arXiv:2101.00044v3 [math.AG] 9 Aug 2021 hr h iadtvt en htteeaentrlisomorph natural are there that means bi-additivity the where n.Dlge[ Deligne one. oicto ftepairing the of gorification sthat is them. epofo eorem of proof e tagtowr,btw utcnedwt h atta oeof some that fact the with contend must we but straightforward, 2. eorem 1. eorem vntog hr r utpeapoce to approaches multiple are there though Even n trpeet [ represents it and following: Ψ - / Let 1 phrases. and words Key 2010 Date Let [ in proved essentially is is Let ntestatement, the In eorem ( sacneuneo h iadtvt ftecppout efu e cup-product. the of bi-additivity the of consequence a is ( uut11, August : K ahmtc ujc Classification. Subject Mathematics !," easot ait vrafield a over variety smooth a be c nerto ln h brmpfrgre htctgrfistewel the categorifies that gerbes for map fiber the along integration Abstract. neet n,aogtewy epoiea xml fterapplicat their of example an provide we way, the along and, interest, 9 9 ( u osrcinpoie ulacuto h idiiiyproper biadditivity the of account full a provides construction Our . 1 eteuulZrsisefaahdt h presheaf the to aached sheaf Zariski usual the be -, 1 eso htDlgesln bundle line Deligne’s that show We efntra ecito ftelwdge asi h ea spect Leray the in maps degree on low the correspondences of of description categorification functorial e the to is application main Our ethe be a ojcue yM ani [ Patnaik M. by conjectured was efunctor e naysot variety smooth any On K SGA73 9 2021. ) sioopi oteCo group Chow the to isomorphic is SGA73 K Let IE NERTO FGRE N EIN IEBUNDLES LINE DELIGNE AND GERBES OF INTEGRATION FIBER 2 , Gerb Ψ c grecrepnigt h u-rdc fln bundles line of cup-product the to corresponding -gerbe leri yls ebs ihrcategories. higher gerbes, cycles, Algebraic Del87 1 - : sotie ycmiigtefloigeorems following the combining by obtained is / - Ψ XVIII , ( - - → : a osrce iadtv uco fPcr categories Picard of functor bi-additive a constructed has ] ( / Tors Tors ( K ( atrzsa opsto fab-diiefunctor bi-additive a of composition a as factorizes h eafml fsot rjciecre,adlet and curves, projective smooth of family a be 2 ! ) AR16 TOEADOAD N IAJNRAMACHANDRAN NIRANJAN AND ALDROVANDI ETTORE - 13 h nerto fagreaogtefiesof fibers the along gerbe a of integration the §1.3] - + eoe h Picard the denotes ( ( CH ! 42,1F2 52,55N15. 55P20, 14F42, 14C25, . Tors ′ < " , ,btfrtebadtvt,wihw drs eo.Biaddit below. address we which biadditivity, the for but ], < Pat08 1 × ) over × ) ( - i . n let and , × ) →h −→ h Tors ( Tors Remark , ," !, ≃ hr xssantrlb-diiefunctor bi-additive natural a exists there , < c CH i ∗ - Ψ × ) - ," !, a eotie rmthe from obtained be can - CH : ( 1. ( 1 / c ( ( 21.3.2]. Tors - < Introduction < h + i : se§ (see ) ) 2 - -aeoyo ebson gerbes of 2-category → ) ( −→ −→ . - → ∪ ∪ ! 9 ( → ) ( 1 ′ 2.2 - Tors " , ( CH Gerb < ) easot rjciemrhs frltv dimension relative of morphism projective smooth a be ) ) i fcodimension- of CH nw n rsn rmteLryseta euneof sequence spectral Leray the from arising one known l h ro fb-diiiyi o-rva nec n of one each in non-trivial is bi-additivity of proof the isms 2 −→ ( , ( * ∪ o oteBae group. Brauer the to ion ( - - h niisivle r ihrctgre rstacks. or categories higher are involved entities the isof ties 1 a eunefor sequence ral ( ) ( h 7→ < K Gerb ( ," !, −→ h efpouto curve. a of self-product the ) ) c nctor 2 , ! , K ∗ )
h ," !, and 2 − − → 9 i . ∫ CH -gerbe ( Ψ c →h −→ * ( - 2 K i ≃ " ∫ ) Tors . 1 , / c ( ( 2 3 eapi fln ude on bundles line of pair a be on ( aeoie h uhowr map pushforward the categorifies ) ( and , ! - 9 c ," !, . ,! ", ) !," ylson cycles ( and htw eeo so independent of is develop we that . ∪ - ihband with ( ai euto Bloch-illen of result basic A . n nadtv functor additive an and ) osrce n[ in constructed i 4 c < " : = . us, . ) h . on ," !, - - u anrsl sthe is result main Our . K i vee as (viewed Ψ , 2 - h iadtvt of bi-additivity the ; / vt radtvt is additivity or ivity AR16 ( satal cate- a actually is - i an via ] . < ∫ -torsors). c : 2 ETTORE ALDROVANDI AND NIRANJAN RAMACHANDRAN
eorem 3. For c : - → ( as above of relative dimension one, there exists a natural additive functor
: Gerb- (K2) −→ Tors( (<) . ¹c e proof of eorem 3 consists in writing the maps in the low degree part of the Leray spectral sequence for c : - → ( directly in terms of the (higher) stacks they classify. While this can be traced back in some implicit form to [Gir71, §V.3.1-2], we reprise it here as we need in particular an explicit description of the functors involved. In particular, the integration map is given by taking the sheaf of connected components of the pushforward gerbe from - to (. We describe the integration map in greater generality, by working with a general site morphism. Finally, we have eorem 4. One has a natural isomorphism
!," h!,"i . ¹c ∗ 9 Let . be a smooth proper variety . over ; let CH (. ) = ⊕9 CH (. ) for the total Chow group of . . Recall that one has a homomorphism [Ful98, Example 16.1.2(c)] CHdim . (. × . ) → End(CH∗(. )) (1) of rings; the ring structure on the former is given by the composition of correspondences. e categorification of (1) is of great interest. We use eorem 1 to provide a categorification (eorem 18)of(1) when dim. = 1. It should be remarked that the problem of categorification of (1) seems to be formidable when dim. > 1: for a surface . , one needs to endow the Picard 2-category Gerb. ×. (K2) with a ring structure. Let CH1 (. ) denote the Picard category of line bundles on . . If is a smooth projective curve over an algebraically closed field, then CH1 ( × ) can be naturally enhanced to a ring category and the natural functor CH1 ( × ) → End(CH1 ()) is a functor of ring categories (eorem 18). While there are several generalizations [Gor09, Blo89, BS88, Elk89, Fra91, Fra90, Blo84, Eri13] of Deligne’s con- struction, they are all restricted to line bundles or codimension one. However, eorem 1 suggests new generaliza- tions [AR21] of Deligne’s construction: if 5 : . → ( is smooth proper of relative dimension two, there exists a natural bi-additive functor Ψ2 K K K . /( : Gerb. ( 2)× Gerb. ( 2) −→ Gerb( ( 2), which is a categorification of the pairing
5∗ CH2 (. )× CH2 (. ) −→ CH4 (. ) −→ CH2 (() . Organization. In section 3 we analyze in some detail the low-degree terms exact sequence of the Leray spectral sequence for c : - → (. While this is all well known from [Gir71, §V.3.1-2], we expand on it as several details were famously le as an exercise ([Gir71, Exercice 3.1.9.2]). Since we describe the maps in the sequence fairly explicitly, as an example we use them to illustrate an application to the Brauer group, which is of independent interest. In section 4 we prove eorems 2 and 3, and, finally, we prove eorem 4, the comparison with Deligne’s construction, in section 5. We end with a proof of the main result (eorem 18) about the categorification of correspondences in §7. e requisite results from the theory of Picard categories are in §6. Notations. For any sheaf of abelian groups on a site we denote by Tors() the Picard category of -torsors and by Tors() the corresponding stack. Similarly, one categorical level up, for Gerb() and Gerb(), which denote the 2-Picard category of -gerbes and the corresponding 2-stack. For any stack F, we denote by c0 (F) its sheaf of connected components and by F the category of its global sections, that is Hom(pt, F), where pt is the terminal sheaf. Acknowledgements. anks to Gerard Freixas I Montplet for alerting us to [BS88], and Sasha Beilinson for his help with understanding [BS88] and for helping us with the proof of Proposition 15. is work was partly supported by a Travel Award Grant from the Florida State University College of Arts and Sciences, which we gratefully acknowledge.
2. Norms and Deligne’s construction
We recall some properties of Deligne’s functor Ψ- /( and the line bundle h!,"i on (. FIBER INTEGRATION OF GERBES AND DELIGNE LINE BUNDLES 3
2.1. Norms and finite maps. Let 6 : + → , be a finite and flat morphism of varieties. Given a line bundle ! on + , its norm (relative to 6) is a line bundle #+ /, (!) on , . One has an additive functor of Picard categories [Del87, §7.1]
#+ /, : Tors+ < −→ Tors, < .
