The Absolute Anabelian Geometry of Canonical Curves

Shinichi Mochizuki

Abstract. In this paper, we continue our study of the issue of the extent to which a hyperbolic curve over a nite extension of the eld of p-adic numbers is determined by the pronite group structure of its etale fundamental group. Our main results are that: (i) the theory of correspondences of the curve in particular, its arithmeticity is completely determined by its fundamental group; (ii) when the curve is a canonical lifting in the sense of p-adic Teichm uller theory, its isomorphism class is functorially determined by its fundamental group. Here, (i) is a consequence of a p-adic version of the Grothendieck Conjecture for algebraic curves proven by the author, while (ii) builds on a previous result to the eect that the logarithmic special ber of the curve is functorially determined by its fundamental group. 2000 Mathematics Subject Classication: 14H25, 14H30 Keywords and Phrases: hyperbolic curve, etale fundamental group, anabelian, correspondences, Grothendieck Conjecture, canonical lifting, padic Teichm uller theory Contents: 1. Serre-Tate Canonical Liftings 2. Arithmetic Hyperbolic Curves 3. Hyperbolically Ordinary Canonical Liftings Introduction Let XK be a hyperbolic curve (cf. 0 below) over a eld K of characteristic 0. Denote its algebraic fundamental group by XK . Thus, we have a natural surjection XK GK
Typeset by AMS-TEX

Shinichi Mochizuki

of XK onto the absolute Galois group GK of K . When K is a nite extension of Q or Qp , and one holds GK xed, then it is known (cf. [Tama], [Mzk6]) that one may recover the curve XK in a functorial fashion from XK . This sort of result may be thought of as a relative result (i.e., over GK ). In the present paper, we continue our study begun in [Mzk7] of absolute analogues of such relative results. Since such absolute analogues are well understood in the case where K is a nite extension of Q (cf. the Introduction to [Mzk7]), we concentrate on the p-adic case. In the p-adic case, it is proven in [Mzk7] (cf. [Mzk7], Theorem 2.7) by applying the work of [Tama] and the techniques of [Mzk5] that (if XK has stable reduction, then) the logarithmic special ber of XK i.e., the special ber, equipped with its natural log structure (cf. [Kato]), of the stable model of XK over the ring of integers OK may be recovered solely from the abstract pronite group XK . This result prompts the question (cf. [Mzk7], Remark 2.7.3): What other information e.g., the isomorphism class of XK itself can be recovered from the pronite group XK ? In this present paper, we give three partial answers to this question (cf. [Mzk7], Remark 2.7.3), all of which revolve around the central theme that: When XK is, in some sense, canonical, there is a tendency for substantial information concerning XK e.g., its isomorphism class to be recoverable from XK . Perhaps this tendency should not be surprising, in light of the fact that in some sense, a canonical curve is a curve which is rigid, i.e., has no moduli, hence should be determined by its special ber (cf. Remark 3.6.3). Our three partial answers are the following: (a) The property that the Jacobian of the XK be a Serre-Tate canonical lifting is determined by XK (Proposition 1.1). (b) The theory of correspondences of XK in particular, whether or not XK is arithmetic (cf. [Mzk3]) is determined by XK (cf. Theorem 2.4, Corollary 2.5). (c) The property that XK be a canonical lifting in the sense of the theory of [Mzk1] (cf. also [Mzk2]) is determined by XK ; moreover, in this case, the isomorphism class of XK is also determined by XK (cf. Theorem 3.6). At a technical level, (a) is entirely elementary; (b) is a formal consequence of the p-adic version of the Grothendieck Conjecture proven in [Mzk6], Theorem

Canonical Curves

A; and (c) is derived as a consequence of the theory of [Mzk1], together with [Mzk7], Theorem 2.7. Finally, as a consequence of (c), we conclude (cf. Corollary 3.8) that the set of points arising from curves over nite extensions of Qp whose isomorphism classes are completely determined by XK forms a Zariski dense subset of the moduli stack over Qp . This result (cf. Remark 3.6.2) constitutes the rst application of the p-adic Teichm uller theory of [Mzk1], [Mzk2], to prove a hitherto unknown result that can be stated without using the terminology, concepts, or results of the theory of [Mzk1], [Mzk2]. Also, it shows that unlike (a), (b) which only yield useful information concerning XK in very rare cases (c) may be applied to a much larger class of XK (cf. Remarks 1.1.1, 2.5.1, 3.6.1). Acknowledgements: I would like to thank A. Tamagawa for useful comments concerning earlier versions of this manuscript.

Shinichi Mochizuki Section 0: Notations and Conventions

We will denote by N the set of natural numbers, by which we mean the set of integers n 0. A number eld is dened to be a nite extension of the eld of rational numbers Q. Suppose that g 0 is an integer. Then a family of curves of genus g XS is dened to be a smooth, proper, geometrically connected morphism X S whose geometric bers are curves of genus g. Suppose that g, r 0 are integers such that 2g 2 + r > 0. We shall denote the moduli stack of r -pointed stable curves of genus g (where we assume the points to be unordered) by Mg,r (cf. [DM], [Knud] for an exposition of the theory of such curves; strictly speaking, [Knud] treats the nite etale covering of Mg,r determined by ordering the marked points). The open substack Mg,r Mg,r of smooth curves will be referred to as the moduli stack of smooth r -pointed stable curves of genus g or, alternatively, as the moduli stack of hyperbolic curves of type (g, r ). A family of hyperbolic curves of type (g, r ) XS is dened to be a morphism which factors X Y S as the composite of an open immersion X Y onto the complement Y \D of a relative divisor D Y which is nite etale over S of relative degree r , and a family Y S of curves of genus g. One checks easily that, if S is normal, then the pair (Y, D) is unique up to canonical isomorphism. (Indeed, when S is the spectrum of a eld, this fact is well-known from the elementary theory of algebraic curves. Next, we consider an arbitrary connected normal S on which a prime l is invertible (which, by Zariski localization, we may assume without loss of generality). Denote by S S the nite etale covering parametrizing orderings of the marked points and trivializations of the l-torsion points of the Jacobian of Y . Note that S S is independent of the choice of (Y, D), since (by the normality of S ) S may be constructed as the normalization of S in the function eld of S (which is independent of the choice of (Y, D) since the restriction of (Y, D) to the generic point of S has already been shown to be unique). Thus, the uniqueness of (Y, D) follows by considering the classifying morphism (associated to (Y, D)) etale covering of (Mg,r )Z[ 1 parametrizing orderings of the from S to the nite l] marked points and trivializations of the l-torsion points of the Jacobian [since this covering is well-known to be a scheme, for l suciently large].) We shall refer to Y (respectively, D; D; D) as the compactication (respectively, divisor at innity; divisor of cusps; divisor of marked points) of X .

Canonical Curves

A family of hyperbolic curves X S is dened to be a morphism X S such that the restriction of this morphism to each connected component of S is a family of hyperbolic curves of type (g, r ) for some integers (g, r ) as above.

Shinichi Mochizuki Section 1: Serre-Tate Canonical Liftings

In this , we observe (cf. Proposition 1.1 below) that the issue of whether or not the Jacobian of a p-adic hyperbolic curve is a Serre-Tate canonical lifting is completely determined by the abstract pronite group structure of its arithmetic pronite group. Let p be a prime number. For i = 1, 2, let Ki be a nite extension of Qp , and (Xi )Ki a proper hyperbolic curve over Ki whose associated stable curve has stable reduction over OKi . Denote the resulting stable model of (Xi )Ki over OKi by (Xi )OKi . Assume that we have chosen basepoints of the (Xi )Ki (which thus induce basepoints of the Ki ) and suppose that we are given an isomorphism of pronite groups (X1 )K1 (X2 )K2 , which (by [Mzk7], Lemmas 1.1.4, 1.1.5) induces a commutative diagram: (X1 )K1 G K1

(X2 )K2 G K2

Proposition 1.1. (Group-Theoreticity of Serre-Tate Canonical Liftings) The Jacobian of (X1 )K1 is Serre-Tate canonical if and only if the same is true of the Jacobian of (X2 )K2 . Proof. Indeed, this follows from the fact that the Jacobian of (Xi )Ki is a Serre-Tate canonical lifting if and only if its p-adic Tate module splits (as a GKi -module) into a direct sum of an unramied GKi -module and the Cartier dual of an unramied GKi -module (cf. [Mess], Chapter V: proof of Theorem 3.3, Theorem 2.3.6; [Mess], Appendix: Corollary 2.3, Proposition 2.5). Remark 1.1.1. As is shown in [DO] (cf. also [OS]), for p > 2, g 4, the Serre-Tate canonical lifting of the Jacobian of a general proper curve of genus g in characteristic p is not a Jacobian. Thus, in some sense, one expects that: There are not so many curves to which Proposition 1.1 may be applied. From another point of view, if there exist innitely many Jacobians of a given genus g over nite extensions of Qp which are Serre-Tate canonical liftings, then one expects cf. the Andr e-Oort Conjecture ([Edix], Conjecture 1.3) that every irreducible component of the Zariski closure of the resulting set of points in the moduli stack of principally polarized abelian varieties should be a

Canonical Curves

subvariety of Hodge type. Moreover, one expects that the intersection of such a subvariety with the Torelli locus (i.e., locus of Jacobians) in the moduli stack of principally polarized abelian varieties should typically be rather small. Thus, from this point of view as well, one expects that Proposition 1.1 should not be applicable to the overwhelming majority of curves of genus g 2.

Shinichi Mochizuki Section 2: Arithmetic Hyperbolic Curves

In this , we show (cf. Theorem 2.4 below) that the theory of correspondences (cf. [Mzk3]) of a p-adic hyperbolic curve is completely determined by the abstract pronite group structure of its arithmetic pronite group. We begin by reviewing and extending the theory of [Mzk3], as it will be needed in the discussion of the present . Let X be a normal connected algebraic stack which is generically schemelike (i.e., admits an open dense algebraic substack isomorphic to a scheme). Then we shall denote by Loc(X ) the category whose objects are (necessarily generically scheme-like) algebraic stacks Y that admit a nite etale morphism to X , and whose morphisms are nite etale morphisms of stacks Y1 Y2 (that do not necessarily lie over X !). Note that since these stacks are generically scheme-like, it makes sense to speak of the (1-)category of such objects (so long as our morphisms are nite etale), i.e., there is no need to work with 2-categories. Given an object Y of Loc(X ), let us denote by Loc(X )Y the category whose objects are morphisms Z Y in Loc(X ), and whose morphisms, from an object Z1 Y to an object Z2 Y , are the morphisms Z1 Z2 over Y in Loc(X ). Thus, by considering maximal nontrivial decompositions of the terminal object of Loc(X )Y into a coproduct of nonempty objects of Loc(X )Y , we conclude that the set of connected components of Y may be recovered functorially! from the category structure of Loc(Y ). Finally, let us observe that Loc(X )Y may be identied with the category Y) Et( of nite etale coverings of Y (and Y -morphisms). We would also like to consider the category Loc(X ) whose objects are generically scheme-like algebraic stacks which arise as nite etale quotients (in the sense of stacks!) of objects in Loc(X ), and whose morphisms are nite etale morphisms of algebraic stacks. Note that Loc(X ) may

Canonical Curves

be constructed entirely category-theoretically from Loc(X ) by considering the category of objects of Loc(X ) equipped with a (nite etale) equivalence relation. (We leave it to the reader to write out the routine details.) Denition 2.1. (i) X will be called arithmetic if Loc(X ) does not admit a terminal object. (ii) X will be called a(n) (absolute) core if X is a terminal object in Loc(X ). (ii) X will be said to admit a(n) (absolute) core if there exists a terminal object Z in Loc(X ). In this case, Loc(X ) = Loc(Z ), so we shall say that Z is a core. Remark 2.1.1. Let k be a eld. If X is a geometrically normal, geometrically connected algebraic stack of nite type over k , then we shall write Lock (X ); Lock (X )

for the categories obtained as above, except that we assume all the morphisms to be k -morphisms. Also, we shall say that X is k -arithmetic, or arithmetic over k (respectively, a k -core, or core over k ), if Lock (X ) does not admit a terminal object (respectively, X is a terminal object in Lock (X )). On the other hand, when k is xed, and the entire discussion takes place over k , then we shall often omit the k - from this terminology. Remark 2.1.2. Thus, when k = C, a hyperbolic curve X is k -arithmetic if and only if it is arithmetic in the sense of [Mzk3], 2. (Indeed, if X is non-arithmetic in the sense of [Mzk3], 2, then a terminal object in Lock (X ) i.e., a (hyperbolic) core is constructed in [Mzk3], 3, so X is non-k arithmetic. Conversely, if X is arithmetic in the sense of [Mzk3], 2, then (cf. [Mzk3], Denition 2.1, Theorem 2.5) it corresponds to a fuchsian group SL2 (R)/{1} which has innite index in its commensurator CSL2(R)/{1} () a fact which precludes the existence of a k -core.) Moreover, issues over an arbitrary algebraically closed k of characteristic zero may always be resolved over C, by Proposition 2.3, (ii), below. Remark 2.1.3. If we arbitrarily choose a nite etale structure morphism to X for every object of Loc(X ), then one veries easily that every morphism of Loc(X ) factors as the composite of an isomorphism (not necessarily over X !) with a (nite etale) morphism over X (i.e., relative to these arbtirary choices). A similar statement holds for Lock (X ).

10

Shinichi Mochizuki

Denition 2.2. Let X be a smooth, geometrically connected, generically scheme-like algebraic stack of nite type over a eld k of characteristic zero. (i) We shall say that X is an orbicurve if it is of dimension 1. (ii) We shall say that X is a hyperbolic orbicurve if it is an orbicurve which admits a compactication X X (necessarily unique!) by a proper orbicurve X over k such that if we denote the reduced divisor X \X by D X , then X is scheme-like near D, and, moreover, the line bundle X /k (D) on X has positive degree. Proposition 2.3. (Independence of the Base Field)

(i) Let k sep be a separable closure of k ; X a geometrically normal, geometrically connected algebraic stack of nite type over k . Then X is a k -core def (respectively, k -arithmetic) if and only if Xk sep = X k k sep is a k sep -core (respectively, k sep -arithmetic). Moreover, if Xk sep admits a nite etale morphism Xk sep Zk sep to a k sep -core Zk sep , then Zk sep descends uniquely to a k -core Z of X . (ii) Suppose that k is algebraically closed of characteristic 0, and that X is a hyperbolic orbicurve. Next, let k be an algebraically closed eld containing k . Then the natural functors Lock (X ) Lock (X k k ); Lock (X ) Lock (X k k )

(given by tensoring over k with k ) are equivalences of categories. In particular, X is a k -core (respectively, k -arithmetic) if and only if X k k is a k -core (respectively, k -arithmetic). Proof. First, we observe that (i) is a formal consequence of the denitions. As for (ii), let us observe rst that it suces to verify the asserted equivalences of categories. These equivalences, in turn, are formal consequences of the following two assertions (cf. Remark 2.1.3): X ) Et( X k k ) is an equivalence of (a) The natural functor Et( categories. etale over X , then (b) If Y1 , Y2 are nite Isomk (Y1 , Y2 ) Isomk (Y1 k k , Y2 k k ) is bijective. The proofs of these two assertions is an exercise in elementary algebraic geometry, involving the following well-known techniques:

Canonical Curves (1) descending the necessary diagrams of nite etale morphisms over k to a subeld K k which is nitely generated over k ; (2) extending orbicurves over K to orbicurves over some k -variety V with function eld K ; (3) specializing orbicurves over V to closed (i.e., k -valued) points v of V ; (4) base-changing orbicurves over V to formal completions Vv of V at closed points v; (5) deforming (log) etale morphisms of orbicurves over v to morphisms over the completions Vv ; (6) algebrizing such deformed morphisms (when the orbicurves involved are proper).

11

This elementary exercise is carried out (for assertion (a) above) in the case when X itself is proper in [SGA1], Expos e X, Theorem 3.8. When X is an arbitrary orbicurve as in the statement of (ii), the same arguments centering around the rigidity of (log) etale morphisms under innitesimal deformations may be used, by considering compactications (X, D) of X as in Denition 2.2, (ii), and replacing etale by etale away from D. Note that we use the assumption that k is of characteristic zero here to ensure that all ramication is tame. Finally, assertion (b) may be deduced by similar arguments by applying, in (5) above, the fact (cf. Denition 2.2, (ii)) that, if Y X is any nite morphism of orbicurves over k , then
(D)|Y ) = 0 H 0 (Y , X/k

(where denotes the OX -dual) in place of the rigidity of (log) etale morphisms used to prove assertion (a). Next, for i = 1, 2, let Ki be a nite extension of Qp (where p is a prime number); let (Xi )Ki be a hyperbolic curve over Ki . Assume that we have chosen basepoints of the (Xi )Ki , which thus induce basepoints/algebraic closures K i of the Ki and determine fundamental groups (Xi )Ki = 1 ((Xi )Ki ) and Galois groups GKi = Gal(K i /Ki ). Thus, for i = 1, 2, we have an exact sequence: 1 Xi (Xi )Ki GKi 1 (where Xi (Xi )Ki is dened so as to make the sequence exact). Here, we shall think of GKi as a quotient of (Xi )Ki (i.e., not as an independent group
def def

12

Shinichi Mochizuki

to which (Xi )Ki happens to surject). By [Mzk7], Lemmas 1.1.4, 1.1.5, this quotient is characteristic, i.e., it is completely determined by the structure of (Xi )Ki as a pronite group. Theorem 2.4. (Group-Theoreticity of Correspondences) Any iso morphism : (X1 )K1 (X2 )K2 induces equivalences of categories: LocK 1 ((X1 )K 1 ) LocK 2 ((X2 )K 2 ); in a fashion that is functorial in . Proof. Since LocK i ((Xi )K i ) may be reconstructed category-theoretically from LocK i ((Xi )K i ) (cf. the discussion at the beginning of the present ), in order to prove Theorem 2.4, it thus suces to show that the isomorphism induces an equivalence between the categories LocK i ((Xi )K i ). Clearly, the class of objects of LocK i ((Xi )K i ) may be reconstructed as the class of objects of the category of nite sets with continuous Xi -action. To reconstruct the morphisms, it suces (cf. Remark 2.1.3) to show that given any two open subgroups H1 , J1 (X1 )K1 which we may assume, without loss of generality, to surject onto GK1 and an isomorphism H1 J1 that arises K1 -geometrically (i.e., from a K1 -scheme-theoretic isomorphism between the curves corresponding to H1 , J1 ), it is necessarily the case that the corresponding isomorphism H2 J2 between open subgroups H2 , J2 (X2 )K2 arises K2 -geometrically. But this follows formally from the p-adic version of the Grothendieck Conjecture proven in [Mzk6], Theorem A: Indeed, H1 J1 necessarily lies over an inner automorphism 1 : GK1 GK1 . In particular, H2 J2 lies over an isomorphism 2 : GK2 GK2 , which is obtained by conjugating 1 by some xed isomorphism (not necessarily geometric!) arising from between the characteristic quotients GK1 GK2 . Since the property of being an inner automorphism is manifestly intrinsic, we thus conclude that 2 is also an inner automorphism. This allows us to apply [Mzk6], Theorem A, which implies that H2 J2 arises K2 -geometrically, as desired. Corollary 2.5. (Consequences for Cores and Arithmeticity) Let : (X1 )K1 (X2 )K2

LocK 1 ((X1 )K 1 ) LocK2 ((X2 )K 2 )

Canonical Curves be an isomorphism. Then:

13

(i) (X1 )K 1 is K 1 -arithmetic (respectively, a K 1 -core) if and only if (X2 )K 2 is K 2 -arithmetic (respectively, a K 2 -core). (ii) Suppose that, for i = 1, 2, we are given a nite etale morphism (Xi )Ki (Zi )Ki to a Ki -core (Zi )Ki . Then the isomorphism extends uniquely to an isomorphism (Z1 )K1 (Z2 )K2 . Proof. Assertion (i) is a formal consequence of Theorem 2.4 and Denition 2.1 (cf. also Remark 2.1.1). In light of Proposition 2.3, (i), assertion (ii) is a formal consequence of Theorem 2.4, at least over some corresponding nite Galois extensions K1 , K2 of K1 , K2 . That the resulting extension (Z1 )K (Z2 )K 1 2 of is unique is a formal consequence of the fact that every open subgroup of (Xi )Ki has trivial centralizer in (Zi )Ki (cf. [Mzk7], Lemma 1.3.1, Corollary 1.3.3). Moreover, it follows formally from this triviality of centralizers that, by choosing corresponding normal open subgroups Hi (Zi )Ki such that Hi XK , we may think of (Zi )Ki (and its various open subgroups) as i subgroups of Aut(Hi ), in a fashion which is compatible with and its various (unique) extensions. Thus, since (Zi )Ki is generated by (Zi )K and (Xi )Ki , we conclude that this extension (Zi )K (Zi)K over the Ki descends to some (Z1 )K1 (Z2 )K2 , as desired.

1 2

Remark 2.5.1. Recall from the theory of [Mzk3] (cf. Remark 2.1.2; Proposition 2.3, (ii), of the discussion above) that (Xi )K i is arithmetic if and only if it admits a nite etale cover which is a nite etale cover of a Shimura curve, i.e., (equivalently) if there exists a Shimura curve in LocKi ((Xi )K i ). As is discussed in [Mzk3], Theorem 2.6, a theorem of Takeuchi states that for a given (g, r ), there are only nitely many isomorphism classes of hyperbolic curves of type (g, r ) (over a given algebraically closed eld of characteristic zero) which are arithmetic. Moreover, a general hyperbolic curve of type (g, r ) is not only non-arithmetic; it is, in fact, equal to its own hyperbolic core (cf. [Mzk3], Theorem 5.3). Thus, for general curves of a given type (g, r ), the structure of the category Loc((Xi )K i ) is not sucient to determine the isomorphism class of the curve. It is not clear to the author at the time of writing whether or not, in the case when (Xi )Ki is arithmetic, the structure of the category Loc((Xi )K i ) is sucient to determine the isomorphism class of (Xi )Ki . At any rate, just as was the case with Proposition 1.1 (cf. Remark 1.1.1), Theorem 2.4 does not allow one to recover the isomorphism class of (Xi )Ki for most (Xi )Ki where here we take most to mean that (at least for (g, r ) suciently large) the set of points determined by the curves for which it is possible to recover the isomorphism class of (Xi )Ki from the pronite group (Xi )Ki via the method

14

Shinichi Mochizuki

in question fails to be Zariski dense in the moduli stack of hyperbolic curves of type (g, r ) (cf. Corollary 3.8 below). Finally, to give the reader a feel for the abstract theory and, in particular, the state of aairs discussed in Remark 2.5.1 above we consider the case of punctured hemi-elliptic orbicurves, in which the situation is understood somewhat explicitly: Denition 2.6. Let X be an orbicurve (cf. Denition 2.2, (i)) over a eld of characteristic zero k . (i) We shall say that X is a hemi-elliptic orbicurve if it is obtained by forming the quotient in the sense of stacks of an elliptic curve by the action of 1. (ii) We shall say that X is a punctured hemi-elliptic orbicurve if it is obtained by forming the quotient in the sense of stacks of a once-punctured elliptic curve (i.e., the open subscheme given by the complement of the origin in an elliptic curve) by the action of 1. Proposition 2.7. (Punctured Hemi-Elliptic Cores) Let k be an algebraically closed eld of characteristic 0; let X be a punctured hemi-elliptic orbicurve over k . Then if X is non-k-arithmetic, then X is a k -core. In particular, if X admits nontrivial automorphisms (over k ), then X is k -arithmetic. Finally, there exist precisely 4 isomorphism classes of k -arithmetic X , which are described explicitly in [Take2], Theorem 4.1, (i). Proof. In the following discussion, we omit the k -. Suppose that X is non-arithmetic. Write Y X for the unique double ( etale) covering by a punctured elliptic curve Y , and Y Z for the unique morphism to the core (i.e., the terminal object in Lock (X ) = Lock (Y )). Thus, the induced morphism Y =Y
crs

crs

on the coarse moduli spaces (cf. [FC], Chapter I, Theorem 4.10) associated to the canonical compactications Y , Z of the orbicurves Y , Z is a nite ramied

Canonical Curves

15

covering morphism whose degree we denote by d from an elliptic curve crs . Note that since Y has only one cusp y (i.e. Y to a copy of P1 k = Z point Y \Y ), and a point of Y is a cusp if and only if its image is a cusp in Z , it follows that Z also has a unique cusp z , and that y is the unique crs arises from a nite point of Y lying over z . Moreover, because Y Z crs etale morphism Y Z , it follows that the ramication index of Y Z is crs the same at all points of Y lying over a given point of Z . Thus, applying the Riemann-Hurwitz formula yields: 0 = 2d +
i

d (ei 1) ei
crs

where the ei are the ramication indices over the points of Z at which the 1 (ei 1). Since covering morphism ramies. Thus, we conclude that 2 = i e i all of the ei are integers, one veries immediately that the only possibilities for the set of ei s are the following: (2, 2, 2, 2); (2, 3, 6); (2, 4, 4); (3, 3, 3) Note that it follows from the fact that y is the unique point of Y lying over z that d, as well as the ramication index at z , is necessarily equal to the largest ei . In the case of (2, 2, 2, 2), we thus conclude that X = Z , so X is a core, as asserted. In the other three cases, we conclude that Y is a nite etale covering of the orbicurve determined by a triangle group (cf. [Take1]) of one of the following types: (2, 3, ); (2, 4, ); (3, 3, ) By [Take1], Theorem 3, (ii), such a triangle group is arithmetic, so X itself is arithmetic, thus contradicting our hypotheses. This completes the proof of the rst assertion of Proposition 2.7. The second (respectively, third) assertion of Proposition 2.7 is a formal consequence of the rst assertion of Proposition 2.7 (respectively, [Take2], Theorem 4.1, (i)).

16

Shinichi Mochizuki Section 3: Hyperbolically Ordinary Canonical Liftings

In this , we would like to work over a nite, unramied extension K of Qp , where p is a prime number 5. We denote the ring of integers (respectively, residue eld) of K by A (respectively, k ). Since A = W (k) (the ring of Witt vectors with coecients in k ), we have a natural Frobenius morphism A : A A which lifts the Frobenius morphism k : k k on k . In the following discussion, the result of base-changing over A (respectively, A; A; Zp ) with k (respectively, K ; with A, via A ; Z/pn Z, for an integer n 1) will be denoted by a subscript k (respectively, subscript K ; superscript F ; subscript Z/pn Z). Let (X S = Spec(A), D X ) be a smooth, pointed curve of type (g, r ) (for which D is the divisor of marked points), where 2g 2+ r > 0 cf. 0. In the following discussion, we would like to consider the extent to which (X, D) is a canonical lifting of (Xk , Dk ), in the sense of [Mzk1], Chapter III, 3; Chapter IV. We refer also to the Introduction of [Mzk2] for a survey of p-adic Teichm uller theory (including the theory of [Mzk1]). Lemma 3.1. (Canonicality Modulo p2 ) Suppose that YK XK is a nite ramied morphism of smooth, proper, geometrically connected curves over K which is unramied away from DK . Denote the reduced induced subscheme associated to the inverse image in YK of DK by EK YK . Suppose further that the reduction Yk Xk modulo p of the normalization Y X of X in YK has the following form: (i) Yk is reduced; (ii) Yk is smooth over k , except for a total of precisely 1 2 (p 1)(2g 2 + r ) ( 2) nodes. Moreover, the order of the deformation of each node determined by Y is equal to 1 (equivalently: Y is regular at the nodes), and the special ber Ek Yk of the closure E Y of EK in Y is a reduced divisor (equivalently: a divisor which is etale over k ) at which Yk is smooth.
def

Canonical Curves (iii) Yk has precisely two irreducible components CV , CF . Here, the morphism CV Xk (respectively, CF Xk ) is an isomorphism (respectively, k -isomorphic to the relative Frobenius morphism Xk /k : F Xk of Xk ). Xk (iv) (Xk , Dk ) admits a nilpotent ordinary indigenous bundle (cf. [Mzk1], Chapter II, Denitions 2.4, 3.1) whose supersingular divisor (cf. [Mzk1], Chapter II, Proposition 2.6, (3)) is equal to the image of the nodes of Yk in Xk .

17

Then (X, D) is isomorphic modulo p2 to the canonical lifting (cf. [Mzk1], Chapter III, 3; Chapter IV) determined by the nilpotent indigenous bundle of (iv). Remark 3.1.1. In the context of Lemma 3.1, we shall refer to the points of Xk which are the image of nodes of Yk as supersingular points and to points which are not supersingular as ordinary. Moreover, the open subscheme of ordinary points will be denoted by
ord Xk Xk

and the corresponding p-adic formal open subscheme of X (the p-adic completion of X ) by X ord . Also, we shall consider X (respectively, Y ) to be equipped with the log structure (cf. [Kato] for an introduction to the theory of log structures) determined by the monoid of regular functions invertible on XK \DK (respectively, YK \EK ) and denote the resulting log scheme by X log (respectively, Y log ). Thus, the morphism of schemes Y X extends uniquely to a morphism of log schemes Y log X log . Proof. First, let us observe that
F (Y log )ord = {(X log )ord k k }

(X log )ord k

log where the isomorphism is the unique isomorphism lying over Xk . Since log ord log ord are log smooth over A, it follows that the inclusion (X ) , (Y ) F {(X log )ord (Y log )ord lifts to a (not necessarily unique!) inclusion k } k F (Y log )ord {(X log )ord Z/p2 Z } Z/p2 Z

whose composite
F log ord )Z/p2 Z log : {(X log )ord Z/p2 Z } (X

18

Shinichi Mochizuki

log ord with the natural morphism (Y log )ord )Z/p2 Z is nevertheless indepenZ/p2 Z (X dent of the choice of lifting of the inclusion. Indeed, this is formal consequence (cf., of the fact that log is a lifting of the Frobenius morphism on (X log )ord k e.g., the discussion of [Mzk1], Chapter II, the discussion preceding Proposition 1.2, as well as Remark 3.1.2 below).

Of course, log might not be regular at the supersingular points, but we may estimate the order of the poles of log at the supersingular points as follows: Since Y is assumed to be regular, it follows that the completion of YZ/p2 Z at a supersingular point is given by the formal spectrum Spf of a complete local ring isomorphic to: RY = (A/p2 A)[[s, t]]/(st p) (where s, t are indeterminates). Thus, modulo p, this completion is a node, with the property that precisely one branch i.e., irreducible component of this node lies on CF (respectively, CV ). (Indeed, this follows from the fact that both CF and CV are smooth over k .) Suppose that the irreducible component lying on CF is dened locally (modulo p) by the equation t = 0. Thus, the ring of regular functions on the ordinary locus of CF restricted to this formal neighborhood of a supersingular point is given by k [[s]][s 1]. The connected component of (Y log )ord Z/p2 Z determined by CF may be thought of as the open subscheme s = 0. Here, we recall that the parameter s in this discussion is uniquely determined up to multiplication by an element of R Y cf. [Mzk4], 3.7. Next, let us write: RX = Im(RY ) RY [s1 ]
def def

for the image of RY in RY [s1 ]. Then since t = s1 p RY [s1 ] = (A/p2 A)[[s]][s1 ] it follows that RX = (A/p2 A)[[s]][s1 p]. In particular, if we think of RX = (A/p2 A)[[s]] RX so RX [s1 ] = RX [s1 ] = RY [s1 ] as a local smooth lifting of CF at , we thus conclude that arbitrary regular functions on Spf(RY ) restrict to meromorphic functions on Spf(RX ) with poles of order 1 at . Thus, since log arises from an everywhere regular morphism Y log X log , we conclude that:
def

Canonical Curves log has poles of order 1 at the supersingular points.

