HAL Id: hal-01768672
https://hal.archives-ouvertes.fr/hal-01768672
Submitted on 19 Apr 2018
HAL is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are pub- lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.
Schemes of dimension 2: obstructions in non Abelian cohomology
Bénaouda Djamai
To cite this version:
Bénaouda Djamai. Schemes of dimension 2: obstructions in non Abelian cohomology. Pure Mathe- matical Sciences, 2017, 6, pp.39 - 45. �10.12988/pms.2017.711�. �hal-01768672�
Obstructions in
Non Abelian Cohomology
Bénaouda DJAMAI UFR de Mathématiques
Université de Lille 1 Sciences et Technologies
F-59665 Villeneuve d’Ascq Cedex France Abstract
Let f : X → Y be a proper and a flat morphism with fibers of dimension1 andX regular of dimension 2. Suppose that the geometric fibers off are connected, and the generic fiber is smooth. In this paper we consider a reductive group G onX and use results of Artin-Tate to find the obstruction for aG-gerb onX to come from anf∗G-gerb onY. Subject Classification: 14F20 Keywords:Non Abelian Cohomology, Gerbs, Obstruction
1 Introduction
Let X be regular scheme of dimension 2 and Y a smooth irreducible curve defined over a perfect field k or the spectrum of the ring of integers of a number field. Suppose that the generic fiber is smooth and that the geometric fibers are connected
We will consider etale topology on X and Y and suppose f∗(OX) ' OY. This is a condition of normalization, because if it is not met, we can consider Spec (f∗(OX)) instead ofY.
This condition implies, in particular, f∗(Gm,X)'Gm,Y.
LetAbe an abelianX-group. From the Leray spectral sequence associated to f, Hp(Y, Rqf∗(A)) ⇒ Hp+q(X, A), we obtain the following exact sequence in lower dimensions:
0 //H1(Y, f∗(A)) //H1(X, A) //H0(Y, R1f∗(A))
//H2(Y, f∗(A)) //H2(X, A)tr //H1(Y, R1f∗(A))
//H3(Y, f∗(A))
(1)
2 Bénaouda DJAMAI
whereH2(X, A)tr :=Ker{H2(X, A)−→H0(Y, R2f∗A)}
In case A =Gm,X and under our assumptions, we have R2f∗Gm,X = 0 by [12, Artin’s Theorem(3.2)], and therefore H2(X, Gm,X)tr = H2(X, Gm,X). So (1) becomes:
0 //P ic(Y) //P ic(X) //P ic(X/Y)
//Br(Y) //Br(X) //H1(Y, P)
//H3(Y, Gm,Y) //H3(X, Gm,X)
(2)
whereP =R1f∗(GmX) etP ic(X/Y) =H0(Y, P), which finally we write 0 //S //Br(Y) //Br(X) //H1(Y, P) //T //0
whereS =Coker(P ic(X)−→P ic(X/Y)andT =Ker{H3(Y, Gm)−→H3(X, Gm)}
(for example, iff admits a section, S =T = 0).
We will now study the analogous of this exact sequence when A is not abelian. More precisely, we will consider a reductive group G and reduce its 2-cohomology to a maximal torusT.
Let L be a lien locally, for the etale topology, represented by a reductive groupG. We know by [3, Prop 3.2 p 75] thatL is represented by a quasi-split reductiveX-groupGL, and therefore, on some extensionGLadmits a maximal torusTL isomorphic to a product of d copies of the multiplicative groupGm.
We can suppose then, by taking a finite etale extension, that G = Ge is a split X-group scheme and thatL is represented by G.e
For the next, we setH2(·,G) =e H2(·, lien(G))e andH2(X,G)e tr the subset of H2(X,G)e composed by the classes of gerbs G bound by lien(G), such thate there exists a refinement R of Y such that, for every U ∈ Ob(R), the fiber of G in f−1(U)is not empty (cf. [6], Chapitre V,3.1.9.3).
LetZ(G)e be the center of Ge and consider next diagram:
H0(Y, R1f∗Z(G))e //
H2(Y, f∗Z(G))e
//H2(X, Z(eG))tr
i(2)∗
//H1(Y, R1f∗Z(eG))
H0(Y, R1f∗Te) //
H2(Y, f∗Te)
O //H2(X,Te)tr
O //H1(Y, R1f∗Te)
O
H0(Y, R1f∗G)e H2(Y, f∗G)e //H2(X,G)e tr //?
