all (2 ‘) €To(N). Note that if 1 is the trivial character, then S2(N, 1) is canonically identified with S)([o(N)), which we will also denote by S2(N). Finally note the direct sum decomposition: Se(T1(N)) = @yS2(N,x), where the sum ranges over all the even Dirichlet characters modulo N.24 1. Elliptic curves and modular forms Exercise 1.14 Show that if f(r) belongs to S2(N), then f(ar) belongs to S2(mN), for each integer a dividing m. Jacobians of modular curves: Let V be the dual space V = S(T)” = Hom(S2(T),©). It is a complex vector space of dimension g = genus(Xr). The integral ho- mology A = H;(Xr,Z) maps naturally to V by sending a homology cycle c to the functional ¢, defined by .(f) = f.wy. The image of A is a lattice in V, i.e., a Z-module of rank 2g which is discrete (cf. [Mul], cor. 3.8). Fix a base point 7 € 1, and define the Abel-Jacobi map ©) : Xr(C) —+ V/A by 43(P)(f) = J wy. Note that this is well-defined, i.e., it does not depend on the choice of path on Xr from 7 to P, up to elements in A. We extend the map ®aj by linearity to the group Div (Xr) of divisors on Xr, and observe that the restriction of 43 to the group Div °(Xr) of degree 0 divisors does not depend on the choice of base-point 7. Moreover we have the Abel-Jacobi theorem: Theorem 1.15 The map ay: Div°(Xp) 4 V/A has a kernel consisting precisely of the group P(Xr) of principal divisors on Xp. Hence &q3 induces an isomorphism from Pic (Xr) := Div °(Xr)/P(Xr) toV/A. For the proof, see [Mul], ch. 3. The quotient V/A is a complex torus, and is equal to the group of complex points of an abelian variety. We denote this abelian variety by Jr, the Jacobian variety of Xp over C. If T = To(N) or T,(N), we will also write Jo(N) or J,(N) respectively for the Jacobian Jr. 1.3 Hecke operators and Hecke theory ‘We maintain our running assumption that I satisfies Ti(N) CP cTo(N). If p is a prime not dividing the level N, we define the Hecke operator Tp on So(P) by the formula m0) = 204 (74) + meorgtor = We give a more conceptual description of T in terms of remark 1.11 in the case T =T(N). We have wns) = Do di(ws)s where $i(r) = 4H, and $oo(t) = (p)pr represent the p +1 curves with - structure that are images of (C/(r,1), 2) by a p-isogeny, and the f are the1. Elliptic curves and modular forms 25 pull-back maps on differential forms on H. (An isogeny of elliptic curves with T-structure is simply an isogeny between the underlying curves which sends one structure to the other.) Such a description makes it evident that T>(f) belongs to S(I'), if f does. In terms of the Fourier expansion of f = ang", the formula for the operator Tp on S2(N, x) is given by: To(f) = So ang”? + px(p) Yang?" pin If p divides N, then we define the Hecke operator Up analogously, by summing again over all the cyclic p-isogenies of I-structures. Since there are only p of them, the formula becomes simpler: a wtf) = 3 04 (4) = Dona". = pin The reader is invited to check that the Hecke operators of the form Tp or U, commute with each other, and also that they commute with the diamond operators introduced in the previous section. We extend the definition of the Hecke operators to operators Tyn, with n> 1, by the inductive formulae Tpnat = TpTpn — (p)pTpn-1, if (p,N) = 1, and Ty = Uf otherwise. We then define the operator T,,, where n = [] pf is written as a product of powers of distinct primes p;, by Tr. = [] Tes This definition makes the Hecke operators multiplicative, i.e., TmTn = Tmn if (m,n) = 1. (A more conceptual definition of the Hecke operator Ty is that Tn(f) is obtained by summing the pullback of wy over the maps describing all the cyclic n-isogenies of I-structures.) The relations among the different Hecke operators can be stated succinctly by saying that they obey the following (formal) identity: ~ Typ a ~ Us Tl @- tee + "TT - op) pI ain The reader can consult [DI], sec. 3 and the references therein (especially [Shi2}, ch. 3 or {Kn}) for more details and different points of view on Hecke opera- tors. Let T be the subring of Ende(S,(F)) generated over C by all the Hecke operators Tp for p{N, U, for qlN, and (d) acting on S2(T'). =otan-*. (1.3.1) Definition 1.16 A modular form f is an eigenform if it is a simultaneous eigenvector for all the Hecke operators in T, i.e., if there exists a C-algebra homomorphism 4: T —+ C such that Tf = X(T), for all T € T. A direct calculation shows that the coefficients a, of an eigenform f can be recovered from the homomorphism X by the formula: an(f) = a1(f)A(Tn)-26 1. Elliptic curves and modular forms It follows that the first Fourier coefficient a; of a non-zero eigenform is always non-zero, and that the non-trivial eigenspaces for T are all one-dimensional: Proposition 1.17 Given a non-zero algebra homomorphism \ : T —+ C, there is ezactly one