19

But then the conclusion of Lemma 3.1 follows formally from [Mzk1], Chapter II, Proposition 2.6, (4); Chapter IV, Propositions 4.8, 4.10, Corollary 4.9. Remark 3.1.2. The fact cf. the end of the rst paragraph of the proof of Lemma 3.1 that the order of a pole of a Frobenius lifting is independent of the choice of smooth lifting of the domain of the Frobenius lifting may be understood more explicitly in terms of the coordinates used in the latter portion of the proof of Lemma 3.1 as follows: Any coordinate transformation s s + p g(s) (where g(s) k [[s]][s 1]) xes since we are working modulo p2 functions of the form sp + p f (s) (where f (s) k [[s]][s 1]). This shows that the order of the pole of f (s) does not depend on the choice of parameter s. In the situation of Lemma 3.1, let us denote the natural morphism of fundamental groups (induced by (X log )ord X log ) by

log (X log )ord = 1 ((X log )ord K ) X log = 1 (XK ) = 1 (XK \DK )

def

def

(for some xed choice of basepoints). Here, we observe that (by the main theorem of [Vala]) (X log )ord is excellent, so the normalization of (X log )ord in log ord ) . Thus, (X log )ord a nite etale covering of (X log )ord K is nite over (X K has a well-behaved theory of nite etale coverings which is compatible with etale localization on (X log )ord . Also, before proceeding, we observe that X log ts into an exact sequence: 1 X log X log GK 1 (where X log is dened so as to make the sequence exact). Lemma 3.2. (The Ordinary Locus Modulo p2 ) Let VFp be a 2dimensional Fp -vector space equipped with a continuous action of X log up to 1 i.e., a representation X log GL 2 (VFp ) = GL2 (VFp )/{1}
def

such that:

20

Shinichi Mochizuki (i) The determinant of VFp is isomorphic (as a X log -module) to Fp (1). (ii) There exists a nite log etale Galois covering (i.e., we assume tame ramication over D)
log X log X

such that the action of X log up to 1 on VFp lifts to a (usual) action (i.e., without sign ambiguities) of X log X log on VFp . This action is uniquely determined up to tensor product with a character of X log of order 2.
log log XK determined by the nite (iii) The nite etale covering YK X log -set of 1-dimensional Fp -subspaces of VFp satises the hypotheses of Lemma 3.1.

(iv) Write
log log (Z )K (X )K

for the nite log etale covering (of degree p2 1) corresponding to the nonzero portion of VFp . Write
log log )K (X )K (Y

for the nite log etale covering of smooth curves which is the composlog ) of the coverings ite (i.e., normalization of the ber product over XK log log log log log (Z )K , YK of XK . (Thus, (Z )K maps naturally to (Y )K , log hence also to YK .) Let us refer to an irreducible component of the log log )K (respectively, (Z )K ) special ber of a stable reduction of (Y over some nite extension of K that maps nitely to the irreducible component CF (cf. Lemma 3.1) as being of CF -type and assolog log )K (respectively, (Z )K ). Then any irreducible comciated to (Y log )K is etale and free of nodes ponent of CF -type associated to (Z over the ordinary locus of any irreducible component of CF -type log )K . associated to (Y Then (after possibly tensoring VFp with a character of X log of order 2) the

log ord ord on (X )K determined by the (X log )ord -module VFp etale local system EF p log ord arises from a (logarithmic) nite at group scheme on (X ) . Moreover, the (X log )ord -module VFp ts into an exact sequence:

) (1) VFp VFetl 0 0 (VFetl p p

Canonical Curves

21

etl where VF is a 1-dimensional Fp -space up to 1 whose (X log )ord -action p ord )k (X )k , and the denotes arises from a nite etale local system on (X the Fp -linear dual.

Proof. As was seen in the proof of Lemma 3.1, we have an isomorphism

(X log )ord k

which thus determines a decomposition of (Y log )ord into two connected components. Moreover, the second connected component on the right-hand side corresponds to a rank one quotient VFp QFp which is stabilized by the action of (X log )ord , while the rst connected component on the right-hand side 2 parametrizes splittings of this quotient VFp QFp . Here, we observe that Q Fp admits a natural (X log )ord -action (i.e., without sign ambiguities). Now any choice of isomorphism between {(X log )ord }F and the rst connected component of (Y log )ord determines a lifting of Frobenius log : {(X log )ord }F (X log )ord which is ordinary (by the conclusion of Lemma 3.1 cf. [Mzk1], Chapter IV, Proposition 4.10). Thus, by the general theory of ordinary Frobenius liftings, log determines, in particular, a (logarithmic) nite at group scheme annihilated by p which is an extension of the trivial nite at group scheme Fp by the nite at group scheme determined by the Cartier dual of some etale ord local system of one-dimensional Fp -vector spaces on Xk (cf. [Mzk1], Chapter III, Denition 1.6); denote the (X log )ord -module corresponding to this etale local system by Fp . Moreover, it is a formal consequence of this general thelog ord )K determined by considering ory that log K is precisely the covering of (X splittings of this extension. Thus, since the Galois closure of this covering has Galois group given by the semi-direct product of a cyclic group of order p with a cyclic group of order p 1, we conclude (by the elementary group theory of such a semi-direct product) that we have an isomorphism of (X log )ord -modules: (QFp )2 (1) = Fp (1), i.e., (QFp )2 = F p
We thus conclude that the local system EF on (X log )ord K determined by this p (logarithmic) nite at group scheme arising from the general theory satises: log ord |(X log )ord Fp Q EF = EF Fp p p
K log

22

Shinichi Mochizuki Next, let us write Q for the character (valued in Fp ) of (X log )ord corre

sponding to QFp . Now it follows formally from condition (iv) of the statement log ord )K determined by Ker(Q ) of Lemma 3.2 that the nite etale covering of (X
log ord WQ (X )K ord is dominated by the composite of some nite etale covering of X and a constant covering (i.e., a covering arising from a nite extension L of K ). Thus, the only ramication that may occur in the covering WQ arises from ramication of the constant covering, i.e., the nite extension L/K . Moreover, (since (QFp )2 = Fp ) the covering determined by Ker(2 Q ) is unramied, so, 1 in fact, we may take L to be the extension K (p 2 ). This implies that we may write

Q = Q Q where the covering determined by the kernel of Q (respectively, Q ) is nite ord etale over X (respectively, the covering arising from base-change from K to L). On the other hand, since Q extends naturally to X log , we may assume

(without loss of generality cf. condition (ii) of the statement of Lemma 3.2) ord that Q is trivial, hence that QFp arises from an etale local system on (X )k . log ord E But this together with the isomorphism E | log ord = F Q
Fp

implies the conclusion of Lemma 3.2.

(X )K

Fp

Fp

Lemma 3.3. (Global Logarithmic Finite Flat Group Scheme) Let VFp be a 2-dimensional Fp -vector space equipped with a continuous action of etale X log up to 1 which satises the hypotheses of Lemma 3.2. Then the log local system EFp on (X )K determined by the X log -module VFp arises from
log . a (logarithmic) nite at group scheme on X

Proof. Write Z X
log for the normalization of X in the nite etale covering of (X )K determined by the local system EFp . Since X is regular of dimension 2, it thus follows that Z is nite and at (by the Auslander-Buchsbaum formula and Serres criterion for normality cf. [Mtmu], p. 114, p. 125) over X . Moreover, since ZK is already equipped with a structure of (logarithmic) nite at group scheme, which, by the conclusion of Lemma 3.2, extends naturally over the

Canonical Curves

23

generic point of the special ber of X since it extends (by the proof of Lemma 3.2) to a regular, hence normal, (logarithmic) nite at group scheme over the ordinary locus it thus suces to verify that this nite at group scheme structure extends (uniquely) over the supersingular points of X . But this follows formally from [Mtmu], p. 124, Theorem 38 i.e., the fact (applied to X , not Z !) that a meromorphic function on a normal scheme is regular if and only if it is regular at the primes of height 1 and the fact that Z (hence also Z X Z ) is nite and at over X . Lemma 3.4. (The Associated Dieudonn e Crystal Modulo p) Let VFp be a 2-dimensional Fp -vector space equipped with a continuous action of X log up to 1 which satises the hypotheses of Lemma 3.2. Suppose, further, that the X log -module (up to 1) VFp satises the following condition: (M ) The GK -module M = H 1 (X log , Ad(VFp )) (where Ad(VFp ) End(VFp ) is the subspace of endomorphisms whose trace = 0) ts into an exact sequence: 0 G1 (M ) M G2 (M ) 0 where G2 (M ) (respectively, G1 (M )) is isomorphic to the result of tensoring an unramied GK -module whose dimension over Fp is equal to 3g 3 + r with the Tate twist Fp (2) (respectively, Fp (1)). Then the MF -object (up to 1) (cf. [Falt], 2) determined by the X log module (up to 1) VFp (cf. Lemma 3.3) arises from the (unique) nilpotent ordinary indigenous bundle of Lemma 3.1, (iv). Proof. First, we recall from the theory of [Falt], 2, that the conclusion of Lemma 3.3 implies that VFp arises from an MF -object (up to 1) on X log , as in the theory of [Falt], 2. Here, the reader uncomfortable with MF log objects up to 1 may instead work with a usual MF -object over X log equipped with an action of Gal(X /X log ) up to 1. Write Fk for the log vector bundle on (X )k underlying the MF -object on X determined by EF p . Thus, Fk is a vector bundle of rank 2, whose Hodge ltration is given by a subbundle F 1 (Fk ) Fk of rank 1. Moreover, the Kodaira-Spencer morphism of this subbundle, as well as the Hasse invariant
1 X F (Fk ) Fk def

Fk /F 1 (Fk ) = F 1 (Fk )

24

Shinichi Mochizuki

(where X is the Frobenius morphism on X ; and the injection is the morphism that arises from the Frobenius action on the MF -object in question), is generically nonzero. Indeed, these facts all follow immediately from our analysis of EFp over the ordinary locus in the proof of Lemma 3.2. Next, let us observe that this Hasse invariant has at least one zero. Indeed, if it were nonzero everywhere, it would follow formally from the general theory of MF -objects (cf. [Falt], 2) that the X log -module (up to 1) VFp admits a X log -invariant subspace of Fp -dimension 1 cf. the situation over the ordinary locus in the proof of Lemma 3.2, over which the Hasse invariant is, in fact, nonzero. On the other hand, this implies that YK XK admits a section, hence that Y is not connected. But this contradicts the fact (cf. the proof of Lemma 3.1) that the two irreducible components CV , CF of Yk (cf. Lemma 3.1, (iii)) necessarily meet at the nodes of Yk . (Here, we recall from Lemma 3.1, (ii), that there exists at least one node on Yk .) In particular, it follows from the fact that the Hasse invariant is generically nonzero, but not nonzero everywhere that the degree of the line bundle F 1 (F ) on (X )k is positive. Note that since F 1 (F )2 descends naturally to a line bundle L on Xk , we thus obtain that 1 deg(L) 2g 2 + r (where the second inequality follows from the fact that the Kodaira-Spencer morphism in nonzero). Now, we conclude from the p-adic Hodge theory of [Falt], 2; [Falt], 5, Theorem 5.3 that the condition (M ) implies various consequences concerning the Hodge ltration on the rst de Rham cohomology module of Ad of the MF -object determined by VFp , which may be summarized by the inequality: h1 (Xk , L1) = h0 (Xk , L OX X ) 3g 3 + r (where hi denotes the dimension over k of H i ) cf. [Mzk1], IV, Theorem 1.3 (and its proof), which, in essence, addresses the Zp analogue of the present Fp -vector space situation. (Note that here we make essential use of the hypothesis p 5.) Thus, (cf. loc. cit.) we conclude that (since deg(L) > 0) the line bundle L OX X on Xk is nonspecial, hence (by the above inequality) that:

deg(L) = deg(L OX X ) deg(X ) = h0 (Xk , L OX X ) + (g 1) 2(g 1) 3g 3 + r (g 1) = 2g 2 + r

Canonical Curves

25

Combining this with the inequalities of the preceding paragraph, we thus obtain that deg(L) = 2g 2 + r , so the MF -object in question is an indigenous bundle, which is necessarily equal to the indigenous bundle of Lemma 3.1, (iv), since the supersingular locus of the former is contained in the supersingular locus of the latter (cf. [Mzk1], Chapter II, Proposition 2.6, (4)). Lemma 3.5. (Canonical Deformations Modulo Higher Powers of p) Let VFp be a 2-dimensional Fp -vector space equipped with a continuous action of X log up to 1 which satises the hypotheses of Lemmas 3.2, 3.4. Suppose that, for some n 1: (i) (X, D) is isomorphic modulo pn to a canonical lifting (as in [Mzk1], Chapter III, 3; Chapter IV). (ii) VFp is the reduction modulo p of a rank 2 free Z/pn Z-module VZ/pn Z with continuous X log -action up to 1. Then (X, D) is isomorphic modulo pn+1 to a canonical lifting, and the X log set P(VZ/pn Z ) (of free, rank one Z/pn Z-module quotients of VZ/pn Z ) is isomorphic to the projectivization of the canonical representation modulo pn (cf. [Mzk1], Chapter IV, Theorem 1.1) associated to (X, D). Finally, if the determinant of VZ/pn Z is isomorphic to (Z/pn Z)(1), then the X log -module (up to 1) VZ/pn Z is isomorphic to the canonical representation modulo pn . Proof. First, we observe that the case n = 1 is a formal consequence of Lemmas 3.1, 3.4. The case of general n is then, in essence, a formal consequence of the theory of [Mzk1], Chapter V, 1 cf. especially, Theorem 1.7, and the discussion following it. We review the details as follows: The space of deformations (of the projectivization) of the X log -module up to 1 VFp may be thought of as a formal scheme R which is noncanonically isomorphic to Spf(Zp [[t1 , . . . , t2(3g 3+r)]]) (where the ti s are indeterminates) cf. the discussion preceding [Mzk1], Chapter V, Lemma 1.5. Write RPD for the p-adic completion of the PD-envelope of R at the Fp -valued point dened by VFp . Note that R and RPD are equipped with a natural GK -action. Then according to the theory of loc. cit., there is a GK -equivariant closed immersion of formal schemes (cf. Remark 3.5.1 below) PD : Spf(DGal ) RPD

26

Shinichi Mochizuki

where DGal is noncanonically isomorphic to the p-adic completion of the PDenvelope at the closed point of a power series ring Zp [[t1 , . . . , t3g 3+r ]] equipped with a natural GK -action. Now it follows from the theory of loc. cit., the induction hypothesis on n, def and the assumptions (i), (ii) of Lemma 3.5, that VZ/pn1 Z = VZ/pn Z Z/pn1Z corresponds (at least projectively) to the canonical representation modulo pn1 , hence determines a GK -invariant rational point n1 RPD (Z/pn1 Z) that lies in the image Im(PD ) of PD . Thus, the point n RPD (Z/pn Z) determined by VZ/pn Z may be regarded as a GK -invariant deformation of n1. Note that the set of deformations of n1 naturally forms a torsor T over the Fp -vector space M of Lemma 3.4. Since n1 Im(PD ), this torsor is equipped with a natural GK -stable subspace T T

(consisting of the deformations that lie Im(PD )) which is (by the theory of loc. cit.) a torsor over G1 (M ). In particular, this subspace determines a GK -invariant trivialization of the G2 (M )-torsor (T )T G2 (M ).

given by the change of structure group M

Now let us observe that by (M ) and the fact that p 5 so the square of the cyclotomic character GK Fp is nontrivial G2 (M ) has no nontrivial Galois invariants, i.e.: G2 (M )GK = 0 Thus, we conclude that is the unique GK -invariant point of T , hence that the image in T of the point n T determined by n is necessarily equal to , i.e., n T or, in other words, n Im(PD ). On the other hand, if we interpret the Galois-theoretic fact that n1 lifts to a GK -invariant n Im(PD ) in terms of the original nite etale coverings combinatorial information concerning which the Galois theory is intended to encode then we obtain the following conclusion: The A-valued point

Canonical Curves

27

N (A)
ord (cf. [Mzk1], Chapter III, 2) of the Zp -smooth p-adic formal scheme N = Ng,r determined by (X, D) and the nilpotent ordinary indigenous bundle modulo p of Lemma 3.4 not only lies by assumption (i) of the statement of Lemma 1 of the canonical 3.5 in the image of N (A) under the (n 1)-st iterate n N Frobenius lifting def

N : N N on (cf. [Mzk1], Chapter III, Theorem 2.8) but also in the image of N (A) unn1 der the n-th iterate n N of N . (Indeed, the restriction of the covering N n (respectively, N ) to admits a section, determined by the GK -invariant rational point n1 (respectively, n ).) Thus, we conclude that X log is canonical modulo pn+1 . Finally, since G1 (M )GK = 0 we conclude that n is the unique GK -invariant lifting of n1 to Z/pn Z, hence that VZ/pn Z corresponds (projectively) to the canonical representation modulo pn , as desired. Remark 3.5.1. In some sense, it is natural to think of Im(PD ) (cf. the proof of Lemma 3.5) as the crystalline locus in the space of all representations RPD . Theorem 3.6. (Group-Theoreticity of Canonical Liftings) Let p 5 be a prime number. For i = 1, 2, let Ki be an unramied nite extension of Qp , and (Xi , Di ) a smooth pointed curve of type (gi , ri ) over OKi , where 2gi 2 + ri > 0. Assume that we have chosen basepoints of the (Xi )Ki \(Di )Ki (which thus induce basepoints of the Ki ); denote the resulting fundamental group by X log . Suppose that we have been given an isomorphism of proi nite groups: X log X log
1 2

Then: (i) (X1 , D1 ) is a canonical lifting (in the sense of [Mzk1], Chapter III, 3; Chapter IV) if and only if (X2 , D2 ) is so. (ii) If at least one of the (Xi , Di ) is a canonical lifting, then the isomorphism on logarithmic special bers of [Mzk7], Theorem 2.7, lifts (uniquely) to an isomorphism (X1 , D1 ) (X2 , D2 ) over OK1 = W (k1 ) W (k2 ) = OK2 .

28

Shinichi Mochizuki

Proof. Let us verify (i). Since, by [Mzk7], Theorem 2.7 (and [Mzk7], Proposition 1.2.1, (vi)), the logarithmic special bers of all stable reductions of all nite etale coverings of (Xi )Ki \(Di )Ki are group-theoretic, it follows that the conditions (i), (ii), (iii), (iv) of Lemma 3.1, as well as the conditions (i), (ii), (iii), (iv) of Lemma 3.2, are all group-theoretic conditions which, moreover, (by the theory of [Mzk1], Chapter III, 3; Chapter IV) are satised by canonical liftings. (Here, relative to the assertion that canonical liftings satisfy Lemma 3.1, (ii), and Lemma 3.2, (iv), we remind the reader that: (1) Since p 5, the special ber of the curve Y of Lemma 3.1 has 2 nodes (cf. Lemma 3.1, (ii)), which implies that the curve Y is stable (i.e., even without the marked points). (2) The smooth locus of any model of a curve over a discrete valuation ring necessarily maps to the stable model (whenever it exists) of the curve cf., e.g., [JO] and, moreover, whenever this map is quasi-nite, necessarily embeds as an open subscheme of the smooth locus of the stable model.) Moreover, (since the cyclotomic character and inertia subgroup are grouptheoretic cf. [Mzk7], Proposition 1.2.1, (ii), (vi) it follows that) condition (M ) of Lemma 3.4 is a group-theoretic condition which (by the theory of loc. cit.) is satised by canonical liftings. Thus, successive application of Lemma 3.5 for n = 1, 2, . . . implies assertion (i) of the statement of Theorem 3.6. Next, let us observe that when one (hence both) of the (Xi , Di ) is a canonical lifting, it follows from Lemma 3.1 that the isomorphism of special bers (X1 , D1 )k1 (X2 , D2 )k2 (lying over some isomorphism k1 k2 ) determined by [Mzk7], Theorem 2.7 is compatible with the nilpotent ordinary indigenous bundles on either side that give rise to the canonical liftings. On the other hand, by the theory of loc. cit., the lifting of (Xi , Di )ki over OKi = W (ki ) is determined uniquely by the fact that this lifting is canonical (i.e., when it is indeed the case that it is canonical!). This completes the proof of assertion (ii) of the statement of Theorem 3.6. Remark 3.6.1. Thus, (cf. Remarks 1.1.1, 2.2.1) unlike the situation with Proposition 1.1, Theorem 2.2: Theorem 3.6 provides, for each hyperbolic (g, r ), lots of examples in particular, an example lifting a general curve of type (g, r ) in characteristic p 5 of (XK , DK ) which are group-theoretically determined solely by the pronite group 1 (XK \DK ).

Canonical Curves

29

In fact, the exact same arguments of the present 3 show that the analogue of Theorem 3.6 also holds for Lubin-Tate canonical curves (for various tones ) i.e., the curves dened by considering canonical points (cf. [Mzk2], Chapter VIII, 1.1) associated to the canonical Frobenius lifting of [Mzk2], Chapter VIII, Theorem 3.1, in the case of a VF-pattern of pure tone . One feature of the Lubin-Tate case that diers (cf. [Mzk7], Proposition 1.2.1, (vi)) from the classical ordinary case of [Mzk1] is that it is not immediately clear that the Lubin-Tate character : GK W (Fq ) (where Fq is a nite extension of Fp ) is group-theoretic. Note, however, that at least for the portion of modulo p GK Fq group-theoreticity at least after passing to a nite unramied extension of the original base eld K (which does not present any problems, from the point of view of proving the Lubin-Tate analogue of Theorem 3.6) is a consequence of the fact that the eld structure on (the union of 0 and) the torsion in Gab K of order prime to p is group-theoretic (cf. [Mzk7], Lemma 2.6, Theorem 2.7). This much is sucient for the Lubin-Tate analogues of Lemmas 3.1, 3.2, 3.3, 3.4, 3.5 (except for the last sentence of the statement of Lemma 3.5, which is, at any rate, not necessary to prove Theorem 3.6), and hence of Theorem 3.6. Finally, once one has proved that the curve in question is (Lubin-Tate) canonical and recovered its canonical representation VZp , at least projectively, then it follows, by considering the W (Fq )-module analogue of the exact sequence of Fp -vector spaces in the conclusion of Lemma 3.2, over the ordinary locus, that (even if one only knows VZp projectively) one may recover the Lubin-Tate character by forming

Hom(third nonzero term of the exact sequence, rst nonzero term of the exact sequence) at least up to un etale twist, i.e., up to multiplication by some unramied character GK W (Fq ) . (We leave the remaining routine technical details of the Lubin-Tate case to the enthusiastic reader.) At the time of writing, it is not clear to the author whether or not it is possible to eliminate this etale twist. Since this etale twist corresponds to the well-known dependence of the Lubin-Tate group on the choice of uniformizer, this indeterminacy with respect to an etale twist may be thought of as being related to the fact that (at the

30

Shinichi Mochizuki

present time) the author is unable to prove the group-theoreticity of the natural uniformizer p cf. [Mzk7], Remark 2.7.2. At any rate, the theory of the present 3 constitutes the case of tone 0. Moreover, one checks easily by considering the FL-bundle (as in [Mzk1], Chapter II, Proposition 1.2) determined by the lifting modulo p2 that canonical curves of distinct tones are never isomorphic. Thus, (at least when r = 0) the Lubin-Tate canonical curves give rise to (strictly) more examples of (proper) hyperbolic XK which are group-theoretically determined solely by the pronite group 1 (XK ). This prompts the following interesting question: Is this list complete? i.e.: Do there exist any other curves (XK , DK ) that are group-theoretically determined solely by the pronite group 1 (XK \DK )? At the time of writing, the author does not even have a conjectural answer to this question. Remark 3.6.2. One interesting aspect of the theory of the present 3 is that, to the knowledge of the author: It constitutes the rst application of the p-adic Teichm uller theory of [Mzk1], [Mzk2], to prove a hitherto unknown result (cf. Corollary 3.8 below) that lies outside i.e., can be stated without using the terminology, concepts, or results of the theory of [Mzk1], [Mzk2]. Moreover, not only does this constitute the rst application of p-adic Teichm uller theory to prove a new result (cf. the proof of the irreducibility of the moduli stack via p-adic Teichm uller theory in [Mzk2], Chapter III, 2.5 an application, albeit to an old result), it is interesting relative to the original philosophical motivation for this theory, involving the analogy to uniformization theory/Teichm uller theory over the complex numbers, which was to construct a p-adic theory that would allow one to prove a p-adic version of the Grothendieck Conjecture as in [Mzk6] cf. [Mzk2], Introduction, 0.10, for more on these ideas. Remark 3.6.3. One interesting point of view is the following: For a hyperbolic curve XK over a nite extension K of Qp , consideration of the pronite group XK should be thought of as the arithmetic analogue of considering a hyperbolic curve (which is given a priori) over Fp [[t]][t 1] where t is an indeterminate which, perhaps, should be thought of as the symbol p in the absolute, i.e., stripped

Canonical Curves of its structure morphism to a specic copy of Fp [[t]][t 1], or, alternatively, when we allow the indeterminate t to vary freely (in Fp [[t]]).