(3)
where // is the Giraud’s relation defined in [6], and the question mark is becauseH1(k, R1π∗GX) does not exist, since R1π∗GX has no more a group structure.
By [4, Cor.(3.4) P 74], for every geometric point of y ∈Y, all the classes of the fiber(Ret2f∗G)y are trivial. Hence:
Proposition 1.1. H2(X,G)e tr =H2(X,G).e
2 Main results
We can now complete the last line of diagram (3):
H2(Y, f∗G)e //H2(X,G)e tr =H2(X,G)e
by defining an obstruction set to pull back a class fromH2(X,G)e toH2(Y, f∗G).e Consider diagram:
H2(Y, f∗Z(G))e u //H2(X, Z(G))e tr
λ //H1(Y, R1f∗Z(G))e
µ
H2(X,Te)tr =Br(X)d
//H1(Y, R1f∗eT) =H1(Y, P)d
H2(Y, f∗eG) //H2(X, eG)tr //?
(4) Let q ∈ H2(X,G)e tr. Because H2(X,G)e tr is stable under the simply tran-
sitive action of H2(X, Z(G))e tr ([6]), we have q = α·ε where ε = [T ors G]e ∈ H2(X,G)e and α∈H2(X, Z(G)) =e H2(X, Z(G))e tr.
Theorem 2.1. The map
H2(X,G)e −→Im(λ)⊂H1(Y, R1f∗Z(G))e q λ(α)
defines the appropriate obstruction, i.e, the sequence
H2(Y,G)e −→H2(X,G)e −→Im(λ)−→0 is exact.
It should be noted in passing that when Im(λ) = 0, we find proposi- tion(3.1.10), chap.V of [6], and more precisely Corollary(3.1.10.3).
In particular:
4 Bénaouda DJAMAI
Corollary 2.2. Under assumption of corollary [4, Cor.(3.4) P 74], if the fibers of f :X −→Y are rational, then every class q∈ H2(X,G)e comes from a class q0 ∈H2(Y, f∗G)e
Proof. For every point y∈Y, by setting f−1(y) =Fy, we have : (R1f∗ZG)e y = lim←−−
U3y
H1(f−1(U), Z(G)) =e H1(Fy, Z(G))e 'Hom
π1(Fy), f∗Z(G)e
whereU runs over etale neighborhoods of y.
But Hom
π1(Fy), f∗Z(G)e
=Hom
π1(P1), f∗Z(G)e
= 0.
Corollary 2.2 extends results of corollaries(3.13) and (3.14 of [4], where the residue fieldk where supposed algebraically closed or finite.
Now let µ : H1(Y, R1f∗Z(G))e −→ H1(Y, R1f∗(Te)). We can consider also as set of obstruction the imageIm(µ◦λ)⊂H1(Y, R1f∗(Te)).
Proposition 2.3. The sequence
H2(Y,G)e −→H2(X,G)e −→Im(µ◦λ)−→0
is exact. Obstruction defined by Im(µ◦λ) is the same as the one defined by Im(λ).
However, we will see in theorem(2.5) later that it is more flexible then that defined byIm(λ).
Proof. If q = α· ε ∈ H2(X,G)e is pulled back to q0 = β ·ε0 ∈ H2(Y, f∗G)e with ε0 = [T ors(Y, f∗G)]e and β ∈ H2(Y, f∗Z(G)), thene λ(α) =λ(u(β)) = 0 ∈ H1(Y, R1f∗Z(G)), hencee µ◦λ(α) = 0.
Conversely, suppose that µ◦λ(α) = 0.
We have a canonical isomorphism of sheaves of sets between R1f∗(Te) and the sheaf of maximal sub-gerbs of the direct image of the gerbT ors(X,Te)by f [6, Lemme 3.1.5].
Moreover, if [G] ∈ H2(X,Te) (resp. H2(X,Te)tr) is the class of an X-gerb G, the sheaf Ger(G)of maximal sub-gerbs of the direct image f∗(G) ofG par f is a pseudo-torsor (resp. a torsor) underR1f∗(Te) [6, Ex. 3.1.9.2].
letGTe be the gerb representing the image of α in H2(X,Te)tr =H2(X,Te).
Since (µ◦λ)(α) = 0, the sheaf Ger(G
Te) of maximal sub-gerbs of the direct image f∗(GTe) admits a section. If G represents the class q, there exists an X-morphism of gerbs:
GTe −→ G induced by the inclusion:
T ,e →Ge
The gerb morphism G
Te −→ G induces in turn anY-morphism of sheaves Ger(G
Te)−→Ger(G) whereGer(G) is the sheaf of maximal sub-gerbs of f∗G
So every Y-section ofGer(G
Te)induces anY-section of Ger(G), i.e,q can be pulled back to a class in H2(Y,G).e
Under assumptions of [4, Cor.(3.4) P 74]), suppose now that f admits a section. We have then the following exact sequence:
0 //Br(Y) //Br(X) //H1(Y, P) //H3(Y, Gm) //H3(X, Gm,X)
{{
which, compared with exact sequence [12, Theorem(3.1)]:
0→Br(Y)→Br(X)→X1(K, J)→0 provides the isomorphism: X1(K, J)'H1(Y, P)
Here, K is the function filed of Y, J is the jacobian variety of the generic fiber andX1 is the Tate-Shafarevich group.
From diagram (4), we get :
Im(µ◦λ)⊂H1(Y, R1f∗(Te)'H1(Y, Pd)'X1(K, J)d Theorem 2.4. In the exact sequence:
H2(Y, f∗G)→H2(X, G)→Im(µ◦λ)→0
the sub-group Im(µ ◦ λ) ⊂ X1(K, J)d is a sub-group of obstruction to a class q = α ·ε ∈ H2(X, G) to come from a class in H2(Y, f∗G). The map H2(X,G)e −→Im(µ◦λ)can be interpreted as the application which associates to a class G the class of the sheaf of maximal sub-gerbs Ger(GTe) of the direct image of G
Te by f.
One can summarize this situation in the following diagram : G ∈H2(X,G)e
++
//[Ger(G
Te)]∈Im(µ◦λ)⊂H1(Y, R1f∗Te))
Ger(G)∈?
Let’s recall that the Artin-Tate conjecture states thatX1(K, J)is finite. Our obstructions takes then values in a finite group. Therefore, if the order Z(G)e and the order of X1(K, J) are coprime, this obstruction vanishes. Hence:
Theorem 2.5. Under Artin-Tate conjecture, if |Z(G)|e and the order of X1(K, J) are coprime, where K denotes the functions field of Y, then all the classes inH2(X,G)e come from H2(Y, f∗G).e
Example 2.6. If Ge = SLn, we have: Te = (Gm)n−1 and Z(G) =e µn. So there is an infinity of integers for which theorem 2.5 is true.
6 Bénaouda DJAMAI
References
[1] M. Demazure & A. Grothendieck. Propriétés générales des schémas en groupes, volume 151 of Lecture Notes in Mathematics. Springer Verlag, 1970.
[2] Jean-Claude Douai. Cohomologie galoisienne des groupes semi-simples definis sur les corps globaux. C.r Acad. Sci, A 281:1077–1080, 1975.
[3] Jean Claude Douai. 2-cohomolgie galoisienne des groupes semi-simples.
Th‘ese de doctorat, Université de Lille1, 1976.
[4] Jean Claude Douai. Suites exactes déduites de la suite spectrale de leray en cohomoligie non abélienne. Journal of Algebra, 79:68–77, 1981.
[5] Jean Giraud. Méthode de la descente. Mémoires de la S.M.F, 2, 1964.
[6] Jean Giraud. Cohomologie non abelienne, volume 179 of Grundlheren Math. Wiss. Springer-Verlag, 1971.
[6] A. Grothendieck.Revetements étales et groupe fondamental,Lecture Notes in Mathematics. 224 Springer Verlag, 1971.
[7] A. Grothendieck. Groupe de Brauer II. Dix exposés sur la cohomologie des schémas, 1968.
[7] A. Grothendieck. Groupe de Brauer III. Dix exposés sur la cohomologie des schémas, 1968.
[8] A. Grothendieck and J.-L. Verdier. Théorie des topos et cohomologie étale des schémas, volume 269 of Lecture Notes in Mathematics. Springer Verlag, 1972.
[9] J.S Milne.Comparison of the Brauer group with the Tate-Safarevic group.
J. Fac. Sci. Univ. Tokyo (Shintani Memorial Volume) IA 28 (1982), 735- 743. .
[10] Jean Pierre Serre.Cohomologie Galoisienne. Lecture Notes in Mathemat- ics. Spinger, 1997.
[11] S.S. Shatz. Profinite groups, Arithmetic and Geometry. Annals of Math- ematics Studies, Princeton University Press, 1972.
[12] J. Tate. On the conjecture of Birch and Swinerton-Dyer and a geometric analog. Séminaire N. Bourbaki, exp. no306, p. 415-440., 1964-1966.