2.2. Characterization of Deligne’s functor Ψ- /( . Let ⊂ - be an effective relative Cartier divisor [Sta20, Tag 056P] of c : - −→ (. Namely, is an effective Cartier divisor on - and the induced morphism c : → ( is finite and flat. For any line bundle " on -, the norm #/( (") is a line bundle on (. Deligne’s construction Ψ- /( is characterized by [SGA73, XVIII 1.3.16]: (i) functoriality, and (ii) for any section of ! with zero set an effective Cartier divisor on -, a canonical isomorphism #/( " h!,"i . (2) Ψ Another approach to - /( from [SGA73, XVIII 1.3.17.2] is the following: if and are effective relative Cartier divisors on -, then L hO(), O()i det Rc∗(O() ⊗ O()). (3)
3. Fiber Integration of gerbes and the Leray spectral seence Let be an abelian sheaf on -. e spectral sequence 8,9 8 9 8+9 2 = H ((, R c∗) ⇒ H (-, ) (4) has as low-term exact sequence [Mil80, Appendix II, page 309] 1,0 1 0,1 2,0 2 1,1 0 −→ 2 −→ −→ 2 −→ 2 −→ 1 −→ 2 , (5) where 1 1 2 2 0 2 = H (-, ) , 1 = Ker(H (-, ) −→ H ((, R c∗)) , 0 2 0,2 and, of course, H ((, R c∗) = 2 . e maps above arise from functors between categories of torsors and gerbes, as shown in [Gir71, pp. 324–327]. For our own purposes, and also to rephrase the arguments in loc. cit. in a more transparent way, we turn to an explicit description of these functors. Our arguments below (in the Zariski topology) are easily seen to be also valid in the etale´ or analytic topology. In fact, at the beginning they are valid for any morphism between sites whose underlying functor is assumed for simplicity to preserve finite limits, and we shall begin our discussion in such generality. 3.1. Site morphisms, push-forwards and pullbacks of stacks. Let c : D → C be a morphism of (small) sites. We let D = c−1 : C → D denote the underlying functor. us, c is a morphism of sites if composition along D preserves sheaves, and this operation has a le adjoint that is exact [SGA73]; this is implied by the property that D preserves coverings and if both C and D have finite limits D preserves them [Jar15],[Sta20, Tag 00X0]. Let G be a category over D. Its push-forward c∗ G along c is defined by
c∗ G = C ×D G as a category over C via the first projection. It is fibered (resp. a stack) if so is G. On the other hand, let ? : F → C ∗ ′ ′ ′ be a stack. e inverse image c F is a pair (F ,q), where F is a stack over C, and q : F → c∗F a stack morphism such that, for any stack G over D, the following composite functor ′ ′ HomD (F , G) −→ HomC (c∗F ,c∗ G) −→ HomC (F,c∗ G) , is an equivalence of categories [Gir71, Def.´ 3.2.1]. Here Hom denotes the category of stack morphisms. us, the inverse image is truly only defined up to equivalence. While specific formulas to compute a modelof c∗F do exist [Sta20, Tag 04WJ], the universal property is sufficient to characterize its connected components. Recall that c0 (F) is the sheaf corresponding to the presheaf of connected components: to any object * ∈ C it assigns the set of connected components c0 (F* ) of the fiber category F* [Bre94, Chap. 7]. (In ref. [Gir71, n. 2.1.3.3] this is the “sheaf of maximal sub-gerbes of F.”) We have [Gir71, Prop. 2.1.5.5 (iii)] an isomorphism of sheaves over D: ∗ ≃ ∗ c0 (c F) −→ c (c0 (F)) . is follows from the fact that if G,~ are any two objects of F over * ∈ C, then there is a sheaf isomorphism ∗ ≃ ′ ′ c HomF (G,~) −→ HomF′ (G ,~ ) , 4 ETTORE ALDROVANDI AND NIRANJAN RAMACHANDRAN
′ ′ F′ where the objects G ,~ of c −1 (* ) are constructed via the above universal property (ibid.). As a consequence, since a gerbe is locally connected, we have that the inverse image of a gerbeis a gerbe[Gir71, Cor. 2.1.5.6]. In fact, assuming, as we shall do in later sections, that F has band , for an abelian sheaf over C, then c∗F has band c∗. On the other hand, even if G → D is a gerbe, its push-forward will not necessarily be so. In other words, c0 (c∗ ( G)) may turn out to be a nontrivial sheaf over C. More precisely, we have the following statement. Lemma 5 ([Gir71, Exercice 3.1.9.2]). Let c : D → C be a site morphism as above. Let G be an -gerbe on D, where 1 2 is an abelian sheaf. en c0 (c∗ ( G)) is a pseudo R c∗-torsor. It is a torsor if and only if the class [ G] ∈ H (D, ) lies 2 0 2 2 in the kernel of the map H (D, ) → H (C, R c∗), and hence in the term denoted 1 above. 2 1 An -gerbe G is horizontal if its class [ G] lies in 1. If G is horizontal, then c0 (c∗ ( G)) is anR c∗-torsor. By H8 (C, −) (same for D) we denote the cohomology of the terminal sheaf pt. In the concrete case of the Zariski sites, where pt is represented by the site’s terminal object, this reduces to the groups considered at the beginning of this section. ′ Definition 6. Let Gerb- (K2) be the full sub(2-)category of Gerb- (K2) consisting of horizontal gerbes. e functor ′ 1 Θc : Gerb- (K2) → Tors( (R c∗K2) sends a gerbe G to c0 (c∗ ( G)). 2 1,1 We write \ : 1 → 2 for the induced map. (In the previous statement, as well as in several that follows, the relevant (2-)categories can be upgraded to the correspoonding (2-)stacks.) 3.2. Proof of Lemma 5. is section is devoted to a complete proof of Lemma 5. Several points of the proof will be explicitly needed in sections 4 and 5 below.2 It is convenient to express the sites’ topologies in terms of local epimorphisms [SGA73, KS06], and take hypercov- ers of those, in particular Cechˇ nerves. For simplicial objects we use the “opposite index convention” [Dus02] (and ∗ reverse the order of the maps for cosimplicial ones) when pulling back by simplicial maps: 38 (−) = (−)[=]\8, where [=] is the ordinal [=] = {0 < 1 < ··· < =}. 3.2.1. Objects with operators. Let F be a stack over a site C, and let be a sheaf of groups over C. e stack Op( , F) has objects the pairs (G,[), where G ∈ F* , and [ : |* → Aut* (G). Morphisms from (G,[) to (~, \) are arrows U : G → ~ in F* compatible with the structure: U ◦ [(6) = \ (6) ◦ U, for all sections 6 ∈ |* .
Lemma 7 ([Gir71, III№ 2.3]). ere is a stack morphism C : Tors( )×C Op( , F) → F. % is “twisting” morphism assigns to each pair (%, (G,[)) over * ∈ C an object of F* , variously denoted as G or % ∧ G. Proof. If % = , the trivial -torsor, we set G = G. To a morphism (6, U) : ( , (G,[))→ ( , (G ′,[′)) (here 6 ∈ is identified with an automorphism of the trivial torsor) we assign the morphism G → G ′ given by U ◦[(6) = [′(6)◦U. In general, we regard % ∈ Tors( )|* and G ∈ F* as defined by descent data relative to an acyclic fibration n : +• → * ∗ ∗ covering * . e pullbacks n G and n % |+0 to +0 are glued over +1 by isomorphisms
U : G1 → G0 , 6 : → , where 6 ∈ (+1) is an isomorphism between trivial |+1 -torsors, satisfying the cocycle conditions U02 = U01 ◦ U12 and 601612 = 602 over +2. at (G,[) is an object of Op( , F) is expressed by the condition U ◦ [1(−) = [0 (−) ◦ U over +1. Pullbacks to+2 along the facemaps 30,31,32 yield morphisms (68 9, U8 9 ) : ( , (G 9 ,[ 9 ))→( , (G8,[8)), 0 ≤ 8 < 9 ≤ 2, F in Op( , )+2 such that U8 9 ◦ [ 9 (68 9 ) = [8 (68 9 ) ◦ U8 9 , in addition to the other cocycle conditions. en we have
U02 ◦ [2(602) = U01 ◦ U12 ◦ [2 (601) ◦ [2(612) = (U01 ◦ [1(601))◦(U12 ◦ [2(612)) , showing the gluing data U ◦ [1 (6) = [0 (6) ◦ U : G1 → G0 satisfy the cocycle identity and therefore define an object % G of F* . e most important properties of the twisted objects are listed in the following lemma implicit in [Gir71], where for any two objects G,~ ∈ F* we denote by Isom* (G,~) the sheaf of isomorphisms from G to ~. Note that Isom* (G,~) is a right Aut* (G)-torsor. (In fact it is an (Aut* (~), Aut* (G))-bitorsor, but we shall not need this fact.)
2As all the details are famously not available in the original reference as well as in the literature, we felt compelled to include them here. FIBER INTEGRATION OF GERBES AND DELIGNE LINE BUNDLES 5
Lemma 8 ([Ems17, MR208]). Let F be a stack and a sheaf of groups on C. Let Op( , F) the stack of objects with -action. e twisting morphism C has the following properties: (1) If % ∈ Tors( )* , and G ∈ F* , then Isom* (G, % ∧ G) % ∧ Aut* (G) ; Aut (~) ≃ (2) If % = Isom* (~, G), then there is a canonical isomorphism % ∧ * ~ −→ G, where the twisting arises from the stack Op(Aut* (~), F|* ) over C/* . Proof. Let (~, \) be another object of Op( , F) over * . Using the same notation as in Lemma 7 for descent data ∗ ∗ relative to +• → * , a morphism & ∧ ~ → % ∧ G of twisted objects over * corresponds to a morphism _ : n ~ → n G
over +0 such that the diagram over +1 _← 1 ~1 → G1 → → ←
V◦\1 (ℎ) ← U◦[1 (6) (6) ~0 ← → G0 _0 commutes. Here V and ℎ represent the descent data and cocycle for ~ and the -torsor &, respectively. In particular, if ~ = G and & is the trivial torsor, we get the simpler relation
U ◦ [1 (6) ◦ _1 = _0 ◦ U . Rewriting it in the more suggestive way −1 [1(6) ◦ _1 = U ◦ _0 ◦ U shows that the _ defines a section of % ∧ Aut* (G), proving the first point. ∗ ∗ If % = Isom* (G,~), _ : n ~ → n G provides a section of % over +0. As _ does not necessarily descend to * , the two
pullbacks _0 and _1 to +1 are related by a diagram of the form _← 1 ~1 → G1 → ← ℎ¯ ←
~1 U → →
V ← ~0 ← → G0 _0
for an appropriate ℎ¯ ∈ Aut(~1) = Aut* (~)(+1). Comparing with (6) (taking \ = id) shows these data descend to an ≃ isomorphism % ∧Aut* (~) ~ −→ G, as wanted. F is an abelian gerbe with band if and only if there is a canonical morphism F → Op(; F), because, in such case, the correspondence ≃ G ∈ F* [G : |* −→ Aut* (G) is functorial: the diagram
|*
[G ← ← [~
→ → Aut (G) ← → Aut (~) * U∗ * commutes whenever U : G → ~ [Bre94, Def. 2.9]. (Note that the above diagram embodies a morphism of Op(; F).) erefore there is a canonical twisting action (%, G) % G resulting from the composite morphism
Tors()×C F −→ Tors()×C Op(; F) −→ F . 3.2.2. e pushforward. Let us return to the situation of the site morphism c : D → C. Recall that D : C → D is the underlying functor of c. Let be an abelian sheaf and G be an -gerbe over D. It is convenient to identify the band with the automorphism sheaves. In this way, Lemma 8, statement (1), simply becomes Isom* (G, % ∧ G) %. 1 e action c0 (c∗ G)× R c∗ → c0 (c∗ G) is induced by the twisting action of Tors() on G on D: if G ∈ GD (* ) 1 represents a section of c0 (c∗ G), and % ∈ Tors()D (* ) represents a class of R c∗ (* ), we let the result of the action be the connected component of the object % ∧ G ∈ GD (* ) c∗ ( G)* . is action is free, because if % ∧ G G, then by Lemma 8 (1) Hom* (G, G ∧ G) % has a global section, hence % |D (* ) . 6 ETTORE ALDROVANDI AND NIRANJAN RAMACHANDRAN
e action is also transitive. Indeed, if the objects G,~ ∈ GD (* ) c∗ ( G)* represent two sections of c0 (c∗ G), by Lemma 8 (2) we have ~ % ∧ G, where % = HomD (* ) (G,~). erefore the section of c0 (c∗ G) over * defined by 1 ~ is obtained from that defined by G via the action of the section of R c∗ (* ) determined by %, as wanted. us, c0 (c∗ G) is a pseudo-torsor. Let * be an objectof C, and denote by* ′ = D(* ) the corresponding object of D. As a gerbe, G is locally nonempty, ′ ′ ′ hence there will be a local epimorphism+ → * covering* with an object G ∈ G+ ′ . eobject G ought to be seen as 2 ′ 2 a trivialization of the restriction G|* ′ , whose characteristic class is an element of H (* , |* ′ ) R c∗ (* ). (In effect, ′ ′ this class can be calculated by computing the 2-cocycle determined by G via a hypercovering +• → * in the usual ′ way [Bre94].) If it is zero, then G* ′ has a global object (that is, G descends to an object over * ), which then provides 2 a section of c0 (c∗ G) over * ∈ C. Clearly, if R c∗ vanishes, this argument shows that c0 (c∗ G) is locally nonempty. On the other hand, if c0 (c∗ G) is locally nonempty, for every object * we can find a local epimorphism + → * such G 2 2 2 that D (+ ) has an object and therefore H (D(+ ), |D (+ ) ) = 0. Now writing R c∗(* ) = lim[+•→* ] H (D(−), |D (−) ) 2 we get R c∗ = 0. is finishes the proof of Lemma 5. 3.3. Maximal subgerbes and pullbacks. e following extra facts (“tautologies” in [Gir71, V № 3.1.8]) are going to be helpful. Recall that we have a site morphism c : D → C. Following Giraud, let us say that a gerbe G over D comes from a gerbe on C, if there is a gerbe F on C and a morphism of gerbes < : c∗F → G over D.
Lemma 9. e gerbe G on D comes from a gerbe on C if and only if the sheaf c0 (c∗ G) admits a section. Proof. If G comes from a gerbe on C, let F be such a gerbe and < : c∗F → G the corresponding morphism. By ∗ adjunction (cf. the universal property that defines the operation c , sect. 3.1) we obtain a morphism = : F → c∗ G. Since F is a gerbe, for the sheaf of connected components we get c0 (=) : pt → c0 (c∗ G), hence a section of c0 (c∗ G). G G F G Conversely, if c0 (c∗ ) has a section, say b : pt → c0 (c∗ ), define = pt ×c0 (c∗ G) c∗ ( ), which is a gerbe on C, and = : F → c0 (c∗ G) as the second projection. e laer is by construction fully faithful, hence, again by adjunction, we have the morphism < : c∗ (F) → G, and so G comes from a gerbe on C. Remark 10. In the previous proof we have used the well known fact (but, again, ultimately due to Giraud [Gir71,III Prop. 2.1.5.3]) that any stack S the projection S → c0 (S) makes it a gerbe on the sheaf of its connectedcomponents. ∗ For a section b ∈ c0 (c∗ G), the pullback b (S) is the corresponding maximal sub-gerbe. 2 Remark 11. Using Lemma 5, we see that Lemma 9 is equivalent to the exactness at H (C,c∗) of the low term sequence 0 1 2 2 ′ · · · −→ H (C, R c∗) −→ H (C,c∗) −→ H (D, ) −→ ··· 2 ′ 2 arising from the Leray spectral sequence we recalled above, where we set H (D, ) = 1. 3.4. Interpretation of the maps. 1,0 1 1 1 ∗ 1 3.4.1. e map 2 → . is is the obvious pull-back map H ((,c∗) → H (-,c c∗) → H (-, ) using the ∗ natural adjunction c c∗ → of sheaves on -. is is just the composite functor ∗ c ∗ Tors( (c∗ ()) −→ Tors- (c c∗ ()) −→ Tors- () for the corresponding gerbes. 1 0,1 1 1 −1 3.4.2. e map → 2 . Using that R c∗ is the sheaf associated to * H (c (* ), ), the map is obtained by considering the class of an object % of Tors- (-) under % 7→ c∗%. 0,1 2,0 3.4.3. e map 2 → 2 . is is the transgression map relative to the standard sequence arising from an injective resolution 0 → → • on -. en it is standard that 0 1 • 1 0 −→ c∗ −→ c∗ −→ / (c∗ ) −→ R c∗ −→ 0 . Viewing it as the splicing of two short exact sequences 0 1 • 1 0 −→ c∗ −→ c∗ −→ −→ 0 , 0 −→ −→ / (c∗ ) −→ R c∗ −→ 0 , the transgression map is the composite 0 2 1 2 H ((, R c∗) −→ H ((,) −→ H ((,c∗) . e laer is obtained by taking the global objects of the composite 2-functor:
−→ Tors( () −→ Gerb( (c∗) . FIBER INTEGRATION OF GERBES AND DELIGNE LINE BUNDLES 7
1 e map on the right is the well known classifying map of the extension of R c∗ by above [Gir71,V№ 3.2] (see also [AR16]). 2,0 2 3.4.4. e map 2 → 1. Analogously to 3.4.1, we have the composite 2-functor ∗ Gerb( (c∗) −→ Gerb- (c c∗) −→ Gerb- () , ∗ where the arrow on the right is “change of band” functor along c c∗ → . Taking isomorphism classes in the 2 2 ∗ 2 global fibers gives the composite H ((,c∗) → H (-,c c∗) → H (-, ). Now, thanks to Lemma 9 the image is 2 actually in 1. 2 1,1 { 3.4.5. e map \ : 1 → 2 . Let G be a stack on - and consider the correspondence G c0 (c∗ G). is correspon- dence is easily seen to be a functorial one between the homotopy category of stacks—as it identifies two naturally isomorphic stack morphisms—on - to that of sheaves on (. erefore, by Lemma 5, and the subsequent sections, ′ 1 ′ it reduces to a functor Ho(Gerb- () ) → Tors( (R c∗ ()), where Gerb- () denotes the subcategory of those - gerbes whose fiber categories over opens of the form c−1 (* ), for every sufficiently small open neighborhood * ⊂ ( around every point of (, are not empty. By taking classes, we get the map. 2 1,1 3.5. Application: Brauer groups. e map \ : 1 → 2 above plays a very important role in many arithmetical applications [Kai16, Lic69, Sko07]. To recall this, let be a perfect field and consider the etale´ sheaf < on) = Spec . Fix an algebraic closure ¯ of and let )¯ = Spec ¯; write Γ for the Galois group of ¯ over . Let 6 : . → ) be a smooth proper map and .¯ = . ×) )¯. 2 2 ¯ e group Het´ (., <) is the Brauer group Br(. ). e group 1 is the relative Brauer group Br(. /. ), namely, the ¯ 2 1,1 kernel of the map Br(. ) → Br(. ). e map 1 → 2 then becomes the map ¯ 1 1 \ : Br(. /. ) −→ Het´ (), R 6∗<) , arising in several contexts. For instance, for any elliptic curve over = ℚ, the map \ gives the well known isomorphism 1 ¯ 1 Br() ≃ H (ℚ,(ℚ)) = Het´ (Spec ℚ,) . e following explicit description of \ seems to be missing in the literature: Given an element U of Br(.¯/. ), pick a < ¯ ¯ ¯ gerbe on . representing U. By definition, the base change on . is trivial. Fix an equivalence 5 : ≃ Tors.¯ (<). ∗ en, given any f ∈ Γ, the gerbe f ¯ is equivalent to ¯ as comes from . . Write 5f for the resulting equivalence of Tors-¯ (<): ∗ 5 ¯ ∗ ¯ f 5 ∗ 5f : Tors.¯ (<) ←− ≃ f −−−→ f Tors.¯ (<) = Tors.¯ (<) . Any self-equivalence [Mil03, §5.1] Tors.¯ (<) ≃ Tors.¯ (<) is a translation by a fixed <-torsor !, namely the self-equivalence is of the form (−) 7→ (−) + !. erefore, if !f is ¯ 1 1 the <-torsor on . corresponding to 5f , then the map f 7→ !f represents the element \ (U) of Het´ (), R 6∗ <). 4. Gerbes and categorical intersection of divisors In this section, we prove eorems 2 and 3. Let . be a smooth variety over . Let K8 denote the Zariski sheaf associated with the presheaf * 7→ 8 (* ). 4.1. Heisenberg groups. For any pair abelian sheaves and on . , we have constructed [AR16] a Heisenberg sheaf , (of nilpotent groups) which fits into an exact sequence
