Deformations ofAbelian Varietiesand SerreTate Theory DrewMoore Deformations of Abelian Varieties and Deformations ofSchemes Serre Tate Theory Deformations ofAbelian Varieties Deformations ofp-Divisible Groups Drew Moore SerreTate Coordinates November 10 Context of this Talk within the Seminar Deformations ofAbelian The main players we’ve been dealing with in this seminar have been Varietiesand SerreTate algebraic varieties over Fp, Zp, Qp, formal schemes over Zp, and rigid Theory analytic varieties over Q , and their analogues for extensions of F p p DrewMoore and Q . Just as a reminder, these categories sit in a commutative p Deformations diagram ofSchemes DofeAfobrmeliaatnions Sch/Zp Sch/Qp Varieties Deformations Sch/Fp ofp-Divisible Groups FSch/Zp Rig/Qp SerreTate Coordinates For any scheme (resp. formal scheme) X over SpecZ (resp. SpfZ ), p p we have its specialization map X→X to its special fiber X ∈Sch/Fp. This talk: take X to be a moduli space of abelian varieties. Pick a closed point x in X (corresponding to an ab var A ). Study the 0 0 preimage of x under the specialization map, i.e. its deformations to 0 characteristic 0. Deformations ofAbelian Varietiesand In Matt’s second talk this quarter, he described the rigid generic fiber SerreTate Theory functor FSch/Zp →Rig/Qp only for p-adic formal schemes, which was DrewMoore Raynaud’s original construction. I.e., those formal schemes for which the topology on the open affines is given by the ideals (pn). Deformations ofSchemes Deformations Example: SpfZ (cid:104)(cid:104)T(cid:105)(cid:105)(cid:32) the closed unit polydisk. ofAbelian p i Varieties The preimage of a point in the special fiber of this example is an Deformations ofp-Divisible open unit polydisk. Would like a formal model for this. Berthelot Groups SCeororredTinaattees extended this functor to all of FSch/Zp, which will solve this. Example: SpfZ [[T]], with ideal of definition (p,T). Then its rigid p i i generic fiber is the open unit polydisk. Conclusion: we expect that the deformations of x to characteristic 0 0 will be parametrized by something like SpfZ [[T]]. p i The Setup: Coefficients and Categories Deformations ofAbelian Let k be a field and W a complete local ring with residue field k. For Varietiesand SerreTate example: Theory k =W =C DrewMoore k =F and W =Z Deformations p p ofSchemes k =F , W =W(F ) Deformations p p ofAbelian Let C be the category Varieties Deformations ofp-Divisible Artin local rings R with maxop Groups C = ideal m equipped with an SerreTate R Coordinates isomorphism R/m →k R I will call C the category of Artin schemes. Similarly C(cid:98)is the category defined similarly, but replace “Artin” with “complete Noetherian.” Objects of C(cid:98)are inductive limits of objects of C, and are affine formal schemes. Why C and C(cid:98)? Deformations ofAbelian Suppose we are in the situation k =W =C. Let E be an elliptic Varietiesand 0 SerreTate curve over k =C. Theory DrewMoore E :y2 =x3+a x +b a ,b ∈C 0 0 0 0 0 Deformations ofSchemes We want to study deformations of E . A deformation over C[[t]] Deformations 0 ofAbelian could be something like Varieties Deformations ofp-Divisible E :y2 =x3+a(t)x +b(t) a(0)=a ,b(0)=b Groups 0 0 SerreTate Coordinates We could imagine that a and b have a positive radius of convergence, so then plugging in sufficiently small values of t would give “deformations” E of E . t 0 SpfC[[t]] is an object of C(cid:98). To study deformations of E0 to C[[t]], it is easier to first study deformations to the intermediate rings C[t]/(tn), which correspond to schemes in C. Why C and C(cid:98)? Deformations k will be “where we start” - i.e., we will have an object over k, and ofAbelian Varietiesand we will aim to “deform” it over W. For example, if k =W =F , SerreTate p Theory then we will be restricting to deformations that are characteristic p. DrewMoore But if W =Z , the deformations could be mixed characteristic. p Deformations ofSchemes Geometrically, the isomorphism R/mR →k allows us to concretely