ebook img

HIGHER ARITHMETIC K-THEORY Introduction The aim of this paper is to provide a new definition ... PDF

76 Pages·2012·0.47 MB·English
by  
Save to my drive
Quick download
Download
Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.

Preview HIGHER ARITHMETIC K-THEORY Introduction The aim of this paper is to provide a new definition ...

HIGHER ARITHMETIC K-THEORY YUICHIRO TAKEDA Introduction The aim of this paper is to provide a new de(cid:12)nition of higher K-theory in Arakelov geometry and to show that it enjoys the same formal properties as the higher algebraic K-theory of schemes. LetX beaproperarithmeticvariety,thatis,letX bearegularschemewhichis(cid:13)at,proper and of (cid:12)nite type over Z, the ring of integers. In the research on arithmetic Chern characters of hermitian vector bundles on X, Gillet and Soul(cid:19)e de(cid:12)ned the arithmetic K -group K (X) 0 0 of X [10]. It can be viewed as an analogue in Arakelov geometry of the K -group of vector 0 bundles on a scheme. b After the advent of K (X), its higher extension was discussed in some papers, such as 0 [5, 6, 15]. In these papers one common thing was suggested that higher arithmetic K-theory should be obtained as thbe homotopy group of the homotopy (cid:12)ber of the Beilinson’s regulator map. That is to say, it was predicted that there would exist a group KM (X) for each n 0 n (cid:21) satisfying the long exact sequence Kn+1(X) (cid:26) HD2p(cid:0)n(cid:0)1(X;R(p)) KMn(X) Kn(X) ; (cid:1)(cid:1)(cid:1) ! ! (cid:8)p ! ! ! (cid:1)(cid:1)(cid:1) where Hn(X;R(p)) is the real Deligne cohomology and (cid:26) is the Beilinson’s regulator map. D To get the homotopy (cid:12)ber, a simplicial description of the regulator map is necessary. And it was given by Burgos and Wang in [5]. For a compact complex manifold M, they de(cid:12)ned an exact cube of hermitian vector bundles on M and associated with it a di(cid:11)erential form called a higher Bott-Chern form. This leads us to a homomorphism of complexes ch : ZS (M) D (M;p)[2p+1] (cid:3) (cid:3) ! from the complex ZS (M) associated with the S-construction of the category of hermitian vector bundles on M(cid:3)to the complebx D (M;p) computing the real Deligne cohomology of (cid:3) M, which is de(cid:12)nedbin [3]. It is the main theorem of [5] that the following map coincides with the higher Chern character map with values in Deligne cohomology: (cid:26) : Kn(M) (cid:25)n+1(S(M)) Hurewicz Hn+1(S(M)) H(ch) HD2p(cid:0)n(M;R(p)): ’ (cid:0)! (cid:0)! Applying the theory of higher Bott-Chern forms to an arithmetic variety, we can obtain a b b simplicial description of the regulator map. In this paper, we will give another de(cid:12)nition of higher arithmetic K-theory of a proper arithmetic variety by means of higher Bott-Chern forms, which di(cid:11)ers from the one coming 1991 Mathematics Subject Classi(cid:12)cation. Primary 14G40; Secondary 11G35, 19E08. Key words and phrases. K-theory, Arakelov geometry, higher Bott-Chern form. 1 2 from the homotpy (cid:12)ber of the regulator map. One of the remarkable features of our arith- metic K-theory is that it is given as an extension of the usual K-theory by the cokernel of the regulator map. Beforeexplainingourmethod,letusrecallthede(cid:12)nitionofK (X). Foraproperarithmetic 0 variety X, let Ap;p(X) be the space of real (p;p)-forms ! on X(C) such that F(cid:3) ! = ( 1)p! (cid:0) for the complex conjugation F : X(C) X(C) and let A(bX) = Ap;p(X)=1Im@ +Im@. 1 ! (cid:8)p Then K (X) is de(cid:12)ned as a factor group of the free abelian group generated by pairs (E;!) 0 e of a hermitian vector bundle E on X and ! A(X). The relation on pairs is given by each 2 short ebxact sequence E : 0 E E E 0 as follows: 0 00 ! ! ! ! e (E ;! )+(E ;! ) = (E;! +! +ch(E)); 0 0 00 00 0 00 where ch(E) is the Bott-Chern secondary characteristic class of E and ! ;! A(X). 0 00 e 2 We can interpret the above de(cid:12)nition of K (X) in terms of loops and homotopies on 0 S(X) ,ethe topological realization of the S-construction of the category of hermeitian vector j j bundles on X. Let us consider a pair (l;!), wbhere l is a pointed simplicial loop on S(X) j j abnd ! A(X). Two pairs (l;!) and (l ;! ) are de(cid:12)ned to be homotopy equivalent if there 0 0 2 is a cellular homotopy H : S1 I= I S(X) from l to l such that the Bott-Cbhern 0 (cid:2) f(cid:3)g(cid:2) ! j j secondaryecharacteristic class ch(H) of H, which is de(cid:12)ned in a natural way, is equal to the di(cid:11)erence ! !. Let (cid:25) ( S(X) ;ch) denote thebset of all equivalence classes of such pairs. 0 1 (cid:0) j j Then it carries the structure ofean abelian group and the map b b Ke (X) (cid:25) ( S(X) ;ch) 0 1 ! j j de(cid:12)ned by (E;!) (l ; !), where l is the simplicial loop on S(X) determined by E, is 7! E (cid:0) b E b b e j j shown to be bijective. In this paper, we will generalize the above construction of (cid:25) tob higher homotopy groups 1 to de(cid:12)ne higher arithmetic K-groups. Strictly speaking, we will de(cid:12)ne the n-th arithmetic K-group K (X) as the set of all homotopy equivalence classes of pairs (f;!), where f : n b Sn+1 S(X) is a pointed cellular map and ! is a real di(cid:11)erential form on X(C) modulo ! j j exact formbs. We will employ the theory of higher Bott-Chern forms in de(cid:12)ning a homotopy equivalencbe relation on these pairs. Let us describe the content of the paper in more detail. In 1we introduce some materials used inthe present paper, such as S-construction, cubes x and higher Bott-Chern forms. Furthermore, by renormalizing the higher Bott-Chern forms, we obtain a homomorphism of complexes chGS : ZS (X) A (X)[1]; (cid:3) ! (cid:3) where A (X) is a complex of vector spaces of real di(cid:11)erential forms on X(C) satisfying a (cid:3) b certainHodgetheoreticcondition. In 2weproposethenotionofmodi(cid:12)edhomotopygroups, x which is a higher generalization of the above (cid:25) . In 3 we de(cid:12)ne higher arithmetic K-groups 1 x K (X) as the homotopy groups of S(X) modi(cid:12)ed by the homomorphism chGS. We show (cid:3) j j the following exact sequence concerning K (Xb): n b b K (X) A (X) K (X) K (X) 0; n+1 ! n+1 b ! n ! n ! e b 3 where An+1(X) = An+1(X)=ImdA. When n = 0, this exact sequence has already been obtained in [10]. Moreover, we de(cid:12)ne the Chern form map e chGS : K (X) A (X) n n ! n in the same way as ch : K (X) A(X) in [10]. The group KM (X), which is characterized 0 n ! b as the long exact sequence as mentioned before, is realized as the kernel of chGS. When n we (cid:12)x an F -invariant Kba(cid:127)hler metric h on X(C), we de(cid:12)ne Arakelov K-group K (X) of X n 1 X = (X;h ) as the subgroup of K (X) consisting of all elements x K (X) such that X n n 2 chGS(x) is harmonic with respect to h . The group K (X) (cid:12)ts into the exact sequence n X n Kn+1(X) (cid:26) HD2p(cid:0)bn(cid:0)1(X;R(p)) Kn(X) Kn(X) 0;b ! (cid:8)p ! ! ! where (cid:26) is the Beilinson’s regulator map. The next two sections concern product structure on K (X). In 4 we prove a product (cid:3) x formula for higher Bott-Chern forms. It provides an alternative proof of the fact that the regulator map respects the products. In 5, we de(cid:12)ne abproduct in higher arithmetic K- x theory. We show that this product is graded commutative up to 2-torsion. A striking property of the product is the lack of the associativity. In other words, for x;y;z K (X), 2 (cid:3) (x y) z is not equal to x (y z) in general. We compute this di(cid:11)erence explicitly. (cid:2) (cid:2) (cid:2) (cid:2) Moreover, we show that there is an associative product in the Arakelov K-theory. b In 6, we de(cid:12)ne a direct image morphism in higher arithmetic K-theory. To do this we x employ a higher analytic torsion form of an exact metrized cube de(cid:12)ned by Roessler [14]. Moreover we establish the projection formula. Contents 1. Preliminaries 4 2. Modi(cid:12)ed homotopy groups 13 3. De(cid:12)nition of arithmetic K-groups 18 4. A product formula for higher Bott-Chern forms 25 5. Product 33 6. Direct images 60 Appendix A. Some identities satis(cid:12)ed by binomial coe(cid:14)cients 73 References 76 4 1. Preliminaries 1.1. Conventions on complexes. Let us (cid:12)rst settle some conventions on complexes. For a complex A = (An;d ) of an abelian category A and n Z, the n-th translation A[n] is (cid:3) A (cid:3) 2 de(cid:12)ned as A[n]k = An+k and d = ( 1)nd . For a morphism of complexes u : A B , A[n] A (cid:3) (cid:3) (cid:0) ! the mapping cone Cone(u) is de(cid:12)ned by Cone(u)k = Ak+1 Bk (cid:8) and the di(cid:11)erential d : Ak+1 Bk Ak+2 Bk+1 is de(cid:12)ned by (cid:8) ! (cid:8) d(a;b) = ( d (a);u(a)+d (b)): A B (cid:0) A homological complex is a family An n Z of objects of A with morphisms @ : A A such that @2 = 0. For af hogm2ological complex (A ;@ ), we can de(cid:12)ne aAcompnlex!A nb(cid:0)y1An = A anAd d = @ . The n-th translation of a nhomAological complex (cid:3) n A A A is de(cid:12)ned by A[n] = A(cid:0) and @ = ( 1)n@ . k k n A[n] A (cid:3) (cid:0) (cid:0) 1.2. S-construction. InthissubsectionwerecallS-constructiondevelopedbyWaldhausen [16]. Let [n] be the (cid:12)nite ordered set 0;1; ;n and Ar[n] the category of arrows of [n]. For a small exact category A, let S Afbe th(cid:1)e(cid:1)s(cid:1)et ogf functors n E : Ar[n] A; i j E i;j ! (cid:20) 7! satisfying the following conditions: (1) E = 0 for any 0 i n. i;i (2) For any i j k,(cid:20)E (cid:20) E E is a short exact sequence of A. i;j i;k j;k (cid:20) (cid:20) ! ! For example, S A = 0 , S A is the set of objects of A and S A is the set of short exact 0 1 2 sequences of A. The ffungctor SA : [n] S A n 7! becomes a simplicial set with the base point 0 S A. 0 The set S A can be identi(cid:12)ed with the set of2sequences of injections n E (cid:26) (cid:26) E 0;1 0;n (cid:1)(cid:1)(cid:1) with quotients E E =E for each i < j. By using this identi(cid:12)cation, we can describe i;j 0;j 0;i the boundary maps’and the degeneracy maps of SA as follows: E (cid:26) (cid:26) E ; k = 0; @ (E (cid:26) (cid:26) E ) = 1;2 (cid:1)(cid:1)(cid:1) 1;n k 0;1 (cid:1)(cid:1)(cid:1) 