0 −→ ⊗ −→ , −→ × −→ 0 , (7) providing a categorification of the cup-product H1(., )× H1 (., ) −→ H2 (., ⊗ ) (8) in the following manner. Given an -torsor % and a -torsor &, the × -torsor % × & can be lied locally to a , -torsor in several ways. ese local lis assemble to a ⊗ -gerbe %,& . Much like the bulk of section 3, we can formulate the result we need in much greater generality. As in [AR16, §3], we assume and are abelian objects of a topos T. (In the applications, we assume T to be the topos of sheaves on the Zariski, or other relevant topology, of the scheme.) Recall from loc. cit. that the Heisenberg group , is defined by the group law: (0,1,C)(0′,1′,C ′) = (00′,11′,C + C ′ + 0 ⊗ 1′) , 8 ETTORE ALDROVANDI AND NIRANJAN RAMACHANDRAN where 0, 0′ are sections of , 1,1′ of , and C, C ′ of ⊗ . e extension (7) is set-theoretically split, i.e. there is a section of the underlying map of sheaves of sets. e map 5 : ( × )×( × ) −→ ⊗ , 5 (0,1,0′,1′) = 0 ⊗ 1′, (9) 2 2 is a cocycle representing the class of the extension in H (B×, ⊗ ) H ( ( × , 1), ⊗ ), where on the le we have the cohomology of classifying topos [Gir71], and on the right that of the corresponding Eilenberg-Mac Lane simplicial object of T. In fact these cohomologies are in turn isomorphic to [ ( × , 1), ( ⊗ , 2)], the hom-set in the homotopy category [Bre78, Ill71], and the cocycle 5 coincides with the only non-trivial component of the characteristic map [AR16, Prop. 3.4]. Proposition 12. e functor
2, : Tors()× Tors() −→ Gerb( ⊗ ) , % × & 7−→ %,& T ∼ is bi-additive. On c0, it induces the cup-product map (8), upon choosing = .Zar. Remark 13 (On bi-additivity). Note that for any (abelian) band !, Gerb(!) is really a 2-stack. Hence the notion of 2-additivity should be updated with appropriate 2-coherence data from higher algebra. is is both outside the scope of this note and inconsequential in the case at hand. Alternatively, we can mod out the 2-morphisms and consider Gerb(!) as a Picard 1-stack of T. us, bi-additivity consists of the data of functorial equivalences ≃ 2, (%1 + %2,&) −→ 2, (%1, &)+ 2, (%2,&) ≃ 2, (%, &1 + &2) −→ 2, (%, &1)+ 2, (%, &2)
subject to the condition that decomposing 2, (%1 + %2, &1 + &2) according to the two possible ways determined by the above morphisms gives rise to a commutative (or commutative up to coherent 2-isomorphism) diagram. is would be exactly the kind of diagram familiar from the theory of biextensions [Gro72, Bre83] (see also [Del91]). Proof of Proposition 12. Bi-additivity is essentially already implied by the fact that the cocycle representing the class of the extension is the tensor product, which is bilinear. It is best to look at this in the universal case, namely over (×, 1)—the rest follows by pullback—where the bilinearity of the tensor product has the following interpretation. By functoriality, from the group operation ( is an abelian object) + : × → we get the map + : ( × × , 1) → ( × , 1) corresponding to the Baer sum of torsors. Its composition with the characteristic map 2 : ( × , 1) → ( ⊗ , 2) equals in the homotopy category the sum 21 + 22, where 28, 8 = 1, 2, is the composition
?8 2 ( × × , 1) −→ ( × , 1) −→ ( ⊗ , 2) ; the first map is induced by the projection onto the first (second) factor, as it follows from (9) and the form of 2 computed in [AR16, §3.4]. ∗ ∗ In turn, the map 2 ◦ (+) classifies the extension (+) , , whereas 21 + 22 classifies the extension ?1 , + ∗ ?2 , —the sum is the Baer sum in this case—so that we obtain the isomorphism ∗ ∗ ∗ (+) , ?1 , + ?2 , (10) of central extensions of × × by ⊗ . Similarly for the “variable” . Furthermore, the commutativity of the diagram alluded to in Remark 13 is immediately implied by further pulling back the isomorphism (10)by + : × → , its counterpart for via +, and again using (9).
4.2. Proof of eorem 2. Consider the map ` : < × < → K2 obtained using the identification < ≃ K1 and the multiplication K1 × K1 → K2. e functor ∪ defined as the composite
2< ,< `∗ Tors. (<)× Tors. (<) −−−−−−→ Gerb. (< ⊗ <) −→ Gerb. (K2) is the required bi-additive functor. e functor ∪ is so-named as it categorifies the cup-product (which can be identified with the intersection product 1 1 2 2 1 1 H (., <)× H (., <) −→ H (., K2) ≃ CH (. ) ←− CH (. )× CH (. ) .
Remark 14. e bi-additivity property of the map 2, of Proposition 12 has the following conjectural formal inter- pretation. e maps + and +, plus the commutative diagram in Remark 13 and the proof of Proposition 12 comprise a structure that can be described as the categorification of a biextension, namely a Tors( ⊗ )-torsor (hence an ⊗ -gerbe) H −→ Tors()× Tors() FIBER INTEGRATION OF GERBES AND DELIGNE LINE BUNDLES 9 equipped with partial addition laws + (resp. + ) giving it the structure of an extension of Tors() (resp. Tors()) by Tors( ⊗ ).
4.3. Proof of eorem 3. Our proof will use the results of §3 on (5) for c : - → ( with = K2 on -. Proposition K 15 shows that all 2-gerbes are horizontal (Definition 6). e functor c is then defined as the composition of Θ ∫ 1 Norm Gerb- (K2) −→ Tors( (R c∗K2) −−−−→ Tors( (K1) . ′ Our first step is to show that Gerb- (K2) is all of Gerb- (K2), in other words, every K2-gerbe on - is horizontal. 2 is is proved by showing R c∗K2 = 0 which provides the isomorphism 2 ≃ 2 K 1 −→ H (-, 2) . We start with the following result, implicit in [BS88, A5.1 (iv)], essentially due to Beilinson-Schechtman. 2 Proposition 15 (Beilinson-Schechtman). e sheaf R c∗K2 is zero. is gives a map 2 1 1 \ :H (-, K2) −→ H ((, R c∗K2) (11) using 2 K ≃ 2 1 1 K H (-, 2) ←− 1 −→ H ((, R c∗ 2) . Proof. Let # be the dimension of (, so that - has dimension # + 1. 2 For any B ∈ (, we have to show that the stalk of R c∗K2 at B is zero. By definition, this is the direct limit lim R2c K (* ) = lim H2 (c−1 (* ), K ) = lim CH2 (c−1 (* )) , −−→ ∗ 2 −−→ 2 −−→ B ∈* B ∈* B ∈* where the last equality comes from the Bloch-illen isomorphism (valid for any smooth variety + )
2 ≃ 2 H (+, K2) −→ CH (+ ) . So, we have to show that for any B ∈ (, any open set * containing B, and any codimension two cycle / in c−1 (* ) ⊂ -, there exists an open subset * ′ ⊂ * such that the class of / goes to zero under the map CH2 (c−1 (* )) −→ CH2 (c−1 (* ′)) . is is clear when B is the generic point Spec (() of (: in this case, we take * ′ to be the complement of c (|/ |) in * . Here we have wrien |/ | for the support of /. e next the case is when B is a point of codimension 8 > 0, corresponding to a codimension 8 subvariety + of (. Let us write . ⊂ - for c−1 (+ ); then . is a subset of - with codimension 8. For any open * ⊂ (, the condition B ∈ * means * ∩ + is non-empty. Let * be such an open set. ere are two cases to consider: Case 1: If |/ | is disjoint from . , then we can proceed as before as c (/) is disjoint from + , so we take * ′ to be the complement of c (|/ |) in * . Since * ′ ∩ + = * ∩ + , we see that * ′ ∩ + is non-empty. Since / is in the kernel of the localization sequence for Chow groups CH2 (c−1 (* )) → CH2 (c−1 (* )−|/ |) → 0 , it is also in the kernel of the composite map CH2 (c−1 (* )) → CH2 (c−1 (* )−|/ |) → CH2 (c−1 (* ′)) . is finishes the proof in this case. Case 2: If |/ | is not disjoint from . , we can find a codimension two cycle / ′ in c−1 (* ) with [/] = [/ ′] ∈ CH2 (c−1 (* )) which intersects . transversally. e codimension of the cycle / ′.. is 8 + 2, because its di- mension (= maximum of the dimensions of the irreducible components) is # + 1 −8 − 2 = # − 1 −8. Hence the dimension of the image c (/ ′.. ) is at most # − 1 − 8, and so its support |c (/ ′.. )| is a proper closed subset of + = c (. ). If * ′′ is the complement of |c (/ ′.. )| in * , then the intersection of * ′′ and + is empty. By definition, the cycle / ′ ∩ c−1 (* ′′) is disjoint from . . is means that the image of / ′ (= image of /) under the map CH2 (c−1 (* )) −→ CH2 (c−1 (* ′′)) 10 ETTORE ALDROVANDI AND NIRANJAN RAMACHANDRAN
is a cycle disjoint from . . By Case 1, we can shrink * ′′ further to * ′ such that / ′ (and hence / also) is in the kernel of the map CH2 (c−1 (* ′′)) −→ CH2 (c−1 (* ′)) , as required.
is gives the functor Θ appearing in the definition of Θ 1 Norm : Gerb- (K2) −→ Tors( (R c∗K2) −−−−→ Tors( (K1). ¹c 1 K O∗ Our next step is the definition of the map R c∗ 2 −→ S. Remark 16. e same proof shows that if 5 : . → ) is a smooth proper map of dimension = with . and ) smooth, 9 9 then R 5∗K9 = 0 for all 9 > =. is says that the relative Chow sheaves CH (. /) ) vanish for all 9 > =. 1 K O∗ 4.4. e norm map R c∗ 2 −→ S. is well known map [Ros96, 3.4], [Gil05, pp. 262-264] arises from the covariant functoriality for proper maps of Rost’s cycle modules (Chow groups in our case). We provide the details for the convenience of the reader. Our description proceeds via the Gersten sequence (a flasque resolution of the Zariski sheaf K2 on -)
0 −→ K2 −→ [∗K2,[ −→ 8∗ 1 (:(G)) −→ 8∗ 0 (:(~)) → 0; (12) GM∈- (1) ~M∈- (2) here [ : Spec (-) → - is the generic point of - and - (8) denotes the set of points of codimension 8 of -. For any * open in (, the norm map 1 −1 K O∗ H (c (* ), 2) −→ ( (* ) is obtained as follows. Since the first group is the homology at degree one of (12), we proceed by constructing a map O∗ 8∗ 1 (:(G)) → ( (* ) . G ∈cM−1 (* ) (1) For each such G ∈ c−1 (* ) of codimension one, the map G → c (G) is either finite or not, and it is zero in the second case. In the first case, there is a norm map :(G)∗ → :(c (G))∗ ; since G has codimension one in -, its image c (G) is the generic point of ( and hence the above norm map is a map :(G)∗ → (()∗ . 1 −1 ∗ −1 An element of H (c (* ), K2) arises from a finite collection of functions 5G ∈ :(G) (for G ∈ c (* ) of codimension one which is finite onto its image) which is in the kernel of the map
8∗ 1 (:(G)) → 8∗ 0 (:(~)) . G ∈cM−1 (* ) (1) ~ ∈cM−1 (* ) (2) On each component, this is the ord or valuation map. One checks that this means that the (finite) product of the ∗ O∗ norms of 5G is an element of (() with no poles on * and hence defines an element of ( (* ). is gives the required functor 1 Norm Tors( (R c∗K2) −−−−→ Tors( (K1) ,
completing the definition of the functor c of eorem 3. ∫ 5. Comparison with Deligne’s construction
Given line bundles ! and " (viewed as <-torsors) on -, consider the K2-gerbe !," on -. By Proposition 2 K 2 Θ 15, the element [ !," ] of H (-, 2) actually lives in 1 and hence !," is horizontal. By Lemma 5, ( !," ) is a 1 K R c∗ 2-torsor. By definition, c !," is its pushforward along the norm map of §4.4, ∫ 1 ∗ Norm: R c∗K2 −→ OS , which gives a line bundle (!,") on (. In this section, we show that this gives Deligne’s line bundle h!,"i . Since (!,") is bi-additive and its construction is functorial, this reduces to showing the identity in eorem 4: hO(), O()i (O(), O()) , for any relative Cartier divisors and on - with effective. FIBER INTEGRATION OF GERBES AND DELIGNE LINE BUNDLES 11
5.1. Comparison. To show that hO(), O()i is isomorphic to (O(), O()), one just has to show that they are equal in H1 ((, O∗). is amounts to showing that the diagram below is commutative:
H1 (-, O∗)
[ ← ← ∪
→ → 1 ∗ ← _ 2 H (, O ) → H (-, K2) (13) → →
#/( ← \← 1 ∗ → Norm 1 1 H ((, O ) ← H ((, R c∗K2) e map [ is the restriction to of a line bundle O(). e map ∪ sends O() to its cup-product with O(). e boundary map _ in the localization sequence
0 → K2,- −→ 9∗K2,* −→ 8∗K1, → 0 for -, * = - − , and . e map \ is the map (11) 2 1 1 H (-, K2) −→ H ((, R c∗K2) . e commutativity of (13) is an implicit consequence of the axiomatics of Rost [Ros96], but we provide a direct proof.
5.1.1. e top triangle of (13). We first prove the commutativity of the top triangle of (13). Let {*8 } be a Zariski open cover of - such that and are principal divisors on *8 . Let {58 } be defining equations for and {68 } be defining equations for . en, 58 ∗ 68 ∗ {08 9 := ∈ O (*8 × * 9 )}, {18 9 := ∈ O (*8 × * 9 )} 59 69 are cocycle representatives for O() and O(). By the explicit description [MR208, (1-18)] of the cup-product map in Cechˇ cohomology, the map ∪ sends {18 9 } to the 2-cocycle
{(08 9 ,1 9: )} ∈ 2 (*8 × * 9 × *: ) . (14) ∗ Given a cocycle B8 9 ∈ O (*8 × * 9 × ) relative to the cover {*8 × } of , one computes its image under _ as follows. Pick B˜8 9 ∈ 2 (*8 × * 9 × * ) whose tame symbol along is B8 9 ; then check that its Cechˇ boundary m(B˜8 9 ) (a 2-cochain with values in K2,* ) is zero when viewed as a cochain with values in 8∗K1, . is means that m(B˜8 9 ) is a 2-cocycle with values in K2,- ; this is defined to be the image of B8 9 under _. Let us apply this to compute the image of O()| under _. ¯ Let 18 9 be the image of 18 9 under the map ∗ ∗ O (*8 × * 9 ) → O (*8 × * 9 × ) . ¯ e cocycle {18 9 } represents O()| . To compute its image under _, consider the element (symbol)
C8 9 = (58,18 9 ) ∈ 2 (* × *8 × * 9 ).
We know that 18 9 is a unit in *8 × * 9 and so defines an element of 1 (*8 × * 9 ); we know 58 is the defining equation of on *8 and so it is a unit on * × *8 and thus 58 defines an element of 1 (* × *8). So C8 9 is a well-defined element of 2 (* × *8 × * 9 ). If E denotes the valuation (-)∗ −→ ℤ defined by the divisor , the tame symbol map is the map
E(1) E(0)E(1) 0 2 (* ) → 1 (), (0,1) 7→ (−1) . . 1E(0)
Since E (58 ) = 1 and E (18 9 ) = 0, we see that C8 9 maps to the element 5 0 1×0 8 ¯−1 (−1) . 1 = 18 9 . 18 9 ¯ So the cochain {C8 9 } lis the inverse of the cocycle {18 9 }. Its Cechˇ boundary (which represents the image under _ of ¯ the inverse of {18 9 }) C8 9 − C8: + C 9: = (58,18 9 )−(58,18:)+(59 ,1 9: ) 12 ETTORE ALDROVANDI AND NIRANJAN RAMACHANDRAN
is a 2-cocycle with values in K2. Since 68 18 9 = 69 is a cocycle, the relation 18: = 18 9 + 1 9: ¯ holds. Using this, the image of the inverse of {18 9 } under _ is given by the negative of the element in (14):
59 (58,18 9 )−(58,18 9 )−(58,1 9: )+(59 ,1 9: ) = ( ,1 9: ) = −(08 9 ,1 9: ) . 58 ¯ 2 is says that _ maps {18 9 } to the class of the cup product of O() and O() in H (-, K2) thus completing the proof of the commutativity of the top triangle in (13). 5.1.2. e boom square of (13). We begin with an explicit description of the map 2 1 1 \ :H (-, K2) → H ((, R c∗K2) in (11). 2 2 Let be a K2-gerbe on -. As CH (-) = H (-, K2) (Bloch-illen), we can pick a codimension-two cycle 2 2 representing [ ] on -. As is horizontal, there exists an open cover {+U } of ( such that [ ] = 0 ∈ H (,U, K2), −1 with ,U = c (+U ); note {,U } is an open cover of -. In terms of the Gersten complex ord 0 −→ K2 −→ [∗K2,[ −→ 1 (:(G)) −−→ 0 (:(~)) → 0 , (1) (2) G ∈M,U ~ ∈M,U 2 which computes the cohomology of K2 on ,U , we have that the vanishing in (,U, K2) of the restriction 2U of the codimension-two cycle 2 representing [ ] on ,U . en, there exists an element ℎU ∈ (1) 1 (:(G)) such G ∈,U that ord(ℎU ) = 2U in the sequence on ,U . So ℎU is a collection of divisors in ,U whose associatedL functions cut out together the codimension-two cycle 2. Since ord(ℎU ) = ord(ℎU′ ) on,U ∩,U′ , we see that the element AU,U′ ≔ ℎU −ℎU′ 1 on ,U ∩ ,U′ defines an element of H (,U ∩ ,U′, K2). e cocycle condition is a formal consequence:
AU,U′ + AU′,U′′ + AU′′,U = 0 . 1 Namely, {AU,U′ } defines a Cechˇ 1-cocycle on ( with values in R c∗K2; this is the element \ ( ). Taking norms down to ( gives a Cechˇ 1-cocycle ℎU A˜U,U′ ≔ #/( ℎU′ with values in < on (. is completes the description of the maps in the boom squareof(13). With all this in place, it is now easy to show that the boom squareof(13) commutes. Recall the defining equations 68 of relative to the open cover {*8 } of -. Restricting O() to and applying _ gives the gerbe = O(),O() , by the commutativity of the top triangle of (13). We use the above description to compute the image of the gerbe under
\; we see that ℎU can be taken to be the collection of functions 6¯8,U = 6¯8 |U,8 on U,8 = ∩ ,U ∩ *8 which cuts out the codimension-two cycle corresponding to the intersection of and . e norm down to ( of the corresponding AU,U′ gives the Cech-cocycleˇ with values in K1 of (; this is the image of O() along one part of the boom square in (13). On the other hand, consider the image of O() under the le vertical map of (13). Let
6¯8,U = 6¯8 |U,8 , 4U = #U,8 /( 6¯8,U . Y8 e image of O() under the map #/( is given by the cocycle 4U ∗ 2U,U′ ≔ ∈ O (+U ∩ +U′) . 4U′
It is clear that 2U,U′ is equal to Norm(AU,U′). is shows the commutativity of the diagram (13), since the image of O() along the vertical le map of (13) gives Deligne’s line bundle hO(), O()i , and the image along the other side of (13) is
(O(), O()) = Norm ◦ Θ ◦∪(O(), O()) = Norm ◦ Θ ◦( O(),O() ) = O(),O() . ¹c is proves eorem 4 and therefore eorem 1. FIBER INTEGRATION OF GERBES AND DELIGNE LINE BUNDLES 13
6. Picard stacks and their endomorphisms Here and elsewhere in this paper “Picard stack,” or “Picard category,” means “strictly commutative Picard” in the sense of Deligne [Del73]. Namely, if we denote the monoidal operation of P simply by +: P × P → P, then the symmetry condition given by the natural isomorphisms fG,~ : G + ~ → ~ + G must satisfy the additional condition that fG,G = idG+G . Such stacks have the pleasant property that there exists a two-term complex of abelian sheaves 3 : −1 → 0 such that F is equivalent, as a Picard stack, to the one associated to the action groupoid formed from the complex. We denote this situation by 3 ∼ P ≃ −1 −→ 0 . A classical example arises from the well known divisor exact sequence of Zariski sheaves × 0 → < → [∗ (. ) → (8~)∗ℤ → 0 , (15) ~M∈. 1 where . is a smooth scheme over a field and the sum is over the set . 1 of points of codimension one in . . We get the equivalence: × ∼ Tors. (<) ≃ [∗ (. ) → (8~)∗ℤ ~M∈. 1 CH1 1 In the sequel we shall denote by . the Picard stack on the right hand side of the above relation and by CH (. ) 1 the Picard category of its global sections. erefore we have Tors. (<) ≃ CH (. ). Still from [Del73], we have that morphisms and natural transformations form a Picard stack Hom(P, Q), where the additive structure is defined pointwise: if , are two objects, then ( + )(G) ≔ (G)+Q (G), for any object G of P. It is immediate to verify that this is symmetric and in fact strictly commutative if +Q is.
6.1. Ring structures. We set End(P) = Hom(P,P). By the above considerations, it is a Picard stack, but the composition of morphisms gives it an additional unital monoidal structure, with respect to which End(P) acquires the structure of a stack of ring groupoids—also known as categorical rings—of the sort described in [Ald15] (see also [Dri21]foraresum´ e).´ Note that the “multiplication” monoidal structure, being given by composition of functors, is strictly associative. Remark 17. IfP ≃ [−1 → 0]∼,thenby[AN09] End(P) is equivalent, as a stack of ring groupoids, to Corr(•, •), the stack whose objects are buerfly diagrams: an object is given by an extension 0 → • → → 0 → 0 such that its pullback to −1 via 3 : −1 → 0 is trivial. Morphisms are morphisms of extensions. Corr(•, •) is a stack of ring groupoids: the “+” is given by the Baer sum of extensions; the “×” is given by concatenation of buerflies described in loc. cit. is structure is associative, but not strictly so. 6.2. otients and colimits. Let : P → Q be a morphism of Picard stacks. Its cokernel Coker is the stack associated to the following construction, the details of which can be found in the literature (see, e.g. [Vit02, KV00]).3 Let assume : P → Q is a morphism of Picard categories. e cokernel Coker is a Picard category defined as follows: (1) its class of objects is the same as that of Q; (2) a morphism [5 ,0] : G → ~ is an equivalence class of pairs (5 ,0), where 5 is morphism 5 : G → G + (0) in Q, 0 ∈ Obj P, and two pairs (5 , 0) and (6,1) are equivalent if there exists an arrow D : 0 → 1 in P and the
diagram
G ← 5 → 6
← → ~ + 0) ← → ~ + (1) ~+ (D) commutes. (3) e monoidal structure is defined to be that of Q on objects and by the class of the composite arrow ≃ G + ~ → (G + (0))+(~ + (1)) −→ (G + (0))+(~ + (1)) ≃ ≃ −→ (G + ~)+( (0)+ (1)) −→ (G + ~)+ (0 + 1) if [5 ,0] : G → G ′ and [6,1] : ~ → ~′.
3is is valid for Picard categories and stacks that not necessarily strictly commutative. 14 ETTORE ALDROVANDI AND NIRANJAN RAMACHANDRAN
ere is a canonical functor ? : Q → Coker which is the identity on objects and sends an arrow 5 : G → ~ in Q to
the class of the composite:
← ← 5 ← ≃ ≃
G→ ~→ ~ + 0Q → ~ + (0P) . ≃ ← ere is an isomorphism c : ? ◦ ⇒ 0: P → Q given by c,0 : (0) → 0Q + (0) . It follows that we have an abelian group isomorphism c0 (Coker ) Coker c0 () : c0 (P) → c0 (Q) . As mentioned, if : P → Q is a morphism of Picard stacks, we define Coker to be the Picard stack associated to the pseudo-functor * Coker * : P(* ) → Q(* ) where * is in the base site.
In the following section, we apply this construction to the diagram
→ ← 1 2 P1 → Q← P2
and the resulting morphism 1 + 2 : P1 × P2 → Q, defined on objects by (1 + 2)(G1, G2) = 1 (G1)+ 2 (G2). We shorten or notation and simply write Coker(1 + 2) as Q/(P1 + P2). By the above recollection we have
c0 Q/(P1 + P2) c0 (Q)/(c0 (P1)+ c0 (P2)) . 7. Categorification of correspondences In this section, ( = Spec and a curve is a smooth projective connected one-dimensional scheme over (. For simplicity, we assume (just in this section) that is algebraically closed. e main result (eorem 18) of this section is an application of eorem 1 using §6 to the categorification of well known identities (16) and (19) about correspondences on the self-product of a curve. We will work with the category V whose objects are curves and the maps are (correspondences) Hom V (,) = Div(×) with composition defined by product of correspondences. In the following we are stating our results for Picard categories, but there are parallel statements for the corre- sponding Picard stacks. 7.1. Categorifying CH1 (. ). For any smooth scheme . over (, it follows from (15) that the Chow group CH1 (. ) is 1 1 isomorphic to the Picard group Pic(. ) = H (., <) of . ; the Picard category CH (. ) is canonically equivalent to the Picard category of <-torsors or line bundles on . . e Picard category CH1 (. ) categorifies the Chow group CH1 (. ) of divisors: 1 1 1 (1) CH (. ) = Pic(. ) = H (., <) = c0 (CH (. )); (2) Any map 5 : . → . ′ of smooth schemes defines an additive functor of Picard categories 5 ∗ : CH1 (. ′) → CH1 (. ), ! 7→ 5 ∗! ; ∗ 1 ′ 1 the induced map on c0 is the pullback of divisors 5 :CH (. ) → CH (. ). For a curve , let Pic0 () be the kernel of the degree map Pic() → ℤ. If CH1 ()0 is the sub-Picard category of 1 1 0 0 CH () consisting of line bundles of degree zero, then c0 (CH () ) = Pic ().
7.2. Correspondences. We refer to [Ful98, Chapter 16] for details. Let and be curves and let c and c be the two projections on × . A correspondence U : ⊢ from to is a divisor U on × . It defines a line bundle O(U) on × . e correspondence U acts on divisors: it induces a map ∗ ∗ U : Pic() → Pic() < 7→ (c )∗(c<.U) ∗ which sends a divisor < on to the pushforward along c of the intersection of U and c< on × . It restricts to a map Pic0 () → Pic0 (): if < has degree zero, then so does U∗ (<). We get a homomorphism ) :CH1 ( × ) → Hom(Pic(), Pic()) → Hom(Pic0 (), Pic0 ()) , U 7→ U∗ (16) as U∗ depends only on the class of U in CH1 ( × ). ∗ 1 ∗ 1 Degenerate correspondences [Ful98, Example 16.1.2] constitute the subgroup (,) = c ( ())+c ( ()) of CH1 ( × ). e map ) induces an isomorphism [Sch94, Proposition 3.3, eorem 3.9] CH1 ( × ) ) : → Hom(Pic0 (), Pic0 ()); (17) (,) FIBER INTEGRATION OF GERBES AND DELIGNE LINE BUNDLES 15
see [BL04, Chapter 11, eorem 5.1] for another proof when = ℂ. Over a non-algebraically closed field, the isomorphism holds if and have rational points. Composition of correspondences induces a ring structure on CH1 ( ×) with (,) as an ideal [Ful98, Example 16.1.2]. It is known that • [Ful98, Corollary 16.1.2] the map ) :CH1 ( × ) → End(CH1 ()) , U 7→ U∗ (18) is a homomorphism of rings. • ) induces a ring isomorphism [Ful98, Example 16.1.2(c)] 1 ( × ) ) : → End(Pic0 ()), (19) (,) as is algebraically closed; see [BL04, Chapter 11, eorem 5.1] for a proof when = ℂ. e following result provides a categorification of the above statements. eorem 18. ere is an additive functor of Picard categories )˜ : CH1 ( × ) → Hom(CH1 (), CH1 ()) → Hom(CH1 ()0, CH1 ()0) (20)
which induces (16) on c0. )˜ has the following properties: ∗ ∗ (i) Let "(,) = Coker(c + c ) be the cokernel of the additive functors ∗ 1 1 1 ∗ c : CH () → CH ( × ) ← CH () : c . (21) en )˜ induces an additive functor 1 0 1 0 )˜, : "(,) → Hom(CH () , CH () ) (22)
which, on c0,is(17). 1 (ii) CH ( × ) comes naturally equipped with the structure of a categorical ring (§6.1) which, on c0, is the compo- sition of correspondences. (iii) the functors 1 1 1 0 )˜ : CH ( × ) → End(CH ()), )˜, : "(,) → End(CH () ) (23)
are functors of categorical rings (§6.1). ese induce (18), (19) on c0. Remark 19. e assignment 7→ CH1 () and 6 ∈ Div( × ) 7→ )˜ (O(6)) comprise a pseudo-functor from V (the category of curves and correspondences) to the 2-category of Picard categories (or stacks). 7.3. Proof of eorem 18. e existence of )˜ is provided by the following lemma. Lemma 20. For any correspondence U : ⊢ , the map U∗ :CH1 () → CH1 () is induced by an additive functor U˜∗ : CH1 () → CH1 (). is functor restricts to a functor CH1 ()0 → CH1 ()0. Further, if V is another correspon- ∗ dence, then (U + V) = U˜∗ + V˜∗ as additive functors. ∗ O Proof. is is a simple application of eorem 1. Given a line bundle " on , consider the pair c " and (U) of line bundles on ×;[AR16] constructs a K2-gerbe O ∗ on ×. As this gerbe is horizontal by Proposition ( (U),c ") 15, one can integrate it along c : × → to get a line bundle on :
∗ U˜ " = O ∗ . ( (U),c ") ¹c ∗ Both the additivity ofU ˜∗ and the property (U + V) = U˜∗ + V˜∗ follow from the bi-additivity (eorem 3) of . If the line bundle " on has degree zero, then so does the line bundleU ˜∗" on as its class in 1 () is U∗ (") which has degree zero. is gives us a bi-additive functor of Picard categories CH1 ( × )× CH1 () → CH1 (), (U,") 7→ U˜∗", and an additive functor )˜ : CH1 ( × ) → Hom(CH1 (), CH1 ()) → Hom(CH1 ()0, CH1 ()0) where, for any pair %, % ′ of Picard categories, Hom(%, % ′) is the Picard category of additive functors from % to % ′. 16 ETTORE ALDROVANDI AND NIRANJAN RAMACHANDRAN
Statement (i) of eorem 18 concerns the factorization of )˜ as )˜ : "(,) → Hom(CH1 ()0, CH1 ()0) (24) ˜ ∗ 1 ∗ 1 is, in turn, follows from the triviality of ) on c CH () and c CH (): ˜ ∗ 1 • ) restricted to c CH (). ∗ ˜ If 6 : ⊢ is the pullback cG of a divisor G on , then ) (6) applied to a line bundle ! on is defined as (O(6),c ∗ !) . As the construction of the K2-gerbe is functorial, we have c ∫ ∗ ∗ ∗ = c ; (c G,c !) (G,!) 2 ˜ as (, K2) = 0, the K2-gerbe (G,!) on is trivializable. Since is an additive functor, ) (6)(!) is c trivializable. It follows that ∫ )˜ (6) : CH1 ()0 → CH1 ()0 is the trivial functor. ˜ ∗ 1 • ) restricted to c CH (). ∗ ˜ If 6 : ⊢ is cG of a divisor G on and < = < 9~9 is a divisor on , then ) (6)(<) corresponds to O the d46 <-th power of the line bundle (G) and henceP is trivial when < has degree zero. is can be seen as follows: )˜ (6)(<) is the object corresponding to the line bundle
∗ ∗ ∗ ∗ ⊗< 9 hc<,cGi = ⊗9 hc~9 ,cGi . ↩ ∗ ∗ O Since c : ×~9 → × → is an isomorphism for any closed point ~9 of , one has hc~9 ,cGi = (G) by (2). By bi-additivity, ∗ ∗ O deg < hc<,cGi = ( (G)) . If < has degree zero, then )˜ (6)(<) is trivializable. So the functor 6˜∗ = )˜ (6) : CH1 ()0 → CH1 ()0 is trivial. is completes the proof of (i) of eorem 18.
7.4. Composition. We show that )˜ is compatible with composition of correspondences. Let - = 1 × 2 × 3 be the product of three curves 1,2,3 and let c8 9 : - → 8 × 9 be the projections. If 6 : 2 ⊢ 1 is a correspondence on 1 × 2 and ℎ : 3 ⊢ 2 on 2 × 3, one can compose 6 and ℎ to get a correspondence 6 ◦ ℎ : 3 ⊢ 1 on 1 × 3: ∗ ∗ 6 ◦ ℎ = (c13)∗(c23ℎ.c126), by pulling back 6 and ℎ to - and intersecting them and pushing forward via c13 to 1 × 3. is gives a bi-additive map 1 1 1 ◦ :CH (1 × 2)× CH (2 × 3) → CH (1 × 3). Lemma 21. e above bi-additive map is induced by a bi-additive functor 1 1 1 ◦˜ : CH (1 × 2)× CH (2 × 3) → CH (1 × 3).
∗ ∗ Proof. e functor is defined as follows: e pair c12O(6) and c23O(ℎ) of line bundles on - give rise to a K2-gerbe ∗ ∗ (c12 O(6),c23 O(ℎ)) on -. Since it is horizontal (Proposition 15) for the map (a relative curve) c13 : - → 1 × 3, we ∗ ∗ can integrate it along c13 to obtain a line bundle hc12O(6),c23O(ℎ)i on 1 × 3. e functor ◦˜, in the notation of eorem 1, is ∗ ∗ ∗ ∗ O O ◦˜ : (6,ℎ) 7→ (c12 O(6),c23 O(ℎ)) = hc12 (6),c23 (ℎ)i. ¹c13
It follows from eorem 1 that ◦˜ induces ◦ on c0.
Taking 1 = 2 = 3 = proves (ii) of eorem 18. FIBER INTEGRATION OF GERBES AND DELIGNE LINE BUNDLES 17
Lemma 22. e functor )˜ is compatible with composition: namely, the diagram 1 1 ← ◦˜ 1 CH (1 × 2)× CH (2 × 3) → CH (1 × 3) → →
)˜×)˜ ← )← ˜ , (25) 1 1 1 1 ← 1 1 Hom(CH (2), CH (1)) × Hom(CH (3), CH (2)) → Hom(CH (3), CH (1))
commutes up to natural isomorphisms )˜ (6◦˜ℎ) )˜ (6) ◦ )˜ (ℎ).
Proof. For any smooth projective morphism 5 : . → of relative dimension one and line bundles !1, !2 on . , let h i = !1, !2 5 5 (!1,!2) denote the Deligne line bundle on . Our task is to prove the existence of a natural isomorphism 1 for any ! ∈∫CH (3): ∗ O ∗ O ∗ O ∗ O ∗ hhc12 (6),c23 (ℎ)ic13 , U3!iU1 h (6),W2 h (ℎ), V3!iV2 iW1 (26)
where the maps are
← ← U← 3 V3 W2 1 × 3 → 3, 2 × 3 → 3, 1 × 2 → 2. → → → ← U← 1 V← 2 W1
1 2 1
By additivity in !, it suffices to consider the case ! = O(G) for a closed point G = Spec of 3. We put
]1 : 1 = 1 × G ↩→ 1 × 3, ]2 : 2 = 2 × G ↩→ 2 × 3,
]12 : 1 × 2 1 × 2 × G ↩→ 1 × 2 × 2 = - . By (2), the le-hand-side of (26) is
∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ hhc O(6),c O(ℎ)ic , U !iU # / (hc O(6),c O(ℎ)ic ) ] hc O(6),c O(ℎ)ic . 12 23 13 3 1 1 12 23 13 1 12 23 13
On the other hand, by (2), the right-hand-side of (26) is O ∗ O ∗ O ∗ O O ∗ ∗O O ∗ ∗ O h (6),W2 h (ℎ), V3!iV2 iW1 h (6),W2 #/2 ( (ℎ) )iW1 h (6),W2]2 (ℎ)iW1 h (6), ]12c23 (ℎ)iW1 ,
using 2 for the second isomorphism and the following diagram for the last isomorphism:
← ← 1 × 2 → 1 × 2 × G→ 1 × 2 × 3 = -
→ →
← W2 c← 23
← ← 2 → → 2 × 3 , where the top row is ]12 and the boom one is ]2. As c12 ◦ ]12 is the identity map on 1 × 2, we have O ∗ ∗ O ∗ ∗ O ∗ ∗ O h (6), ]12c23 (ℎ)iW1 h]12c12 (6), ]12c23 (ℎ)iW1 . e required natural isomorphism in (26), namely,
∗ ∗ ∗ ∗ ∗ ∗ ∗ h] c O(6), ] c O(ℎ)iW ] hc O(6),c O(ℎ)ic 12 12 12 23 1 1 12 23 13
follows from functoriality: use the map of relative curves
← ← 1 × 2 → 1 × 2 × G→ 1 × 2 × 3 = -
→ →
← W1 c← 13
← ← 1 → → 1 × 3 , where the top row is still ]12 and the boom one is now ]1. is proves Lemma 22.
Taking = 1 = 2 = 3 in the above lemma, we obtain that ◦˜ : CH1 ( × )× CH1 ( × ) → CH1 ( × ) is a monoidal functor of Picard categories and that )˜ : CH1 ( × ) → End(CH1 ()) 18 ETTORE ALDROVANDI AND NIRANJAN RAMACHANDRAN is a functor of ring categories proving (iii). is finishes the proof of eorem 18.
References [Ald15] Eore Aldrovandi, Stacks of ann-categories and their morphisms, eory and Applications of Categories 30 (2015), 1256–1286, URL: http://www.tac.mta.ca/tac/volumes/30/39/30-39abs.html. 6.1 [AN09] Eore Aldrovandi and Behrang Noohi, Buerflies i: Morphisms of 2-group stacks, Advances in Mathematics 221 (2009), 687–773, arXiv:0808.3627, doi:doi:10.1016/j.aim.2008.12.014. 17 [AR16] E. Aldrovandi and N. Ramachandran, Cup products, the Heisenberg group, and codimension two algebraic cycles, Doc. Math. 21 (2016), 1313–1344. MR 3578206 (document), 1, 3.4.3, 4.1, 4.1, 4.1, 4.1, 7.3 [AR21] , Algebraic cycles and categorification, In preparation. 1 [BL04] Christina Birkenhake and Herbert Lange, Complex abelian varieties, second ed., Grundlehren der Mathematischen Wissenschaen [Fun- damental Principles of Mathematical Sciences], vol. 302, Springer-Verlag, Berlin, 2004, doi:10.1007/978-3-662-06307-1. MR 2062673 7.2, 7.2 [Blo84] S. Bloch, Height pairings for algebraic cycles, J. Pure Appl. Algebra 34 (1984), 119–145, doi:10.1016/0022-4049(84)90032-X. MR 772054 1 [Blo89] , Cycles and biextensions, Algebraic -theory and algebraic number theory (Honolulu, HI, 1987), Contemp. Math., vol. 83, Amer. Math. Soc., Providence, RI, 1989, pp. 19–30, doi:10.1090/conm/083/991974. MR 991974 1 [Bre78] L. Breen, Extensions du groupe additif, Inst. Hautes Etudes´ Sci. Publ. Math. (1978), no. 48, 39–125. MR MR516914 (81f:14011) 4.1 [Bre83] , Fonctions thetaˆ et theor´ eme` du cube, Lecture Notes in Mathematics, vol. 980, Springer-Verlag, Berlin, 1983. MR MR823233 (87d:14036) 13 [Bre94] , On the classification of 2-gerbes and 2-stacks, Asterisque´ (1994), no. 225, 160. MR 95m:18006 3.1, 3.2.1, 3.2.2 [BS88] A. A. Be˘ılinson and V. V. Schechtman, Determinant bundles and Virasoro algebras, Comm. Math. Phys. 118 (1988), no. 4, 651–701, URL: http://projecteuclid.org/euclid.cmp/1104162170. MR 962493 1, 1, 4.3 [Del73] Pierre Deligne, La formule de dualite´ globale,eorie´ des Topos et Cohomologie Etale des Schemas´ (M. Artin, A. Grothendieck, and J. L. Verdier, eds.), vol. Tome 3, Springer Berlin Heidelberg, 1973, pp. 481–587. 6, 6 [Del87] P. Deligne, Le determinant´ de la cohomologie, Current trends in arithmetical algebraic geometry (Arcata, Calif., 1985), Contemp. Math., vol. 67, Amer. Math. Soc., Providence, RI, 1987, pp. 93–177, doi:10.1090/conm/067/902592. MR 902592 1, 2.1 [Del91] , Le symbole moder´ e´, Inst. Hautes Etudes´ Sci. Publ. Math. (1991), no. 73, 147–181, URL: http://www.numdam.org/item?id=PMIHES 1991 73 147 0. MR 1114212 13 [Dri21] Vladimir Drinfeld, On a notion of ring groupoid, 2021, hps://arxiv.org/abs/2104.07090, arXiv:2104.07090. 6.1 [Dus02] J. W. Duskin, Simplicial matrices and the nerves of weak =-categories. I. Nerves of bicategories, vol. 9, 2001/02, CT2000 Conference (Como), pp. 198–308. MR 1897816 3.2 [Elk89] R. Elkik, Fibres´ d’intersections et integrales´ de classes de Chern, Ann. Sci. Ecole´ Norm. Sup. (4) 22 (1989), no. 2, 195–226, URL: http://www.numdam.org/item?id=ASENS 1989 4 22 2 195 0. MR 1005159 1 [Ems17] M. Emsalem, Twisting by a torsor, Analytic and algebraic geometry, Hindustan Book Agency, New Delhi, 2017, pp. 125–152, doi:10.1007/978-981-10-5648-2 9. MR 3728130 8 [Eri13] D. Eriksson, Deligne’s functorial riemann-roch theorem, Various papers on his webpage. 1 [Fra90] J. Franke, Chow categories, Compositio Math. 76 (1990), no. 1-2, 101–162, Algebraic geometry (Berlin, 1988), URL: http://www.numdam.org/item?id=CM 1990 76 1-2 101 0. MR 1078860 1 [Fra91] , Chern functors, Arithmetic algebraic geometry (Texel, 1989), Progr. Math., vol. 89, Birkhauser¨ Boston, Boston, MA, 1991, pp. 75– 152. MR 1085257 1 [Ful98] William Fulton, Intersection theory, second ed., Ergebnisse der Mathematik und ihrer Grenzgebiete. 3. Folge. A Series of Modern Surveys in Mathematics [Results in Mathematics and Related Areas. 3rd Series. A Series of Modern Surveys in Mathematics], vol. 2, Springer-Verlag, Berlin, 1998, doi:10.1007/978-1-4612-1700-8. MR 1644323 1, 7.2, 7.2, 7.2, 7.2 [Gil05] H. Gillet, -theory and intersection theory, Handbook of -theory. Vol. 1, 2, Springer, Berlin, 2005, pp. 235–293, doi:10.1007/3-540-27855-9 7. MR 2181825 4.4 [Gir71] J. Giraud, Cohomologie non abelienne´ , Springer-Verlag, Berlin-New York, 1971, Die Grundlehren der mathematischen Wissenschaen, Band 179. MR 0344253 1, 1, 3, 3.1, 5, 7, 3.2.1, 3.3, 10, 3.4.3, 4.1 [Gor09] S. Gorchinskiy, Notes on the biextension of Chow groups, Motives and algebraic cycles, Fields Inst. Commun., vol. 56, Amer. Math. Soc., Providence, RI, 2009, pp. 111–148, doi:10.1016/s0315-0860(08)00015-3. MR 2562455 1 [Gro72] A. Grothendieck, Biextensions de faisceaux de groupes, Groupes de monodromie en geom´ etrie´ algebrique.´ I, Lecture Notes in Mathematics, vol. 288, Springer-Verlag, 1972, Seminaire´ de Geom´ etrie´ Algebrique´ du Bois-Marie 1967–1969 (SGA 7 I), Dirige´ par A. Grothendieck. Avec la collaboration de M. Raynaud et D. S. Rim, pp. 133–217. 13 [Ill71] L.Illusie, Complexe cotangent et deformations.´ i, Lecture Notes in Mathematics, Vol. 239, Springer-Verlag, Berlin, 1971. MR MR0491680 (58 #10886a) 4.1 [Jar15] J. F. Jardine, Local homotopy theory, Springer Monographs in Mathematics, Springer, New York, 2015, URL: https://doi-org.proxy.lib.fsu.edu/10.1007/978-1-4939-2300-7, doi:10.1007/978-1-4939-2300-7. MR 3309296 3.1 [Kai16] W. Kai, A higher-dimensional generalization of Lichtenbaum duality in terms of the Albanese map, Compos. Math. 152 (2016), no. 9, 1915– 1934, doi:10.1112/S0010437X16007600. MR 3568943 3.5 [KS06] M. Kashiwara and P. Schapira, Categories and sheaves, Grundlehren der Mathematischen Wissenschaen [Fundamental Principles of Mathematical Sciences], vol. 332, Springer-Verlag, Berlin, 2006. MR MR2182076 (2006k:18001) 3.2 [KV00] Stefano Kasangian and Enrico Vitale, Factorization systemms for symmetric cat-groups, eory and Applications of Categories 7 (2000), no. 5, 47–70, URL: http://www.tac.mta.ca/tac/volumes/7/n5/7-05abs.html. 6.2 FIBER INTEGRATION OF GERBES AND DELIGNE LINE BUNDLES 19
[Lic69] S. Lichtenbaum, Duality theorems for curves over ?-adic fields, Invent. Math. 7 (1969), 120–136, doi:10.1007/BF01389795. MR 0242831 3.5 [Mil80] J.S. Milne, Etale´ cohomology, Princeton Mathematical Series, vol. 33, Princeton University Press, Princeton, N.J., 1980. MR 559531 3 [Mil03] , Gerbes and abelian motives, Preprint, available at www.jmilne.org. 3.5 [MR208] Loop spaces, characteristic classes and geometric quantization, Modern Birkhauser¨ Classics, Birkhauser¨ Boston, Inc., Boston, MA, 2008, Reprint of the 1993 edition, doi:10.1007/978-0-8176-4731-5. MR 2362847 8, 5.1.1 [Pat08] M. M. Patnaik, Geometry of loop Eisenstein series, Proest LLC, Ann Arbor, MI, 2008, esis (Ph.D.)–Yale University, URL: http://gateway.proquest.com/openurl?url ver=Z39.88-2004&rft val fmt=info:ofi/fmt:kev:mtx:dissertation&res dat=xri:pqdiss&rft dat=xri:pqdiss:3317198. MR 2711828 1 [Ros96] M. Rost, Chow groups with coefficients, Doc. Math. 1 (1996), No. 16, 319–393. MR 1418952 4.4, 5.1 [Sch94] A. J. Scholl, Classical motives, Motives (Seale, WA, 1991), Proc. Sympos. Pure Math., vol. 55, Amer. Math. Soc., Providence, RI, 1994, pp. 163–187, doi:10.1090/pspum/055.1/1265529. MR 1265529 7.2 [SGA73] eorie´ des topos et cohomologie etale´ des schemas.´ Tome 3, Lecture Notes in Mathematics, Vol. 305, Springer-Verlag, Berlin-New York, 1973, Seminaire´ de Geom´ etrie´ Algebrique´ du Bois-Marie 1963–1964 (SGA 4), Dirige´ par M. Artin, A. Grothendieck et J. L. Verdier. Avec la collaboration de P. Deligne et B. Saint-Donat. MR 0354654 1, 1, 2.2, 2.2, 3.1, 3.2 [Sko07] A. N. Skorobogatov, On the elementary obstruction to the existence of rational points, Mat. Zametki 81 (2007), no. 1, 112–124, doi:10.1134/S0001434607010099. MR 2333868 3.5 [Sta20] e Stacks Project Authors, Stacks Project, https://stacks.math.columbia.edu, 2020. 2.2, 3.1 [Vit02] Enrico M. Vitale, A picard–brauer exact sequence of categorical groups, Journal of Pure and Ap- plied Algebra 175 (2002), no. 1, 383–408, Special Volume celebrating the 70th birthday of Pro- fessor Max Kelly, URL: https://www.sciencedirect.com/science/article/pii/S0022404902001421, doi:https://doi.org/10.1016/S0022-4049(02)00142-1. 6.2
Ettore Aldrovandi, Department of Mathematics, Florida State University, Tallahassee, FL 32306-4510 USA. Email address: [email protected] URL: http://www.math.fsu.edu/˜ealdrov
Niranjan Ramachandran, Department of Mathematics, University of Maryland, College Park, MD 20742 USA. Email address: [email protected] URL: http://www2.math.umd.edu/˜atma/