Deformations pick out the “special fiber” over S =SpecR. ofAbelian Varieties Closed embeddings S →S(cid:48) in C are called thickenings. If the kernel Deformations ofp-Divisible of the corresponding ring map R(cid:48) →R has square zero, then S →S(cid:48) Groups SerreTate is said to be first order. Examples: Coordinates SpecC[t]/(tn)→SpecC[t]/(tn+1) SpecZ/pnZ→SpecZ/pn+1Z Every thickening is a composition of first order thickenings. Fact: C(cid:98)is equivalent to the category of formal schemes equipped with an isomorphism of its special fiber with the point Speck. Deformation Definitions, Definitely Deformations ofAbelian Suppose that X0 is a smooth scheme over k. Let S =SpecR be an Varietiesand SerreTate element of C. An infinitesimal deformation of X0 to S is a smooth Theory S-scheme X →S along with an isomorphism of the special fiber of X DrewMoore with X : 0 Deformations X −∼→= X × k ofSchemes 0 S DofeAfobrmeliaatnions Suppose S →S(cid:48) =SpecR(cid:48) is a thickening in C, and X is smooth Varieties over S. Then a deformation along S →S(cid:48) is a smooth scheme X(cid:48) Dofepfo-rDmivaitsiiobnles over S(cid:48) along with an isomorphism X ∼=X(cid:48)×S(cid:48) S. Groups SerreTate Current goal: understand when we can deform X to S, and if so, in Coordinates 0 how many different ways? We will do this by studying deformations along first order thickenings. Namely, we are interested in the following contravariant functor: Def : C →Sets S (cid:55)→{deformations of X to S}/∼= X0 0 Cohomology and Deformations Deformations Theorem ofAbelian Varietiesand Suppose S →S(cid:48) is a first order thickening, and X is a deformation of SerreTate Theory X to S. Let I be the kernel of R(cid:48) →R. Then: 0 DrewMoore There exists an obstruction Deformations ofSchemes ω(X/S)∈H2(X,T ⊗I) Deformations X/S ofAbelian Varieties which is zero if and only if there exists X(cid:48)/S(cid:48) deforming X to S(cid:48). Deformations ofp-Divisible Moreover, ω is natural with respect to isomorphisms in X/S. Groups SerreTate If ω(X/S)=0, then the set of deformations of X to S(cid:48) (up to Coordinates isomorphism) can be put in bijection with H1(X,T ⊗I) X/S If X(cid:48) is a deformation of X along S →S(cid:48), then the infinitesimal automorphisms of X(cid:48) over X are naturally identified with H0(X,T ⊗I) X/S Proof in special case, idea of proof Deformations ofAbelian For simplicity, suppose we are in the case S =Speck and Varietiesand SerreTate S(cid:48) =Speck[(cid:15)]. Let X/k[(cid:15)] be a deformation of X0/k. Then the ideal Theory sheaf of X (cid:44)→X is just ((cid:15)) considered as an O -module. Hence, the 0 X DrewMoore conormal bundle of X in X is ((cid:15))∼=O 0 X0 Deformations ofSchemes Hence the conormal sequence gives (using ((cid:15))∼=O ), Deformations X0 ofAbelian Varieties X i X Deformations 0 ofp-Divisible (cid:32) 0→O →i∗Ω →Ω →0 Groups X0 X/k[(cid:15)] X0/k SerreTate Coordinates Speck Speck[(cid:15)] On the right, we have an extension of Ω by O , and thus an X0/k X element of Ext1 (Ω ,O )∼=H1(X ,T ). OX0 X0/k X0 0 X0/k Idea of proof: first prove it when X is affine (i.e. there exists a unique deformation), and then use Cech cocycles in the general case. Abelian Schemes Deformations ofAbelian Let S be a scheme (think S ∈C). An abelian scheme over S is an Varietiesand SerreTate S-group scheme A→S which is proper, flat, finitely presented, and Theory has smooth geometrically connected fibers. DrewMoore Deformations When S =Speck is the spectrum of a field, an abelian scheme A /S ofSchemes 0 is just an abelian variety. Deformations ofAbelian Varieties An infinitesimal deformation of A to S ∈C is an abelian scheme 0 Deformations A/S over S along with an isomorphism of the special fiber ofp-Divisible Groups A −∼→= A× S. Important: we are deforming not just the scheme A , SerreTate 0 k 0 Coordinates but the group structure too. Consider the following contravariant functor: (cid:26) (cid:27) infinitesimal deformations Def : C →Sets S (cid:55)→ /∼= A0 of A to S 0
Description: