sign in sign up
<<
Home , Category of sets, Derived functor, Direct image functor, Forgetful functor, Functor, Grothendieck topology

?WHAT IS... a ? Luc Illusie

Basic Examples restriction maps with a family of sections si of E In the early 1960s Grothendieck chose the Greek on the Ui ’s which coincide on the intersections word topos (which means “place”) to denote a math- Ui ∩ Uj . Now, let C be a having finite pro- ematical object that would provide a general frame- jective limits. To give a (sometimes called work for his theory of étale and other a ) on C means to specify, variants related to his philosophy of . Even for each object U of C, families of maps (Ui → U)i∈I, if you do not know what a topos is, you have surely called covering families, enjoying properties anal- come across some of them. Here are two examples: ogous to those of open covers of an open (a) The category of sheaves of sets on a topo- of a topological , such as stability under logical space is a topos. In particular, the category change and composition (see [SGA 4 II 1.3] for a pre- of sets is a topos, for it is the category of sheaves cise definition). Once a topology has been chosen of sets on the one point space. This topos, denoted on C, C is called a site, and one can define a {pt}, is called the punctual topos. of sets on C in the same way as in the case in which (b) Let G be a . The category BG of G-sets, C is the category of open of a topological i. e., sets equipped with a left action of G, is a space: a sheaf of sets E on C is a contravariant func- topos. For G = {1}, BG = {pt}. tor U → E(U) on C (with values in the category of What these categories have in common is that (i) they behave very much like the , sets) having the property that for any covering → and (ii) they possess a good notion of localization. family (Ui U)i∈I, a s of E on U, i. e., an In order to formalize (ii), Grothendieck conceived element of E(U), can be identified via the “restric- the idea of sheaf on a site, which generalizes the tion” maps with a family of sections si of E on the notion of sheaf on a . That led him Ui ’s that coincide on the “intersections” Ui ×U Uj. to the notion of topos, which encompasses (i) and A topos T is a category equivalent to the cate- (ii). gory of sheaves of sets on a site C (which is then Sites and called a defining site for T). Here are some prop- Consider the category C of open subsets of a erties of toposes : topological space X (the being the in- (1) A topos T admits finite projective limits: in clusions of open subsets). A sheaf of sets E on X particular, it has a final object, and it admits fibered is a contravariant U → E(U) on C (with val- products. ues in the category of sets) having the property that (2) If (Ui)i∈I is a family of objects of T, the sum for any open (Ui)i∈I of U, a section s of E on i∈I Ui exists, is “disjoint”, and commutes with U, i. e., an element of E(U), can be identified via the base change. (3) Quotients by equivalence relations exist and Luc Illusie is professor of mathematics at the Université have the same good properties as in the category Paris-Sud. His email address is luc.illusie@math. of sets. u-psud.fr. The author wishes to thank P. Deligne, O. Gabber, A. Jackson, G. Laumon, W. Messing, A. Ogus, A theorem of Giraud [SGA 4 IV 1.2] asserts that and M. Raynaud for their help in the preparation of this the converse is essentially true. Namely, if T is a column. category satisfying (1), (2), (3), and if moreover T

1060 NOTICES OF THE AMS VOLUME 51, NUMBER 9 satisfies a certain technical “smallness” condition, admits enough injectives, and one can consider then T is a topos. the of Γ (T,−). This yields a com- Several unequivalent sites may give rise to the mon generalization of on topo- same topos, as the case of {pt} already shows : both logical spaces and (the functor the one point space and the category of sets, Γ (BG,−) is “taking the invariants under G”). equipped with the topology defined by surjective Toposes Arising from Algebraic families, are defining sites. Grothendieck liked to The most important example, which was the compare this with the fact that a group can be de- main motivation for Grothendieck and which is fined by generators and relations in many differ- also the closest to geometric intuition, is the étale ent ways. The site is some kind of system of gen- topos of a X. It is the topos of sheaves on erators and relations for the topos. And in the the étale site Xet of X. The underlying category of same way in which a group G can be defined by a Xet is the category of schemes Y étale over X (i. e., of generators that is G itself, a topos T can be étale morphisms Y → X). Étale morphisms are the defined by a site whose underlying category is T analogs, in , of morphisms of itself. Covering families are just epimorphic fam- complex analytic spaces that are analytically local ilies. This topology is called the canonical topology. . Covering families of Xet are sur- In the case of BG, the canonical topology is the jective families (Yi → Y )i∈I of (étale) X-schemes. topology defined by surjective families of G-maps. When X is the spectrum of a k with absolute Morphisms, Points, Cohomology Galois group G, the étale topos of X is a variant of A continuous of topological spaces BG, namely the category of (discrete) sets endowed ∗ f : X → Y defines a pair of adjoint (f ,f∗) with a continuous action of G; cohomology in this between the categories of sheaves of sets on X case is the Galois cohomology of k. It is a miracle and Y. The inverse functor f ∗ commutes (as that the consideration of the cohomology of étale a left adjoint) with inductive limits. It also com- toposes with values in Z/n, with n prime to the mutes with finite projective limits. Now, if X, Y are characteristics, has given rise to a Weil cohomology toposes, one defines a f : X → Y as a pair and, eventually, to the proof of the , ∗ of (f : Y → X,f∗ : X → Y) such by Grothendieck and Deligne [D]. Variants of the that f ∗ commutes with finite projective limits. If étale topology, such as the so-called fppf topology, T is a topos, up to a unique of func- play an important role in certain moduli problems tors, there is only one morphism from T to the (Hilbert and Picard schemes, etc.) and also in the punctual topos {pt}. On the other hand, a point of theory of group schemes and arithmetic applica- T is by definition a morphism from {pt} to T. If T tions. is defined by a topological space X, a (usual) point Another interesting example is the crystalline x of X defines a point of T, whose inverse image topos, constructed by Grothendieck and Berthelot, functor is the functor E → Ex. But new phe- which is crucial in differential calculus and the nomena occur. Deligne has constructed examples study of in positive or mixed of toposes without points (he has also given crite- characteristic. The comparison between crystalline ria for the existence of “enough points”) [SGA 4 IV cohomology and p-adic étale cohomology, some- 7, VI 9]. Moreover, if x and y are points of a topos times called p-adic [P], is closely re- T, there may exist nontrivial morphisms (of func- lated to deep problems in . tors) from x to y. In the case of BG, for example, Finally, let me mention that, in the wake of pi- the from BG to {pt} is the inverse oneering work of Lawvere in the late 1960s, a vari- image functor by a point of BG, whose group of ant of the notion of topos, called elementary topos, is G itself! This observation is at has been extensively used in logic for the past the root of Grothendieck’s theory of the funda- thirty years. mental group in algebraic geometry. If X is a topo- logical space and P is a G-torsor on X (i. e., a prin- References cipal G-cover of X), then twisting a G-set E by P (i. [SGA 4] Théorie des topos et cohomologie étale e., forming the corresponding space with fiber des schémas, Séminaire de Géométrie Algébrique E over X) is the by a mor- du Bois-Marie 1963-64, dirigé par M. Artin, phism fP : X → BG, and P → fP establishes a bijec- A. Grothendieck, J.-L. Verdier, SLN 269, 270, 305, tive correspondence between isomorphism classes Springer-Verlag, 1972, 1973. of G-torsors on X and morphisms from X to BG. [D] P. Deligne, La conjecture de Weil, I, Publ. Thus BG plays the role of a . Math. IHÉS 43 (1974), 273–308 ; La conjecture de If T is a topos, the for the Weil, II, Publ. Math. IHÉS 52 (1980), 137–252. unique morphism T →{pt} associates to an object [P] Cohomologies p-adiques et applications arith- E of T the set of its “global sections” métiques, (I), (II) (P. Berthelot, J.-M. Fontaine, Γ (T,E) = Hom(e, E) , where e is the final object of L. Illusie, K. Kato, M. Rapoport, eds), Astérisque T. The category of objects of T 278, 279, SMF, 2002; (III), to appear.

OCTOBER 2004 NOTICES OF THE AMS 1061