Comparison of secondary invariants of algebraic K-theory

In this paper we prove that the multiplicative character of A. Connes and M. Karoubi and the determinant invariant of L. G. Brown, J. W. Helton and R. E. Howe agree up to a canonical homomorphism.

Authors: Jens Kaad (University of Copenhagen)

COMP ARISON OF SECOND AR Y INV ARIANTS OF A LGEB RAIC K -THEOR Y J. KAAD Departmen t of Mathematical Sciences, Univ ers it y of Cop enhagen Univ ersitets park en 5, DK-2100 Cop enhagen, Denmark Abstract In this pap er w e prov e that the multiplic ativ e character of A. Connes and M. K a roubi and the determinan t in v arian t of L. G. Bro wn, J. W. Helton and R. E. How e agree up to a canonical homomorphism. Contents 1. In t r o du ction 2 2. The determinan t inv arian t of an extension b y L 1 ( H ) 4 2.1. The ˇ Cec h complex of a surjectiv e simplicial map 4 2.2. The determinan t in v ariant 5 3. The multipl icativ e c haracter o f a finitely summable F redholm mo dule 6 3.1. The relative Chern character 6 3.2. The multiplic ativ e c haracter 8 4. Comparing the determinan t in v ariant and the m ultiplicativ e c haracter 9 4.1. F actorization through relativ e con tinuous cyclic homology 9 4.2. Determinan ts and the relativ e logarithm 11 4.3. Pro of of the main result 14 References 17 email: kaad@ math.k u.dk . 1 2 J. KAAD 1. Int r oduction In their pap er, [9], A. Connes and M. Karoubi define a m ultiplicativ e ch aracter on algebraic K -theory M F : K 2 p ( A ) → C / (2 π i ) p Z for eac h o dd unital 2 p -summable F redholm mo dule ( F , H ) o v er the unital C -algebra A . The construction uses the re lativ e K -groups of a un ital Banac h algebra and the re lativ e Chern c ha r a cter with v alues in con tin uous cyclic homology . In a different direction, L.G. Brow n has defined a determinan t inv ariant d ( X,ι ) : K 2 ( B ) → C ∗ for eac h exact seq uence of C -a lgebras X : 0 − − − → L 1 ( H ) i − − − → E π − − − → B − − − → 0 equipped with a unital algebra homomorphism ι : E → L ( H ) suc h that (1) ( ι ◦ i )( T ) = T for all T ∈ L 1 ( H ) The construction uses the F redholm determinan t homomorphism det : G → C ∗ e T 7→ e T r ( T ) Here G ⊆ GL ( L ( H )) denotes the op erators of determinan t class. See [3, 4]. In the case where B is commu tativ e, the determinan t in v arian t is related to the work of J. W. Helton and R. E. How e o n traces of comm utators, [12], via the iden tit y d ( X,ι ) { π ( e S ) , π ( e T ) } = e T r [ S,T ] S, T ∈ E Here { π ( e S ) , π ( e T ) } ∈ K 2 ( B ) denotes the Steinberg sym b ol. Notice also the app earance of the determinan t in v ariant in the pap ers of R. W. Carey and J. D. Pincus. See [5, 6]. The purp ose of this pap er is to sho w that the multipli cativ e c haracter and the determinan t in v arian t coincide up to a canonical homomorphism. In particular, to eac h o dd unital 2 - summable F redholm mo dule ( F , H ) o v er a unital C -algebra A w e a sso ciate a n extens ion o f a certain C - algebra B by the op erato r s of trace class (2) X F : 0 − − − → L 1 ( H ) i − − − → E π − − − → B − − − → 0 an injectiv e homomorphism ι : E → L ( H ) satisfying (1 ) , and a surjectiv e homomorphism R F : A → B The main result of the pap er can then b e expresse d as the commu tativit y of the diagram (3) K 2 ( A ) K 2 ( B ) C / (2 π i ) Z C ∗ ✲ ( R F ) ∗ ❄ M F ❄ d ( X F ,ι ) ✲ exp It follows that the m ultiplicativ e c haracter admits calculation in t he commu tativ e case, and that it factorizes through the second algebraic K -gro up of the Calcin a lg ebra C 1 ( H ) = L ( H ) / L 1 ( H ) . COMP ARISON OF SECONDAR Y INV ARI ANTS OF A LGEBRAIC K -THEOR Y 3 Using the theory of cen tral extensions w e can state the main result in a differen t fashion. In [9, Paragraph 5 ] it is show n that the univ ersal m ultiplicativ e c haracter M U : K 2 ( M 1 ) ∼ = H 2 ( E ( M 1 )) → C / (2 π i ) Z ∼ = C ∗ is induced by a cen tral extension (4) ϕ : 1 − − − → C ∗ − − − → Γ − − − → E ( M 1 ) − − − → 1 On the other hand, the univ ersal determinan t inv aria n t d : K 2 ( E ( C 1 ( H ))) ∼ = H 2 ( E ( C 1 ( H ))) → C ∗ could b e understo o d as the ho momorphism induc ed by the cen tra l extensi on (5) ψ : 1 − − − → C ∗ − − − → e Γ E ( q ) − − − → E ( C 1 ( H )) − − − → 1 where e Γ = GL ( L ( H )) / G 0 with G 0 defined as the normal subgroup consisting of op erators of determinan t one. The main theorem then translates in to the follo wing statemen t: The elemen ts in group cohomology with co efficien ts in C ∗ determined b y the cen tral extensions (4) and (5 ) coincide up to a homomorphism E ( R ) ∗ ( ψ ) = ϕ E ( R ) ∗ : H 2 ( E ( C 1 ( H )) , C ∗ ) → H 2 ( E ( M 1 ) , C ∗ ) W e th us obtain refinemen ts o f results giv en in [27]. Here w e hav e applied the univ ersal co efficien t Theorem together with the p erfectness of the elemen tary matrices. The presen t paper is organized as follows . In the first tw o sections w e recall the construction of the determinan t inv ariant and the mul- tiplicativ e c haracter. Our definition of the determinan t in v ariant differs sligh tly f r o m the one giv en in [3, 4 ]. W e use a description of the relativ e homology groups of the surjectiv e homo- morphism E ( q ) : E ( L ( H )) → E ( C 1 ( H )) in terms of an algebraic analogue of t he ˇ Cec h complex of an op en co v er of a top olo gical space. See [13 , 17]. A com binatorial argumen t can b e giv en, pro ving that the t w o definitions yield the same map. The alternativ e a pproa c h seems to b e essen t ia l f or the working of our main pro of . In the last section we giv e a pro of of the main result. There are tw o main ingredien ts. The first ingredien t is a factorization result for the homomorphism τ 1 : H C 1 ( M 1 ) → C ( x 0 , x 1 ) 7→ − 1 4 T r ( F U [ F U , x 0 ][ F U , x 1 ]) through a relativ e contin uous cyclic homology g r o up. This result is essen tially con tained in the pro of of [9, Theorem 5 . 6 ]. The second ingredien t is a n application of the (well kno wn) description of the F redholm determinan t using a logarithm. T o b e precise, for each smo oth map σ : [0 , 1] → G σ (0) = 1 w e can calculate the F redholm determinan t of the endp oin t σ (1) ∈ G , using the in tegral T r ( Z 1 0 dσ dt · σ − 1 dt ) ∈ C 4 J. KAAD W e use this observ ation in a relativ e setting, in v o king the algebraic ˇ Cec h complex a second time. These considerations lead directly to a pro of of the main result. W e would lik e to end this introduction b y mentionin g that ev en though the determinan t in- v arian t has no ob vious generalization to the higher K - groups, w e are still a ble to calculate the m ultiplicativ e c haracter in this setting, o bta ining higher dimensional analogues of the iden tity (exp ◦M F ) { e T , e S } = exp( T r [ P T P , P S P ]) This will b e the sub ject of a forthcoming pap er. A c kno wledgemen ts: I would lik e to thank Ryszard Nest fo r his con tin uous supp ort and man y helpful commen ts. 2. The determinant inv ariant of an extension by L 1 ( H ) The main sub ject o f this section is the definition of the determinan t in v arian t. Our definition relies on conside rations in relativ e homology , whic h will find application throughout the pap er. W e will therefore start by a form ulation of these ideas in a general contex t b efore passing on to t he definition of the determinan t in v ar ia n t. 2.1. The ˇ Cec h complex of a surjectiv e simplicial map. Let X and Y b e simplicial sets and sup p ose that w e hav e a surje ctive simplicial map f : X → Y . W e let Ker ∗ ( f ) denote the k ernel ch ain complex of the asso ciated c hain map f : Z [ X ∗ ] → Z [ Y ∗ ] . F urthermore w e let ˇ C ( f ) denote the bicomplex with ˇ C nm ( f ) = Z [ X n × Y n . . . × Y n X n | {z } m +2 ] The generators are th us ( m + 2) -tuples of elemen ts in X n , ( x 1 , . . . , x m +2 ) , whic h coincide in Y n , f ( x 1 ) = . . . = f ( x m +2 ) . The differen tials are give n b y d : ˇ C nm ( f ) → ˇ C ( n − 1) m ( f ) d ( x 1 , . . . , x m +2 ) = P n i =0 ( − 1) i ( d i ( x 1 ) , . . . , d i ( x m +2 )) and δ : ˇ C nm ( f ) → ˇ C n ( m − 1) ( f ) δ ( x 1 , . . . , x m +2 ) = P m +2 j =1 ( − 1) j +1 ( x 1 , . . . , ˇ x j , . . . , x m +2 ) Here the ˇ · signifies t ha t the term has to b e omitted. W e then hav e a long exact sequen ce of ab elian groups Z [ X n ] Z [ Y n ] ˇ C n 0 ( f ) ˇ C n 1 ( f ) ˇ C n 2 ( f ) . . . ✛ f ✛ ε ✛ δ ✛ δ ✛ δ for eac h n ∈ N 0 . Here the ch ain map ε : ˇ C ∗ 0 ( f ) → Z [ Y ∗ ] is giv en b y ε : ( x 1 , x 2 ) 7→ x 1 − x 2 . In particular, letting Cok er ∗ ( δ f ) denote the cok ernel c hain complex of the chain map δ : ˇ C ∗ 1 ( f ) → ˇ C ∗ 0 ( f ) w e ha ve the result COMP ARISON OF SECONDAR Y INV ARI ANTS OF A LGEBRAIC K -THEOR Y 5 Theorem 2.1. [17] F or e ach n ∈ N 0 we ha v e the isomorphisms H n ( f ) ∼ = H n ( Ker ( f )) ∼ = H n ( Cok er ( δ f )) ∼ = H n ( ˇ C ( f )) Her e H n ( f ) denotes the r elative homolo gy o f the simpli c ial map f : X → Y 2.2. The determinan t in v arian t. Let C 1 ( H ) = L ( H ) / L 1 ( H ) denote the Calkin a lgebra. W e then hav e the quotien t map q : L ( H ) → C 1 ( H ) and an induced surjectiv e group homomorphism E ( q ) : E ( L ( H )) → E ( C 1 ( H )) . Here E ( A ) denotes the elemen ta ry matrices of a unital ring A . W e define the determinan t homomorphism on the ab elian group ˇ C 10 ( E ( q )) b y det : Z [ E ( L ( H )) × E ( C 1 ( H )) E ( L ( H ))] → C ∗ ( g 1 , g 2 ) 7→ det ( g 1 g − 1 2 ) Note that g 1 g − 1 2 is of determinan t class since E ( q )( g 1 ) = E ( q )( g 2 ) . Lemma 2.2. T he determin ant det : Z [ E ( L ( H )) × E ( C 1 ( H )) E ( L ( H ))] → C ∗ induc es a map on ho molo gy det : H 1 ( Cok er ( δ E ( q ) )) → C ∗ Pr o of. This is a matter of c hec king the equalit y det ( g 1 g − 1 2 ) · de t ( g 2 g − 1 3 ) = det ( g 1 g − 1 3 ) for E ( q )( g 1 ) = E ( q )( g 2 ) = E ( q )( g 3 ) and the equalit y det ( g 1 g − 1 2 ) det ( h 1 h − 1 2 ) = det ( g 1 h 1 h − 1 2 g − 1 2 ) for E ( q )( g 1 ) = E ( q )( g 2 ) and E ( q )( h 1 ) = E ( q )( h 2 ) . But they follow easily from w ell kno wn prop erties of the F redholm determinan t, [28, 31].  