eigenform f up to scaling, which satisfies Tf = X(T) f, for all T €T. Sketch of proof. The proof of the existence of f is an exercise in commutative algebra (localize S2(I) at the kernel of ), and the uniqueness is clear from the formula above. o We call an eigenform satisfying a, = 1 a normalized eigenform. Atkin-Lehner theory: It is natural to ask whether Sp(P) can be decomposed into a basis consisting of distinct normalized eigenforms. Unfortunately, this is not always possible, as the following exercise illustrates. Exercise 1.18 Suppose that p* divides N exactly. Let T’ be the algebra of Hecke operators (generated by the operators Ty with q{N/p®, and U, with q\N/p®) acting on S2(N/p*). Let f = T®, ang” be a T’-eigenform of level N/p® in S2(N/p*). Show that the space Sy spanned by the forms f(r), f(p7), f(p*r), and f(p37) is contained in S2(N), and is stable for the action of the Hecke operators Ty, q/N, and U;, q|V. Show that S has no basis of simulta neous eigenforms for the Hecke algebra T of level N, so that the action of T on S; is not semi-simple. Let T° denote the subalgebra of T generated only by the “good” Hecke opera- tors T', with q{N, and (d) Proposition 1.19 If q does not divide N, the adjoint of the Hecke operator T, with respect to the Petersson scalar product is the operator (q)~!T,, and the adjoint of (q) is (q)~*. In particular, the Hecke operators commute with their adjoints. Proof: See [Kn], th. 9.18 and 8.22, or [Ogg]. o Proposition 1.19 implies, by the spectral theorem for commuting operators that commute with their adjoints: Proposition 1.20 The algebra T° is semi-simple (i.e, it is isomorphic to a product C x -++ x C of a certain number of copies of C), and there is a basis of So(L) consisting of simultaneous eigenvectors for the operators Ty. Thus, T° has the merit of being semi-simple, while T is not in general. The cost of replacing T by T°, however, is that one loses “multiplicity one”, i.e., the eigenspaces for T° need not be one-dimensional. For example, the space Sy defined in the previous exercise is contained in a single eigenspace for T° ‘The theory of Atkin-Lehner [AL] gives essentially a complete understanding of the structure of the algebra T, and the structure of the space of eigenforms. To motivate the main result, observe that the problem in the exercise above seems to be caused by forms of level NV that are coming from forms of lower level N/p? by astraightforward operation, and are therefore not “genuinely” of level N. They are the analogues, in the context of modular forms, of non-primitive Dirichlet characters.1. Elliptic curves and modular forms 7 Definition 1.21 We define the old subspace of Sp(I) to be the space spanned by those functions which are of the form g(az), where g is in S2(P'1(M)) for some M < N and aM divides N. We define the new subspace of S2(F) to be the orthogonal complement of the old subspace with respect to the Petersson scalar product. A normalized eigenform in the new subspace is called a newform of level N. The following result is the main consequence of the theory of Atkin-Lehner. It gives a complete answer to the question of what is the structure of the algebra T acting on S2(P). Theorem 1.22 If f is in the new subspace of S:(P) and is an eigenvector for all the operators in T°, then it is also an eigenform for T, and hence is unique up to scaling. More generally, if f is a newform of level N;|N, then the space Sy defined by Sy = {9 € S2(L) such that Tg = A;(T)g, for all T € T°} is stable under the action of all the Hecke operators in T. It is spanned by the linearly independent forms f(az) where a ranges over the divisors of N/Ny. Furthermore, we have S(T) = @5y, where the sum is taken over all neuforms f of some level Ny dividing N. See [AL] for the proof in the case I’ = ['9(WV), and (La2], ch. VIII for the general case, (See also [DI], sec. 6 for an overview.) Exercise 1.23 Consider the case where [ = ['9(22). Show that Xo(22) is of genus 2, and hence that $2(22) has dimension 2. Show that S(22) is equal to Sy, where f = (n(r)n(117))? is a newform of level 11, so that in particular there are no newforms of level 22 on I’. Show that T° is isomorphic to C in this case, and that T is isomorphic to the semisimple algebra C x C. Action on homology and Jacobians: Note that the Hecke operators act on V =S,(L)Y by duality. One checks (cf. [Kn], props. 11.23, 11.24) that the sublattice A of V is stable under the action of all the Hecke operators T,, and of the diamond operators (d). Therefore the operators T;, and (d) give rise to endomorphisms of the torus V/A, and hence the Jacobian variety Jr, in a natural way. The involution r ++ —7 gives rise to an involution on Xp(C) (which is complex conjugation on the model of Xr over R deduced from the @ model defined in section 1.5). Since complex conjugation is continuous it also acts on the integral homology A = Hi(Xr(C),Z). Let A* and Aq be the sublattices of A on which complex conjugation acts by +1 and —1. These are sublattices of A of rank g which are stable under the Hecke operators, since complex conjugation commutes with the Hecke action. A more algebraic description of the action of T, on Jr is given via the notion of an algebraic correspondence. A correspondence on a curve X is a divisor C on X x X which is taken modulo divisors of the form {P} x X and X x {Q}. Let 71 and 7 denote the projections of X x X onto each factor. Then the correspondence C' induces a map on divisors of X, by setting C(D) = m(x;"(D)-C).28 1, Elliptic curves and modular forms (For the definition of the intersection D, - D2 of two divisors, see [Wel].) The map C preserves divisors of degree 0, and sends principal divisors on X to prin- cipal divisors. It gives a well defined algebraic endomorphism of the Jacobian variety Jac(X). Given a correspondence C, its transpose CY is defined to be the divisor of X x X obtained by interchanging the two factors of X x X. One can define a natural notion of composition for correspondences, and the set of correspondences forms a ring. The general theory of correspondences and the proofs of the above facts are given in [Wel], particularly the second chapter. The Hecke correspondence Ty is defined as the closure in Xp x Xr of the locus of points (A, B) in Y¥p x Yp such that there is a degree n cyclic isogeny of elliptic curves with I-structure from A to B. For example, if p is a prime not dividing N, then Tp is an algebraic curve in X(N) x X1(N) which is birational to Xr,(w)ro(p)- The induced map on divisors in this case satisfies T,((E, P)) = }-(E/C, P mod C) where the sum runs over the subgroups C’ of E having order p. Note also that if (A, B) belongs to Tp, then the isogeny dual to A —> B gives a p-isogeny from B to (p)A, so that TY = (py "Tp. 1.4 The L-function associated to a cusp form For this section, let f in S2(T,(N)) be a cusp form with Fourier expansion at ico given by 7, @nq". One has the following estimate for the size of the Fourier coefficients an: Theorem 1.24 The coefficients an € C satisfy the inequality lanl < e(f)o0(n) va, where c(f) is a constant depending only on f, and oo(n) denotes the number of positive divisors of n. Sketch of proof: This follows from proposition 1.51 of section 1.7 which relates the p-th Fourier coefficients of eigenforms, for p a prime not dividing the level of T, to the number of points on certain abelian varieties over the finite field F,. The estimates of Hasse and Weil for the number of points on abelian varieties over finite fields (stated in theorem 1.4 of section 1.1 for the special case of elliptic curves; see [Wel], §IV for the general case) thus translate into asymptotic bounds for the Fourier coefficients of these eigenforms. We note that the cruder estimate |ay| = O(n), which is enough for the purposes of this section, can be derived by a more elementary, purely analytic argument; cf. [Ogg], ch. IV, prop. 16. o ‘The L-function associated to f is defined by the formula: L(f,s) = So ann”*. As in exercise 1.8, one can show that the infinite sum defining L(f, s) converges absolutely in the right half-plane Re(s) > $. A much better insight is gained into the function L(f,s) by noting that it is essentially the Mellin transform of1, Elliptic curves and modular forms 29 the modular form f. More precisely, if we set A(f,s) = N*/?(2n)-*I'(s)L(f,s), then we have Athos) =e [ slinhytayly (1.43) Exercise 1.25 Check the formula above. This integral representation for L(f,s) gives the analytic continuation of L(f,s) to the entire complex plane. The modular invariance of f translates into a functional equation for L(f,s): more precisely, let wy be the Atkin- Lehner involution defined by wn(r) = —1/Nr. The reader may check that wy induces an involution of Xr, and hence of 2'(Xp) = S9(T). One finds that L(f, 8) satisfies the functional equation: A(f, 8) = —A(ww(f),2- 8). For a proof of this, see [Ogg], ch. V, lemma 1. Eigenforms for T in S2(N, x) have a great importance in the theory because their associated L-functions have an Euler product expansion, in addition to an analytic continuation and functional equation: Theorem 1.26 If f = Dang” is a normalized eigenform in S2(N,x) for all the Hecke operators, then the L- Se L(f,s) = Sann* has the Euler product expansion [[ G - epp7* + x(p)p'-**)“! T] 1 - app")? pin iN Proof: This follows directly from equation (1.3.1) of section 1.3. a If f is a newform of level N, then it is also an eigenform for wy, so that the functional equation may be viewed as relating L(f,s) and L(f,2—-s). We can also state the following more precise version of theorem 1.24 (see lemma 3.2 of [Hi2] for example for parts (b), (c) and (d)). ‘Theorem 1.27 Suppose that f is a newform of level Ny and let Ny denote the conductor of its character x. (a) If p does not divide Ny then lap| < 2B. (b) If p\\Ny and p does not divide Ny then a2 = xo(p) where xo is the prim- itive character associated to x. . (c) If p divides Ny and p does not divide Ny/Ny then \ap| = VB. (d) If p? divides Ny and p divides Ns/Ny then ay = 0. 