0;n E (cid:26) (cid:26) E (cid:26) E (cid:26) (cid:26) E ; k 1 ( 0;1 0;k 1 0;k+1 0;n (cid:1)(cid:1)(cid:1) (cid:0) (cid:1)(cid:1)(cid:1) (cid:21) and 0 (cid:26) E (cid:26) (cid:26) E ; k = 0; s (E (cid:26) (cid:26) E ) = 0;1 (cid:1)(cid:1)(cid:1) 0;n k 0;1 (cid:1)(cid:1)(cid:1) 0;n (E (cid:26) (cid:26) E (cid:26)id E (cid:26) (cid:26) E ; k 1: 0;1 0;k 0;k 0;n (cid:1)(cid:1)(cid:1) (cid:1)(cid:1)(cid:1) (cid:21) Theorem 1.1. [16, 1.9] There is a homotopy equivalence map between the topological x realizations of simplicial sets SA BQA ; j j ’ j j 5 where BQA is the classifying space of the Quillen’s Q-construction of A [13]. Therefore for n 0, it follows that (cid:21) (cid:25) ( SA ;0) K (A): n+1 n j j ’ 1.3. Exact n-cubes. Let us recall the notion of an exact n-cube. For more details, see [4, 5]. Let < 1;0;1 > be the ordered set consisting of three elements and < 1;0;1 >n its n-th power(cid:0). For a small exact category A, a functor :< 1;0;1 >n A i(cid:0)s called an n-cube of A. Let denote the image of an object (F(cid:11) ; (cid:0);(cid:11) ) of <! 1;0;1 >n. For integers i and j sFa(cid:11)t1is;(cid:1)f(cid:1)(cid:1)y;(cid:11)inng 1 i n and 1 j 1, a1n(cid:1)(cid:1)(n(cid:1) n1)-cube(cid:0)@j is de(cid:12)ned (cid:20) (cid:20) (cid:0) (cid:20) (cid:20) (cid:0) iF by (@j ) = . It is called a face of . For an object (cid:11) of < 1;i0F;1(cid:11)1>;(cid:1)(cid:1)n(cid:1);(cid:11)1n(cid:0)a1nd anF(cid:11)in1;t(cid:1)(cid:1)e(cid:1)g;(cid:11)ei(cid:0)r1i;j;s(cid:11)ai;t(cid:1)(cid:1)i(cid:1)s;(cid:11)fyn(cid:0)in1g 1 i n, a 1-cube @(cid:11)F called an edge of is (cid:0) (cid:0) (cid:20) (cid:20) icF F de(cid:12)ned by : F(cid:11)1;(cid:1)(cid:1)(cid:1);(cid:11)i(cid:0)1;(cid:0)1;(cid:11)i;(cid:1)(cid:1)(cid:1);(cid:11)n(cid:0)1 ! F(cid:11)1;(cid:1)(cid:1)(cid:1);(cid:11)i(cid:0)1;0;(cid:11)i;(cid:1)(cid:1)(cid:1);(cid:11)n(cid:0)1 ! F(cid:11)1;(cid:1)(cid:1)(cid:1);(cid:11)i(cid:0)1;1;(cid:11)i;(cid:1)(cid:1)(cid:1);(cid:11)n(cid:0)1 An n-cube is said to be exact if all edges of are short exact sequences. Let C A denote n F F the set of all exact n-cubes of A. If is an exact n-cube, then any face @j is also exact. F iF Hence @j induces a map i @j : C A C A: i n ! n(cid:0)1 Let be an exact n-cube of A. For an integer i satisfying 1 i n+1, let s1 be an F (cid:20) (cid:20) iF exact (n+1)-cube de(cid:12)ned as follows: 0; (cid:11) = 1; (s1 ) = i iF (cid:11)1;(cid:1)(cid:1)(cid:1);(cid:11)n+1 (F(cid:11)1;(cid:1)(cid:1)(cid:1);(cid:11)i(cid:0)1;(cid:11)i+1;(cid:1)(cid:1)(cid:1);(cid:11)n+1; (cid:11)i 6= 1 and the morphism (s1 ) (s1 ) is the identity of .iFIn (cid:11)a1d;(cid:1)d(cid:1)(cid:1)i;(cid:11)tii(cid:0)o1n;(cid:0), 1le;(cid:11)ti+s1;(cid:1)(cid:1)1(cid:1);(cid:11)n+b1e!an exiFact(cid:11)1(;n(cid:1)(cid:1)(cid:1)+;(cid:11)i(cid:0)11);-0c;(cid:11)uib+1e;(cid:1)(cid:1)d(cid:1);e(cid:11)(cid:12)n+n1ed as follows: F(cid:11)1;(cid:1)(cid:1)(cid:1);(cid:11)i(cid:0)1;(cid:11)i+1;(cid:1)(cid:1)(cid:1);(cid:11)n+1 i(cid:0) F 0; (cid:11) = 1; (s 1 ) = i (cid:0) (cid:0)i F (cid:11)1;(cid:1)(cid:1)(cid:1);(cid:11)n+1 (F(cid:11)1;(cid:1)(cid:1)(cid:1);(cid:11)i(cid:0)1;(cid:11)i+1;(cid:1)(cid:1)(cid:1);(cid:11)n+1; (cid:11)i 6= (cid:0)1 and the morphism (s 1 ) (s 1 ) is the identity (cid:0)i F (cid:11)1;(cid:1)(cid:1)(cid:1);(cid:11)i(cid:0)1;0;(cid:11)i+1;(cid:1)(cid:1)(cid:1);(cid:11)n+1 ! (cid:0)i F (cid:11)1;(cid:1)(cid:1)(cid:1);(cid:11)i(cid:0)1;1;(cid:11)i+1;(cid:1)(cid:1)(cid:1);(cid:11)n+1 of . Then the map F(cid:11)1;(cid:1)(cid:1)(cid:1);(cid:11)i(cid:0)1;(cid:11)i+1;(cid:1)(cid:1)(cid:1);(cid:11)n+1 sj : C A C A i n ! n+1 is also de(cid:12)ned. An exact cube written as sj is said to be degenerate. Let ZC A be the free abelian group geneirFated by C A. Let D ZC A be the subgroup n n n n (cid:26) generated by all degenerate exact n-cubes. Let ZC A = ZC A=D and n n n n 1 @ = ( 1)i+j+1@je: ZC A ZC A: (cid:0) i n ! n(cid:0)1 i=1j= 1 XX(cid:0) e e Then ZC A = (ZC A;@) becomes a homological complex. n We can(cid:3)construct an exact (n 1)-cube Cub(E) associated with an element E S A in n an indeuctive waey. In the case o(cid:0)f n = 1, let us de(cid:12)ne Cub(E) = E for E S A2. When 1 2 6 an exact (m 1)-cube is associated with any element of S A for m < n, an (n 1)-cube m Cub(E) assoc(cid:0)iated with E S A is de(cid:12)ned as (cid:0) n 2 @ 1Cub(E) = s1 s1(E ); 1(cid:0) n(cid:0)2(cid:1)(cid:1)(cid:1) 1 0;1 @0Cub(E) = Cub(@ E); 1 1 @1Cub(E) = Cub(@ E): 1 0 Then Cub : S A C A induces a homomorphism of complexes n n 1 ! (cid:0) Cub : ZS A[1] ZC A: (cid:3) ! (cid:3) Remark: The di(cid:11)erential of ZC A de(cid:12)ned aboveeis the minus of the one de(cid:12)ned in [5]. But since the di(cid:11)erential ofZS A[1](cid:3)is alsothe minus ofthe one ofZS A, the homomorphism (cid:3) (cid:3) Cub is compatible with the di(cid:11)eerentials of complexes. 1.4. Deligne cohomology. Let M be a complex algebraic manifold of dimension n, that is, let M be the analytic space consisting of all C-valued points of a smooth algebraic variety of dimension n over C. Then higher Chern character map from higher algebraic K-theory to a reasonable cohomology theory ch : K (M) H2j i(M;(cid:0)(j)) i i (cid:0) ! (cid:8)j has been developed in [8]. In the case of i = 0, it is given by the Chern characters of vector bundles. But when i > 0 and M is compact, ch with values in the singular cohomology is i known to be trivial. In order to obtain a nontrivial higher Chern character map, we should consider Deligne cohomology. Fromnow onwe assume thatM is compact. Let (cid:10) bethe complex ofanalytic (cid:3)M sheaves of holomorphic di(cid:11)erential forms and let Fp(cid:10) be the de Rham (cid:12)ltration of (cid:10) , (cid:3)M (cid:3)M that is, Fp(cid:10) = (0 (cid:10)p (cid:10)p+1 (cid:10)n ): (cid:3)M ! M ! M ! (cid:1)(cid:1)(cid:1) ! M The real Deligne complex R(p)D on M is a complex of sheaves de(cid:12)ned as follows: Cone(R(p)(cid:8)Fp(cid:10)(cid:3)M (cid:0)"(cid:0)!(cid:19) (cid:10)(cid:3)M)[(cid:0)1]; where " and (cid:19) are natural inclusions. It is obvious that R(p)D is quasi-isomorphic to the complex 0 ! R(p) ! OM ! (cid:10)1M ! (cid:1)(cid:1)(cid:1) ! (cid:10)pM(cid:0)1 ! 0; where R(p) is of degree zero. The real Deligne cohomology Hi (M;R(p)) is de(cid:12)ned as the D hypercohomology of R(p), that is, HDi (M;R(p)) = Hi(M;R(p)D): In fact, Deligne cohomology can be de(cid:12)ned when M is neither smooth nor compact [7]. Moreover, it can be a target of a nontrivial higher Chern character map, that is, there is a homomorphism chi : Ki(M) HD2p(cid:0)i(M;R(p)); ! (cid:8)p which is far from trivial. 7 Let Ep(M) be the space of real smooth di(cid:11)erential forms of degree p on M and Ep(M) = R Ep(M) C. Let Ep;q(M) be the space of complex di(cid:11)erential forms of type (p;q) on M. We R (cid:10)R set En 1(M)(p 1) Ep0;q0(M); n < 2p; R(cid:0) (cid:0) \p0+q(cid:8)0=n 1 (cid:0) Dn(M;p) = 8 p0<p;q0<p >>><E2Rp(M)(p)\Ep;p(M)\Kerd; n = 2p; 0; n > 2p and de(cid:12)ne a di(cid:11)erential dD :>>>Dn(M;p) Dn+1(M;p) by : ! (cid:25)(d!); n < 2p 1; (cid:0) (cid:0) dD(!) = 8 2@@!; n = 2p 1; (cid:0) (cid:0) ><0; n > 2p 1; (cid:0) where (cid:25) : En(M) Dn(M;p) is the canonical projection. > ! : Theorem 1.2. [3, Thm.2.6] Let M be a compact complex algebraic manifold. Then (D(cid:3)(M;p);dD) becomes a complex of R-vector spaces and there is a canonical isomorphism Hn(D(cid:3)(M;p);dD) HDn(M;R(p)) ’ if n 2p. (cid:20) 1.5. Higher Bott-Chern forms. Inthissubsection werecallthehigherBott-Chernforms developed by Burgos and Wang. For more details, the reader should consult the original paper [5] or a survey [4]. A hermitian vector bundle E = (E;h) on a complex algebraic manifold M is an algebraic vector bundle E on M with a smooth hermitian metric h. On a hermitian vector bundle E, there is a unique connection that is compatible with both rE the metric and the complex structure. Let K denote the curvature form of . The Chern E rE form of E is de(cid:12)ned as ch (E) = Tr(exp( K )) D2p(M;p): 0 (cid:0) E 2 (cid:8)p Hereafter we assume that M is compact. An exact metrized n-cube on M is an exact n-cube made of hermitian vector bundles on M. Let = E be an exact metrized n-cube (cid:11) F f g on M. For an n-tuple (cid:11) = ((cid:11) ; ;(cid:11) ) with 1 (cid:11) 1 and an integer i satisfying (cid:11) = 1, 1 n k i (cid:1)(cid:1)(cid:1) (cid:0) (cid:20) (cid:20) there is a surjection E E . is called an emi-n-cube if the metric (cid:11)1;(cid:1)(cid:1)(cid:1);(cid:11)i(cid:0)1;0;(cid:11)i+1;(cid:1)(cid:1)(cid:1);(cid:11)n ! (cid:11) F on any E with (cid:11) = 1 coincides with the metric induced from E by the (cid:11) i (cid:11)1;(cid:1)(cid:1)(cid:1);(cid:11)i(cid:0)1;0;(cid:11)i+1;(cid:1)(cid:1)(cid:1);(cid:11)n above surjection. Let (x : y) be the homogeneous coordinate of P1. Let (1) be the tautological line bundle O on P1 with the Fubini-Study metric. For a hermitian vector bundle E on M, let E(1) be the hermitian vector bundle on M P1 given by (cid:25) E (cid:25) (1). Let (cid:27) ;(cid:27) H0(P1; (1)) be (cid:2) 1(cid:3) (cid:10) 2(cid:3)O x y 2 O global sections of (1) determined by x and y respectively. O For an emi-1-cube E : E E E , a map : E E (1) E (1) on M 1 0 1 1 0 1 P1 is de(cid:12)ned by e ((cid:19)(e)(cid:0) !(cid:27) ;e !(cid:27) ), where (cid:19) is the(cid:0)inj!ection E (cid:8) (cid:0) E . The 1(cid:2)- x y 1 0 transgression bundle7!tr (E) of(cid:10)Eis a h(cid:10)ermitian vector bundle on M P1 gi(cid:0)ven!by the cokernel 1 (cid:2) of with the induced metric. It follows from the de(cid:12)nition that tr (E) E and 1 M x=0 0 j (cid:2)f g ’ 8 tr (E) E E . Hence tr (E) can be viewed as a family of hermitian vector 1 M y=0 1 1 1 bundlejs (cid:2)onf Mgp’aram(cid:0)et(cid:8)rized by P1 connecting E with E E . 0 1 1 (cid:0) (cid:8) The n-transgression bundle of an emi-n-cube is de(cid:12)ned by iterating the above process. The 1-transgression tr ( ) of an emi-n-cube is an emi-(n 1)-cube on M P1 de(cid:12)ned as 1 F F (cid:0) (cid:2) tr ( ) = tr (@(cid:11) ( )) 1 F (cid:11) 1 nc F for (cid:11) < 1;0;1 >n 1. The n-transgression bundle of is de(cid:12)ned as (cid:0) 2 (cid:0) F ntimes tr ( ) = tr tr tr ( ); n 1 1 1 F (cid:1)(cid:1)(cid:1) F which is a hermitian vector bundle on M z(P1)}n|. { Let z = x=y be the Euclidean coordina(cid:2)te of P1 and z the i-th Euclidean coordinate of i (P1)n. For an integer i satisfying 1 i n, a di(cid:11)erential form with logarithmic poles Si on (P1)n is de(cid:12)ned as (cid:20) (cid:20) n dz dz dz(cid:22) dz(cid:22) Si = ( 1)(cid:27)log z 2 (cid:27)(2) (cid:27)(i) (cid:27)(i+1) (cid:27)(n): n (cid:0) j (cid:27)(1)j z ^(cid:1)(cid:1)(cid:1)^ z ^ z(cid:22) ^(cid:1)(cid:1)(cid:1)^ z(cid:22) (cid:27)X2Sn (cid:27)(2) (cid:27)(i) (cid:27)(i+1) (cid:27)(n) Let us de(cid:12)ne the Bott-Chern form of an emi-n-cube as F 1 ch ( ) = ch (tr ( )) T D2p n(M;p); n 0 n n (cid:0) F (2(cid:25)p(cid:0)1)n Z(P1)n F ^ 2 (cid:8)p where ( 1)n n T = (cid:0) ( 1)iSi: n 2n! (cid:0) n i=1 X Proposition 1.3. Let be an emi-n-cube on M. Then F ch (sj ) = 0 n+1 iF for 1 i n+1 and j = 1. (cid:20) (cid:20) (cid:6) Proof: Let us (cid:12)rst consider the case of n = 0. For a hermitian vector bundle E on M, it follows that s 1E = 0 E id E . Hence the 1-transgression bundle tr(s 1E) is isometric (cid:0)1 ! ! (cid:0)1 to E(1). If r : P1 (cid:16)P1 is an involu(cid:17)tion given by r (z) = z 1, then r (E(1)) = E(1). Hence (cid:3) (cid:0) (cid:3) ! we have 1 ch (s 1E) = ch (E(1))log z 2 1 (cid:0)1 2(cid:25)p 1 P1 0 j j (cid:0) Z 1 = r ch (E(1))log z 2 (cid:3) 0 2(cid:25)p 1 P1 j j (cid:0) Z 1 (cid:0) (cid:1) = (cid:0) ch (E(1))log z 2 0 2(cid:25)p 1 P1 j j (cid:0) Z = ch (s 1E); (cid:0) 1 (cid:0)1 therefore ch (s 1E) = 0. In the same way we can show ch (s1E) = 0. 1 (cid:0)1 1 1 9 Let us move on to the general case. For 1 i n+1, let r : (P1)n+1 (P1)n+1 denote i (cid:20) (cid:20) ! an involution given by z ; i = j; j r z = 6 i(cid:3) j z 1; i = j: ( j(cid:0) Then we have r (tr (sj )) = tr (sj ) and r T = T . Therefore we can show i(cid:3) n+1 iF n+1 iF i(cid:3) n+1 (cid:0) n+1 ch (sj ) = 0 in the same way as above. (cid:3) n+1 iF Let us extend the de(cid:12)nition of the Bott-Chern form to an arbitrary exact metrized n- cube. Let be an exact metrized n-cube, not necessarily emi. For an integer i satisfying F 1 i n, (cid:21)1 is de(cid:12)ned as (cid:20) (cid:20) iF E ; (cid:11) = 1 or 0; ((cid:21)1 ) = (cid:11) i (cid:0) iF (cid:11) E ; (cid:11) = 1; ( 0(cid:11) i whereE isthesamevectorbundleasE withthemetricinducedfromE . Let (cid:21)2 0(cid:11)be an exact metrized n-cube de(cid:11)(cid:12)ned as (cid:11)1;(cid:1)(cid:1)(cid:1);(cid:11)i(cid:0)1;0;(cid:11)i+1;(cid:1)(cid:1)(cid:1);(cid:11)n iF E ; (cid:11) = 1; (cid:11)1;(cid:1)(cid:1)(cid:1);(cid:11)i(cid:0)1;1;(cid:11)i+1;(cid:1)(cid:1)(cid:1);(cid:11)n i (cid:0) ((cid:21)2 ) = E ; (cid:11) = 0; iF (cid:11) 8 0(cid:11)1;(cid:1)(cid:1)(cid:1);(cid:11)i(cid:0)1;1;(cid:11)i+1;(cid:1)(cid:1)(cid:1);(cid:11)n i ><0; (cid:11) = 1 i (cid:21)an1d the(cid:21)m2 or.pAhinsmemEi-(cid:11)n1-;(cid:1)c(cid:1)(cid:1)u;(cid:11)bi(cid:0)e1;(cid:21)1;(cid:11)i+i1s;>:(cid:1)(cid:1)d(cid:1);e(cid:11)(cid:12)nn!edEas0(cid:11)1;(cid:1)(cid:1)(cid:1);(cid:11)i(cid:0)1;1;(cid:11)i+1;(cid:1)(cid:1)(cid:1);(cid:11)n is the identity. We set (cid:21)iF = iF (cid:8) iF F (cid:21) (cid:21) (cid:21) ; n 1; n n 1 1 (cid:21) = (cid:0) (cid:1)(cid:1)(cid:1) F (cid:21) F ; n = 0: ( F De(cid:12)nition 1.4. The Bott-Chern form of an exact metrized n-cube is an element of D2p n(M;p) de(cid:12)ned as F (cid:0) (cid:8)p 1 ch ( ) = ch (tr ((cid:21) )) T : n 0 n n F (2(cid:25)p(cid:0)1)n Z(P1)n F ^ Remark: For an emi-n-cube , the Bott-Chern form ch ( ) de(cid:12)ned in Def.1.4 is the n F F same form as the one we have de(cid:12)ned before, because there is a decomposition (cid:21) = (a degenerate emi-n-cube): F F (cid:8) ForacompactcomplexalgebraicmanifoldM,letP(M)bethecategoryofhermitianvector bundles on M and S(M) the S-construction of P(M). We note that S(M) is homotopy equivalent to the S-construction of vector bundles bon M by the map forgetting metrics. In particular, it followsbthat b b (cid:25) ( S(M) ) K (M): n+1 n j j ’ b 10 LetZC (M) = ZC P(M)andZCemi(M)thesubcomplex ofZC (M)generatedbyemi-cubes on M. (cid:3)Then (cid:3)(cid:21) induces a h(cid:3)omomorphism of complexes (cid:3) F 7! F e b e b e b e b (cid:21) : ZC (M) ZCemi(M): (cid:3) ! (cid:3) e b e b Theorem 1.5. [5] If is an exact metrized n-cube on M, then we have F dDchn( ) = chn 1(@F): F (cid:0) Hence the higher Bott-Chern forms induce a homomorphism of complexes ch : ZC (M) D (M;p)[2p]: (cid:3) (cid:3) ! (cid:8)p Moreover, the following map e b Kn(M) = (cid:25)n+1(S(M)) Hurewicz Hn+1(S(M)) Cub Hn(ZC (M)) ch HD2p(cid:0)n(M;p) (cid:0)! (cid:0)! (cid:3) ! (cid:8)p agrees with the higherbChern character withbvalues in the Deebligne cohomology. This is the main theorem of [5]. Here let us prove dDchn( ) = chn 1(@ ) in a di(cid:11)erent way from [5]. To do this we introduce another description ofFthe logari(cid:0)thmFic forms Si. n For integers (cid:11) ; ;(cid:11) with 1 (cid:11) n, a k-form with logarithmic poles ((cid:11) ; ;(cid:11) ) on 1 k i 1 k (P1)n is given as (cid:1)(cid:1)(cid:1) (cid:20) (cid:20) (cid:1)(cid:1)(cid:1) ((cid:11) ; ;(cid:11) ) = dlog z 2 dlog z 2: 1 (cid:1)(cid:1)(cid:1) k j (cid:11)1j ^(cid:1)(cid:1)(cid:1)^ j (cid:11)kj Let ((cid:11) ; ;(cid:11) )(i;k i) denote the (i;k i)-part of ((cid:11) ; ;(cid:11) ). Then we have 1 k (cid:0) 1 k (cid:1)(cid:1)(cid:1) (cid:0) (cid:1)(cid:1)(cid:1) n Si = (i 1)!(n i)! ( 1)(cid:11)+1log z 2(1; ;(cid:11); ;n)(i 1;n i): n (cid:0) (cid:0) (cid:0) j (cid:11)j (cid:1)(cid:1)(cid:1) (cid:1)(cid:1)(cid:1) (cid:0) (cid:0) (cid:11)=1 X b Lemma 1.6. It follows that n @Si = i!(n i)!(1 n)(i;n i) +(n i) ( 1)(cid:11)@@log z 2Si ; n (cid:0) (cid:1)(cid:1)(cid:1) (cid:0) (cid:0) (cid:0) j (cid:11)j n 1;(cid:11)b (cid:0) (cid:11)=1 X n @Si = (i 1)!(n i+1)!(1 n)(i 1;n i+1) (i 1) ( 1)(cid:11)@@log z 2Si 1 ; n (cid:0) (cid:0) (cid:1)(cid:1)(cid:1) (cid:0) (cid:0) (cid:0) (cid:0) (cid:0) j (cid:11)j n(cid:0)1;(cid:11)b (cid:0) (cid:11)=1 X where Si is the logarithmic form on (P1)n with the same expression as Si for the n 1;(cid:11)b n 1 coordinate(cid:0)(z ; ;z ;z ; ;z ). (cid:0) 1 (cid:11) 1 (cid:11)+1 n (cid:1)(cid:1)(cid:1) (cid:0) (cid:1)(cid:1)(cid:1) We will prove more general identities in Lem.4.3. Let us return to the proof of the identity dDchn( ) = chn 1(@ ). Since (cid:21) : ZC (M) ZCemi(M) is a homomorphism of complexes, F (cid:0) F (cid:3) ! (cid:3) e b e b

Description:
Introduction. The aim of this In this paper, we will give another definition of higher arithmetic K-theory of a proper arithmetic variety by point is expressed by n-tuple of real numbers (t1, ททท ,tn) with 0 < ti < 1. Let T be a pointed
See more

The list of books you might like

Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.