31

Thus, from this point of view, it is natural to expect that the hyperbolic curves XK most likely to be recoverable from the absolute datum XK are those which are dened over some (ctitious) absolute eld of constants inside K hence which have moduli that are invariant with respect to changes of variable t t + t2 + . . . . In particular, since one expects canonical curves to be arithmetic analogues of curves dened over the constant eld (cf. [Mzk2], Introduction, 2.3), it is perhaps not surprising that they should satisfy the property of Theorem 3.6. Moreover, this point of view suggests that: Perhaps it is natural to regard Theorem 3.6 as the proper analogue for hyperbolic canonical curves of the fact that Serre-Tate canonical liftings (of abelian varieties) are dened over number elds (cf. [Mzk2], Introduction, 2.1, Open Question (7)). Indeed, one way of showing that Serre-Tate canonical liftings or, indeed, arbitrary abelian varieties with lots of endomorphisms are dened over number elds is by thinking of the algebraic extension Qp of Qp over which such abelian varieties are a priori dened in the absolute, i.e., as a transcendental extension of Q and considering what happens when one transports such abelian varieties via arbitrary eld automorphisms of Qp . Such eld automorphisms are reminiscent of the changes of variable appearing in the approach to thinking aboutrecovering XK from the absolute datum XK described above. Remark 3.6.4. The rigidity of canonical curves in the sense that they are determined by the existence of the unique GK -invariant lifting VZp of the representation VFp (cf. Lemma 3.5 and its proof) is reminiscent, at least at a technical level, of the theory of deformations of representations applied in Wiles famous proof of the modularity conjecture (cf. [Wiles]). It would be interesting if this analogy could be pursued in more detail in the future. Remark 3.6.5. Just as the Serre-Tate canonical coordinates are used in [Mzk5], 9, to prove a weak p-adic Grothendieck Conjecture-type result for hyperbolic curves over p-adic elds whose Jacobians have ordinary reduction, the techniques of the present may be applied by using the canonical coordinates of [Mzk1] to prove a similar (but in some sense even weaker) p-adic Grothendieck Conjecture-type result for hyperbolic curves over absolutely unramied p-adic elds which are isomorphic to canonical curves (as in [Mzk1]) modulo p2 . Thus, the true signicance of the theory of the present lies in its wide applicability in the canonical lifting case (cf. Corollary 3.8 below),

32

Shinichi Mochizuki

a feature which diers substantially from the theory in the case of ordinary Jacobians (cf. Remark 1.1.1). Denition 3.7. Let YL be a hyperbolic curve over a nite extension L of Qp . Then we shall say that YL is absolute if for every other hyperbolic curve YL over a nite extension L of Qp , 1 (YL ) = 1 (YL ) (as pronite groups) implies that YL is isomorphic as a Qp -scheme to YL . Also, we shall refer to points in moduli stacks of hyperbolic curves over Qp that are dened by absolute hyperbolic curves as absolute. Corollary 3.8. (Application of p-adic Teichm uller Theory)

(i) A general pointed smooth curve (Xk , Dk ) of type (g, r ), where 2g 2 + r > 0, over a nite eld k of characteristic p 5 may be lifted to a pointed smooth curve (XK , DK ) over the quotient eld K of the ring of Witt vectors A = W (k) such that the hyperbolic curve XK \DK is absolute. (ii) In particular, for each (g, r ), p 5, there exists a Zariski dense hence (at least when 3g 3 + r 1) innite set of absolute points, valued in absolutely unramied nite extensions of Qp , of the moduli stack of hyperbolic curves of type (g, r ) over Qp . Proof. Assertion (i) follows formally from Theorem 3.6; [Mzk7], Lemmas 1.1.4, 1.1.5; and [Mzk7], Proposition 1.2.1, (v). Assertion (ii) follows formally from assertion (i) and the following elementary argument: The scheme-theoretic closure of the points of assertion (i) in the (compactied) moduli stack over Zp forms a Zp -at proper algebraic stack Z with the property that Z Fp is equal to the entire moduli stack over Fp , hence is smooth of dimension 3g 3 + r . But this implies by Zp -atness that Z Qp is also of dimension 3g 3+ r , hence equal to the entire moduli stack over Qp , as desired.

Canonical Curves Bibliography

33

Unpublished RIMS preprints are available as .ps les at: http://www.kurims.kyoto-u.ac.jp/ kenkyubu/paper/all.html [DM] P. Deligne and D. Mumford, The Irreducibility of the Moduli Space of Curves of Given Genus, IHES Publ. Math. 36 (1969), pp. 75-109. [DO] B. Dwork and A. Ogus, Canonical liftings of Jacobians, Compositio Math. 58 (1986), pp. 111-131. [Edix] B. Edixhoven, On the Andr e-Oort conjecture for Hilbert modular surfaces, Moduli of abelian varieties (Texel Island, 1999), Progr. Math. 195, Birkh auser (2001), pp. 133-155. [Falt] G. Faltings, Crystalline Cohomology and p-adic Galois Representations, Proceedings of the First JAMI Conference, Johns Hopkins University Press (1990), pp. 25-79. [FC] G. Faltings and C.-L. Chai, Degenerations of Abelian Varieties, SpringerVerlag (1990). [Groth] A. Grothendieck, Letter to G. Faltings (June 1983) in Lochak, L. Schneps, Geometric Galois Actions; 1. Around Grothendiecks Esquisse dun Programme, London Math. Soc. Lect. Note Ser. 242, Cambridge Univ. Press (1997). [JO] A. J. de Jong and F. Oort, On Extending Families of Curves, Journal of Alg. Geom. 6 (1997), pp. 545-562. [Kato] K. Kato, Logarithmic Structures of Fontaine-Illusie, Proceedings of the First JAMI Conference, Johns Hopkins University Press (1990), pp. 191224. [Knud] F. F. Knudsen, The Projectivity of the Moduli Space of Stable Curves, II, Math. Scand. 52 (1983), 161-199. [Mtmu] H. Matsumura, Commutative Algebra (Second Edition), The Benjamin/Cummings Publishing Company (1980). [Mess] W. Messing, The Crystals Associated to Barsotti-Tate Groups; with Applications to Abelian Schemes, Lecture Notes in Mathematics 264, SpringerVerlag (1972). [Mzk1] S. Mochizuki, A Theory of Ordinary p-adic Curves, Publ. of RIMS 32 (1996), pp. 957-1151. [Mzk2] S. Mochizuki, Foundations of p-adic Teichm uller Theory, AMS/IP Studies in Advanced Mathematics 11, American Mathematical Society/International Press (1999).

34

Shinichi Mochizuki

[Mzk3] S. Mochizuki, Correspondences on Hyperbolic Curves, Journ. Pure Appl. Algebra 131 (1998), pp. 227-244. [Mzk4] S. Mochizuki, The Geometry of the Compactication of the Hurwitz Scheme, Publ. of RIMS 31 (1995), pp. 355-441. [Mzk5] S. Mochizuki, The Pronite Grothendieck Conjecture for Closed Hyperbolic Curves over Number Fields, J. Math. Sci., Univ. Tokyo 3 (1996), pp. 571-627. [Mzk6] S. Mochizuki, The Local Pro-p Anabelian Geometry of Curves, Invent. Math. 138 (1999), pp. 319-423. [Mzk7] S. Mochizuki, The Absolute Anabelian Geometry of Hyperbolic Curves, RIMS Preprint No. 1363 (June 2002). [NTM] H. Nakamura, A. Tamagawa, and S. Mochizuki, The Grothendieck Conjecture on the Fundamental Groups of Algebraic Curves, Sugaku Expositions 14 (2001), pp. 31-53. [OS] F. Oort and T. Sekiguchi, The canonical lifting of an ordinary Jacobian variety need not be a Jacobian variety, J. Math. Soc. Japan 38 (1986), pp. 427-437. [SGA1] Rev etement etales et groupe fondamental, S eminaire de G eometrie Alg ebrique du Bois Marie 1960-1961 (SGA1), dirig e par A. Grothendieck, augment e de deux expos es de M. Raynaud, Lecture Notes in Mathematics 224, Springer-Verlag (1971). [Take1] K. Takeuchi, Arithmetic Triangle Groups, Journ. Math. Soc. Japan 29 (1977), pp. 91-106. [Take2] K. Takeuchi, Arithmetic Fuchsian Groups with Signature (1; e), Journ. Math. Soc. Japan 35 (1983), pp. 381-407. [Tama] A. Tamagawa, The Grothendieck Conjecture for Ane Curves, Compositio Math. 109 (1997), pp. 135-194. [Vala] P. Valabrega, A Few Theorems on Completion of Excellent Rings, Nagoya Math J. 61 (1976), pp. 127-133. [Wiles] A. Wiles, Modular elliptic curves and Fermats last theorem, Ann. of Math. 141 (1995), pp. 443-551. Research Institute for Mathematical Sciences Kyoto University Kyoto 606-8502, JAPAN