1.5 Modular curves and modular forms over Q Modular curves: For I’ between [(N) and I(N), the modular curve Xr has a model over Q. We describe such a model in the case of I = ['9(N); the construction for general [’ follows from similar considerations.30 1. Elliptic curves and modular forms The key remark here is that, as was noted in section 1.2, the complex points on the curve Yo(N) have a natural interpretation as moduli of elliptic curves together with a cyclic subgroup of order N, given by sending the point 7 € H/To(N) to the pair (C/(1,7), (1/N)). Consider the “universal elliptic curve” 3 1728" ~ j- 1728" Ej:y+2y= It is an elliptic curve over the function field Q(j), with j-invariant j. Let d be the order of P!(Z/NZ), and let C1,... ,Ca denote the set of all cyclic subgroups of B; of order N, defined over Q(j), an algebraic closure of Q(j). Fix one of these subgroups, C. The Galois group Gal (Q(j)/Q(J)) permutes the C; in a natural way. Let Fy be the smallest extension of Q{j) (viewed as embedded in Q%j)) with the property that o(C) = C, for all o € Gal (Q(j)/Fw). It can be seen (cf. (Shi2], thm. 6.6) that the Galois action on the C; is transitive so that Fy /Q(j) is of degree d, and that it is a regular extension, i.e., Fy NQ = Q Geometrically, Fy can be viewed as the function field of a curve X g over Q, with the inclusion of Q(j) corresponding to a map from X7g to the projective “jeline” over Q. The pair (E},C) is an elliptic curve over Q(j) with a subgroup of order N defined over Fy. Using (E;,C), each complex point of X gives an elliptic curve over C with a subgroup of order N, provided the point does not lie over j = 0, 1728 or co. The resulting map to Xo(N) extends to an isomorphism from X to Xo(N). The curve X thus constructed, together with this identification, is the desired model of X(N) over Q. More concretely, the functions j = j(r) and jy = j(N7) are related by a polynomial equation $y(j,jn) = 0 with coefficients in Q, of bidegree d. The field Fy is the function field of the affine curve &y(X,Y) = 0, and the mapping 7 ++ (j(7),j(Nz)) gives a birational equivalence between 1/To(N) and the complex curve defined by the equation $y. In practice it is not feasible to write down the polynomial $y, except for certain very small values of N. To study the models over Q of Xo(N), more indirect methods are needed, which rely crucially on the moduli interpretation of Xo(V). Similar remarks hold for X(N). Models over Z: The work of Igusa {Ig}, Deligne-Rapoport [DR], Drinfeld [Dr], and Katz-Mazur [KM] uses the moduli-theoretic interpretation to describe a canonical proper model for Xr over Spec Z. These models allow us to talk about the reduction of Xr over finite fields F,, for p prime. The curve has good reduction at primes p not dividing N, with the “non-cuspidal” points of Xp r, corresponding to elliptic curves over I, with [-structure. The singular fibers at primes p dividing N can also be described precisely; an important special case (see [DR)) is that of P'o(N) with p exactly dividing N. For further discussion and references, see [DI], sec. 8, 9. From now on, when we write Xr, Xo(N), or X;(NV), we will mean the curve over Q which are the models described above for the complex curves defined in section 1.2. Remark 1.28 When considering q-expansions, it is more convenient to use a different set of models over Z for these complex curves. We define X#(N) in the case of P(N) as a model over Z which parametrizes pairs (E,i) where é1. Elliptic curves and modular forms 31 is an embedding of jw in the (generalized) elliptic curve E. (So X#(N) is the model denoted X,,(V) in [DI], sec. 9.3, assuming N > 4.) For I’ between To(N) and P(N), we define X/ as the corresponding quotient of X(N). This model has good reduction at primes p not dividing N, but unlike the models mentioned above, its fibers at primes p dividing N are smooth and irreducible, but not proper. In the case of ! = ['9(N), the curve Xf-q can be identified with Xo(N). However, this is not the case in general: the cusp 0 is a rational point of X#q but not necessarily of Xr. Jacobians: Weil’s theory [Wel] of the Jacobian shows that the Jacobians Jr defined in section 1.2 as complex tori also admit models over @ When we speak of Jp, Jo(N) and J,(N) from now on, we will refer to these as abelian varieties defined over Q. Thus, the points in Jr(K), for any Qalgebra K, are identified with the divisor classes on Xr of degree 0, defined over K’. We let Jryz, denote the Néron model of the Jacobian Jr over Spec (Z). Using this model we define Jp/4 for arbitrary rings A. In particular we can consider Jp, , the reduction of the Jacobian in characteristic p, which is closely related to the reduction of the integral model of the curve Xr mentioned above. In particular, if p does not divide the level of’, then Jp/p, can be identified with the Jacobian of Xr/p,. For a treatment of the case Pr = [9(N) with p exactly dividing N, see the appendix of [Maz]; for more general discussion and references, see [DI], sec. 10, especially sec. 10.3. Hecke operators: The Hecke operators have a natural moduli interpretation, which was already touched upon in section 1.3. In particular, one finds that the operator T, arises from a correspondence which is defined over Q, and gives rise to an endomorphism of the Jacobian Jr which is defined over Q. This in turn gives rise to an endomorphism of the Néron model Jr /g, and we can then con- sider the endomorphism T,, on the reduction of the Jacobian in characteristic p. Recall that if p is a prime not dividing V, we may identify this reduction with the Jacobian of X/p,. (Cf. [MW], ch. 2, sec. 1, prop. 2.) Furthermore, one can show that the moduli-theoretic interpretation of the Hecke operator remains valid in characteristic p; i.e. the endomorphism T,, of Jr/e, is induced by a map on divisors satisfying, for all ordinary elliptic curves A with P-structure: T(A) = Dil A), where the sum is taken over all cyclic isogenies of degree n. (See [DI], sec. 10.2 for further discussion and references.) This description allows one to analyse, for example, the Hecke operator T, over Fp, when (p,N) = 1. Let us work with T = Ty(N), to illustrate the idea. For a variety X over f,, let ¢x be the Frobenius morphism on X defined by raising coordinates to the pth power. Thus if (E, P) corresponds to a point of X1(N)p,, then dg is an isogeny of degree p from (B, P) to the pair (Beor Poo) = $x,(n)(E,P). The graph of $x, qn) in (Xi(V) x Xa(N))/e, is a correspondence of degree p, which we call F. Let J’ be the transpose of this correspondence. The endomorphism F of Jr induced by F is the Frobenius endomorphism ¢,, and the endomorphism F” is the dual endomorphism (in the sense of duality of abelian varieties). Now consider the divisor F'((E, P)) = (Eas Pi) ++ + (Eps Pp)32 1. Elliptic curves and modular forms where the (Ej, P;) are elliptic curves with I-structure in characteristic p. Since 6, is an isogeny of degree p from (Ej, P;) to (E,P), we also have the dual isogeny from (E, P) to (Ei,pP;). If Eis ordinary at p, then the points (Exo, Poo); (E1,pP1); ..., (EpsPPp) are a complete list of the distinct curves with I- structure which are p-isogenous to (E, P). Hence one has the equality of divi- sors on X;(N) e,: T,((E, P)) = (Eoos Poo) + (Ei, pP1) + +++ + (Ep, PPp) = (F + (p)F")((E, P)). Since the ordinary points are dense on X;(N),g,, we deduce that Tp = (F + (p)F') as endomorphisms of J; (N) p,. This equation, known as Eichler-Shimura congruence relation, plays a central role in the theory. (For more details, see [DI], sec. 10.2, 10.3.) Theorem 1.29 If p{N then the endomorphism T» of Jrje, satisfies Tp = F + (p)F". Remark 1.30 This was proved by Eichler [Ei] to hold for all but finitely many p in the case of (NV), and by Shimura ([Shil], see also (Shi2], ch. 7) in the case of [,(N). The fact that it holds for all p not dividing N follows from work of Igusa [Ig]. Modular forms: In the same way that modular curves have models over Q and over Z, the Fourier coefficients of modular forms also have natural rationality and integrality properties. We start by sketching the proof of: Theorem 1.31 The space S2(I) has a basis consisting of modular forms with integer Fourier coefficients. Proof: The Hecke operators act on the integral homology A* in a way that is compatible with the action on S(T) and respects the natural (Poincaré) duality between these two spaces. Hence, if {An}nen is a system of eigenvalues for the Ty, then the Aq are algebraic integers in some finite extension K of Q, and the system {A%}nen is a system of eigenvalues for the T;, for any Galois automorphism o of Q/Q. Hence, we have shown: Proposition 1.32 If f € S:(M,x) is a newform of some level M dividing N, then its Fourier coefficients lie in a finite extension K of Q. Moreover, if 0 € Gal(Q/Q) is any Galois automorphism, then the Fourier series {7 obtained by applying o to the Fourier coefficients is a newform in Sp(M, x°). The explicit description of S2(I) given in section 1.3 implies that S,(P) is spanned by forms having Fourier coefficients which are algebraic integers in some finite (Galois) extension K of Q, and that the space of forms with Fourier coefficients in K is stable under the natural action of Gal (K/Q) on Fourier expansions. An application of Hilbert’s theorem 90 shows that S2([) has a basis consisting of forms with rational Fourier expansions, and the integral- ity of the Fourier coefficients of eigenforms yields the integrality statement of theorem 1.31. a1. Elliptic curves and modular forms . 33 We define S(T, Z) to be the space of modular forms with integral Fourier coefficients in S2(P). Theorem 1.31 states that S(I',Z) @C = S2(P). Given any ring A, we define S_(P, A) = So(P,Z) ® A, and define Sp(N, A) and S2(N, x, A) (where x now is a character with values in A*) in the obvious way. If A is contained in C, the g-expansion principle below allows us to identify S2(I', A) with the set of modular forms in S2(T) with Fourier coefficients in A. The q-expansion principle: Because the modular curve X(N) has a model over Q, the space of modular forms So(N) = 9"(Xo(.V)) has a natural rational structure, given by considering the differential forms on Xo(N) defined over Q. ‘The fundamental g-expansion principle (see [DR], ch. 7, or [Kat], sec. 1.6) says that these algebraic structures are the same as those obtained analytically by considering q-expansions at oo. More generally, using the model X#, we obtain the q-expansion principle over Z for cusp forms on [ (cf. [DI], sec. 12.3). Theorem 1.33 The map S2(I) —> 9'(Xr) defined by f +4 wy induces an isomorphism from Sp(T,Z) to9'(Xf). Furthermore, if A is flat over Z or N is invertible in A, then the induced map S(L, A) > 01(Xf 4) is an isomorphism. Furthermore, if A is a subring of C, then this isomorphism identifies Sp(T, A) with set of modular forms in S2(P) having coefficients in A. 1.6 The Hecke algebra It follows directly from the formulas for the Hecke operators acting on q- expansions that the T, leave So(I'9(N),Z) stable, as well as the subspace of So(N,x) with coefficients in Z[x]. Using the q-expansion principle (the- orem 1.33), one can also show ({DI], sec. 12.4) that the diamond operators preserve the spaces of cusp forms on I’ with integral Fourier expansions, and hence that the space S2(P, Z) is preserved by all the Hecke operators. We define Tz to be the ring generated over Z by the Hecke operators T, and (d) acting on the space S(I',Z). More generally, if A is any ring, we define Ta to be the A-algebra Tz ® A. This Hecke ring acts on the space Sp (I, A) in a natural way. Before studying the structure of the Hecke rings T4 as we vary the rings 4, we note the following general result (Cf. [Shi], ch. 3.): Lemma 1.34 The space 52(I,A)Y = Homa(So(I°, A), A) is a free Ta-module of rank one. Sketch of proof: One checks that the pairing-Tz x So(I',Z) —+ Z defined by (T, f) + a (Tf) sets up a perfect, Tz-equivariant duality between Tz and S(T, Z). The result for arbitrary A follows. o Hecke rings over C: If A = C, then the structure of the ring T = Te is completely described by theorem 1.22. More precisely, if Tey denotes the image of the Hecke algebra acting on the space S, defined in section 1.3, then Te = @sTes, where the direct sum ranges over all distinct newforms f of some level Ny dividing N. Furthermore, the algebra Tey can be described explicitly. In34 1. Elliptic curves and modular forms particular, if f is a newform of level N then Tey is isomorphic to C, but if Ny is not equal to N then the ring Tey need not be a semi-simple algebra over C. Lemma 1.34 in the case A = € says that V = S2(I) is a free Te-module of rank one, but we also have: Lemma 1.35 The module S,(P) is a free Te-module of rank one. Proof: Let g1,.-. ge be a complete system of newforms of levels Ni, ..., Ne dividing N. One can check that the form 9 = 91(N/Nir) + +++ + ge(N/Ner) generates $2(I) as a Te-module. The map Tc —> S2(P) defined by T+ Tg gives an isomorphism from Te to S2(L), as Tc-modules. o Remark 1.36 Lemmas 1.34 and 1.35 imply that Tc is a Gorenstein C-algebra (of finite rank), i.e., Home(Tc,C) is isomorphic to Te as a Te-module. Hecke rings over Q: Let [f] be the Galois orbit (under the action of Gg = Gal (Q/Q)) of a normalized newform f of some level N dividing N, and let Ky be the field extension of Q generated by the Fourier coefficients of f. The space ®ge{4}5o is a vector space of dimension [K's : Qloo(N/N;), which is spanned by modular forms with rational Fourier coefficients. Let Sj) be the Q-subspace of forms in ®ge{s)S with rational Fourier coefficients. The space Si) is stable under the action of Tg, and letting Tgiyj be the image of Tg acting on Sty, we obtain the direct sum decomposition To = @)Toins where the sum is taken over the distinct Gg-orbits of normalized newforms f of some level Ny dividing N. If Ny is equal to N, then the algebra Tgip) is isomorphic to the field Ky. If Ny is a proper divisor of N, then, as in the complex case, the algebra Tg 4) is a (not necessarily semi-simple) algebra over Q of rank oo(N/N,)[Ky : Q). The nature of the fields Ky, and in general the structure of Ta, is very poorly understood at this stage; for example, one does not know how to characterize the number fields that occur as a Ky for some f (but they are all known to be totally real or CM fields). The ring Tg acts naturally on the rational homology Hi(Xr,Q) = A@Q, and we have Lemma 1.37 The module A @ Q is free of rank two over Ta. Sketch of proof: The modules A+ @ C = V and A~ @C ~ V are free of rank one over Te, by lemma 1.34. This implies that A* @ Q and A~ @ Q are both free of rank one over Tg. ao Hecke rings over Z: The ring Tz is a certain (not necessarily maximal) order in Tg. One still has an injection Tz OyTztsp where now Tz,1y) denotes the ring generated over Z by the Hecke operators acting on Siz. Of course the structure of Tz is even more mysterious than that1. Elliptic curves and modular forms 35 of Tg! The ring Tz acts naturally on A, but it is not the case in general that Ais free of rank two over Tz, i.e., that the integral analogue of lemma 1.37 is true. (See remark 1.42 below.) Hecke rings over Qy: The study of the algebras Tz, and Tg, arises naturally because of the Hecke action on the Tate module 7¢(Jr), ime Ie), Te(Jr) where the inverse limit is taken with respect to the multiplication by € maps. The action of Tz, on Te(Jr) is compatible with that of Gg, and it is this pair of actions on the Tate module which is used to associate two-dimensional Galois representations to modular forms. It will sometimes be more convenient to consider the ring Tg, and its action on V=TN(Jr) Oz. &- We first record a useful duality property enjoyed by the Tate modules. The Weil pairings on the groups Jr{¢*] for n > 1 induce a perfect pairing (,.) 5 TelJr) * Te Jr) > Ze. Since each Hecke operator T is adjoint to wPw where w = wy is the Atkin- Lehner involution, we have the following lemma. Lemma 1.38 The map ++ $s where b.(y) = (x,wy) defines an isomorphism of Tz,-modules, Ti Jr) ¥ Te Ip)’ = Homz,(Te(Jr), Z2), and hence an isomorphism of Tg,-modules Vv & VY = Homa, (V,Q). The following lemma allows us to regard Tg, as a coefficient ring for a two-dimensional Galois representation Gal (Q/Q) + Autre, (V) ¥ GLa(Ta,)- Lemma 1.39 The module V is free of rank 2 over Ta,- Proof: Lemma 1.37 implies directly that the module V = A @ Q, is free of rank two over Ta,. . o Corollary 1.40 The ring Tg, is a Gorenstein Qr-algebra; i.e., Tg, = Home, (Te, &) is free of rank one over Ta,. Proof, Choosing a basis for V over Tg,, we obtain an isomorphism To, © Te, = Te, © Te.-36 1. Elliptic curves and modular forms Decomposing Tg, as [], R; where each factor R; is a finite-dimensional local Q-algebra, we obtain an isomorphism Ri @R = RY ORY for each i. At least one of the four maps RY —> R; deduced from this isomor- phism must be surjective, and by counting dimensions, we see that it must be injective as well. It follows that Tg, is isomorphic to Tg,- o Recall that for primes p not dividing N, the Jacobian Jp has good reduction mod p, and the Eichler-Shimura relation, theorem 1.29, states that on Jryr,, we have Tp = F + (p)F". For primes p not dividing Né, we may identify 7;(Jr) with the é-adic Tate module of the reduction (see (ST]) and consider the Frobenius endomorphism F on the free rank two Tg,-module V. As a consequence of the Eichler-Shimura relation, we find: Theorem 1.41 For p not dividing N€, the characteristic polynomial of F on the Tg,-module V is X? ~T,X + (p)p. Proof: (We are grateful to Brian Conrad for showing us this argument.) Since FF' = p, it follows from the Eichler-Shimura relation that F? TF + (p)p = 0. To conclude that this is in fact the characteristic polynomial, it suffices to compute the trace of F. To do so, we use the Tg, isomorphism vow defined by the modified pairing (-,w-). Under this modified pairing, F is adjoint to wF'w = (p)F’, so the trace of F on V is the same as that of ++ do ((p) F’) on VY. Choosing bases for V and Ty,, one sees that this is the same as the trace of (p)F'. Hence Qtr F = tr F + tr((p)F’) = tr (Tp) = 27). o Hecke rings over Z;: The ring Tz, is free of finite rank over Zz. It therefore decomposes as Tz, =] Ts, where the product runs over the maximal ideals m of Tz,, and Tis the lo- calization of Tz, at m. For each m, Tm is a complete local Zealgebra, free of finite rank as a Zmodule. Remark 1.42 While the analogue of lemma 1.39 does not always hold for Tz, (see [MRi], sec. 13) we shall see that it holds for certain localizations Tm. Results of this type are much deeper than lemma 1.39 and were first obtained by Mazur [Maz]], sec. 14, 15. We shall return in chapter 4 to explain Mazur’s result and its generalizations, which play a role in the arguments of (W3] and (Tw).1. Elliptic curves and modular forms 37 Example 1.43 The curve Xo(19) has genus 1, and Xo(57) has genus 5. By consulting the tables in the Antwerp volume [Ant4] or Cremona’s book [Cr], one finds that there is exactly one newform of level 19, and that there are three newforms of level 57, which all have rational Fourier coefficients. Their Fourier coefficients ap, for the first few primes p, are listed in the following table: Setting f = 194, we find that a basis of simultaneous eigenforms in $2(57,C) for the Hecke operators Tp, ! # 3, 19, and Us, Uso is: f(r) + (1+ V=2)f (87), flr) + (1- V=2)f(37), 57A, 57B, 570. It appears from the table that the Fourier coefficients corresponding to the forms 57B and 57C are congruent modulo 3. This is in fact the case. One finds that the Hecke ring Tz, generated over Zz by the Hecke operators acting on S3(57,Zs) is isomorphic to the subalgebra of Z: {(z,y,2,t, w) such that = w (mod 3)}. The isomorphism sends T> (for p # 3, 19) to the element (ap(19A),a9(19),a5(57A), ap(57B), ap(57C)). It sends Us to (-1 + V—2,-1 — Y=2,-1,1,1) and Uso to (1,1,-1,-1,-1). Thus there are exactly 4 distinct maximal ideals of Tz,, and the localizations at these maximal ideals are the rings Zs, Zs, Zs, and Tr = {(t,w) such that t (mod 3)}, where m is the ideal generated by 3 and Tn — an(57B) for all n. 1.7 The Shimura construction Let f = Slang” be an eigenform on I’ with (not necessarily rational) Fourier coefficients, corresponding to a surjective algebra homomorphism Ay : Tg > K,, where Ks is the field generated over Q by the Fourier coefficients of f. We briefly review in this section a construction of Shimura ({Shi2], ch. 7) which associates to f (or rather, to the orbit [f] of f under Gg) an abelian variety Ag defined over Q and of dimension [Ky:Q. Let Iy C Tz be the ideal ker(Ay) Tz. The image I;(Jr) is a (connected) subabelian variety of Jp which is stable under Ty and is defined over Q. Definition 1.44 The abelian variety Ay associated to f is the quotient Ay = Jr/Tp Jr From this definition one sees that Ay is defined over Q and depends only on [f], and that its endomorphism ring contains Tz/I, which is isomorphic to an order in Ky.38 1. Elliptic curves and modular forms Remark 1.45 Using theorem 1.22, one can show that Jo(V) is isogenous to Ty An” We now describe the abelian variety Ay as a complex torus. Let V; be the subspace of V = S»(P)Y on which T acts by \y. Theorem 1.22 and lemma 1.34 show that V; is a one-dimensional complex vector space. Let 7y be the or- thogonal projection of V to Vj, relative to the Petersson scalar product. The projector my belongs naturally to Tx,. Let [f] be the set of all eigenforms whose Fourier coefficients are Galois conjugate to those of f. The number of forms in [/] is equal to the degree d of Ky over Q. Now, we set Vin =SrcteiMor my = Do ty sell Note that my) is simply the orthogonal projection of V to Vip). Note also that a7] belongs to the Hecke algebra Tg. Lemma 1.46 The abelian variety Ay is isomorphic over C to the complet torus Vin /miy\(A), with the map my): V/A —+ Vigj/my(A) corresponding to the natural projection from Jp to Ay. In particular, one sees that A, is an abelian variety of dimension d = [K; : Ql. Hence if f has rational Fourier coefficients, then the abelian variety Ay is an elliptic curve. This elliptic curve is called the strong modular elliptic curve associated to f if also f is a newform of level N and I’ =To(N). Example 1.47 If = T(26), one checks that the genus of Xp is two, and that there are two distinct normalized eigenforms fy and fy in S(26). From the tables in [Antd] or [Cr], one sees that fy and fy have integral Fourier coefficients, whose values for the primes < 31 are: Hence the abelian varieties Ay, and Ay, are elliptic curves. The above table suggests (and this can be checked directly by looking at the equations for these curves given in [Ant4] or [Cr], or by using the discussion in [DO], lemma 2.1) that the Fourier coefficients of f; and fy are congruent modulo 2. The natural projection Jo(26) —+ Ay, ® Ay, is an isogeny whose kernel is isomorphic to Z/2L x Z/2Z, and Jo(26) is not isomorphic to Ay, ® Ay,. More generally, one knows that a Jacobian can never decompose as a non-trivial direct product of two principally polarized abelian varieties, cf. [Mazl], prop. 10.6. (There are no non-zero homomorphisms from Ay, to Ay,, and so a non-trivial product decomposition of Jp would have to induce a decomposition as a product of principally polarized abelian varieties.) Let 7;(Ay) be the Tate module of the abelian variety Ay, Te(Ay) := lim(Ay) IE"),