By Theorem 2.1, the surjectivit y o f E ( q ) : E ( L ( H )) → E ( C 1 ( H )) implies that ε : H 1 ( Cok er ( δ E ( q ) )) → H 1 ( E ( q )) is an isomorphi sm. F urthermore, by [1 8, Prop osition 1 . 2 . 1 ] w e can identify the ab elian groups K 2 ( C 1 ( H )) a nd H 2 ( E ( C 1 ( H ))) . W e can thus define the univ ersal determinan t in v ariant as t he comp osition d = det ◦ ∂ : K 2 ( C 1 ( H )) ∼ = H 2 ( E ( C 1 ( H ))) → C ∗ Here ∂ : H 2 ( E ( C 1 ( H ))) → H 1 ( E ( q )) is the b oundary map a sso ciated with the group homomor- phism E ( q ) : E ( L ( H )) → E ( C 1 ( H )) . No w, suppo se that w e hav e an exact sequence of C -algebras X : 0 − − − → L 1 ( H ) i − − − → E π − − − → B − − − → 1 equipped with a unital algebra homomorphism ι : E → L ( H ) suc h that ( ι ◦ i )( T ) = T for all T ∈ L 1 ( H ) W e then ha v e an induced homomorphism ι : B → C 1 ( H ) whic h b y functorialit y of algebraic K -theory yields a homomorphism ι ∗ : K 2 ( B ) → K 2 ( C 1 ( H )) 6 J. KAAD W e define the determinan t in v ar ia nt of the pair ( X , ι ) as the comp osition d ( X,ι ) = d ◦ ι ∗ : K 2 ( B ) → C ∗ 3. The mul tiplica tive cha ra cter of a finitel y summable Fredholm module In this section w e recall the construction of the m ultiplicativ e c ha ra cter asso ciated with an o dd, finitely summable F redholm mo dule. T o this end w e will describe the relativ e Chern c ha r a cter and the relativ e K -groups in detail. The references for this section are [9, 16] and [2 9]. Note that the cyclic homology and Ho c hsc hild homolog y groups encounte red in this section, and throughout the pap er a r e the non-Hausdorff ve rsions of c on tinuous cyclic homology a nd c ontinuous Ho c hsc hild homology groups. 3.1. The relativ e Chern c haracter. Let A b e a unital Banac h algebra. W e let ∆ n = { t ∈ [0 , 1] n | n X i =1 t i ≤ 1 } denote the standard n -simplex. The vertic es will b e denoted by 0 , . . . , n ∈ ∆ n . Let R ( A ) denote the simplicial set with R ( A ) n = { σ ∈ GL ( C ∞ (∆ n , A )) | σ ( 0 ) = 1 } in degree n ∈ N 0 and with face op erators and degeneracy o p erators defined by d i ( σ )( t 1 , . . . , t n − 1 ) =  σ (1 − P n − 1 j =1 t j , t 1 , . . . , t n − 1 ) · σ ( 1 ) − 1 for j = 0 σ ( t 1 , . . . , t i − 1 , 0 , t i , . . . , t n − 1 ) for j ∈ { 1 , . . . , n } s j ( σ )( t 1 , . . . , t n +1 ) =  σ ( t 2 , . . . , t n +1 ) for j = 0 σ ( t 1 , . . . , t i − 1 , t i + t i +1 , . . . , t n +1 ) for j ∈ { 1 , . . . , n } Remark the extra f a ctor σ ( 1 ) − 1 in the express ion f or d 0 . Let | R ( A ) | denote the geometric realization of the simplicial set R ( A ) . The fundamen tal group is then given by π 1 ( | R ( A ) | ) = R ( A ) 1 / ∼ where ∼ denotes the equiv alence relation given by smo oth homotopies with fixed endp oin ts. See [22]. Since the commutator subgroup is p erfect and normal, w e can apply the plus construction of D. Quillen, obtaining the p oin ted C W -complex | R ( A ) | + . See [24]. Definition 3.1. [16] By the relativ e K - theory of A we wil l understand the homotopy gr oups of | R ( A ) | + thus b y de fi nition K rel n ( A ) = π n ( | R ( A ) | + ) for e ach n ∈ N 0 . COMP ARISON OF SECONDAR Y INV ARI ANTS OF A LGEBRAIC K -THEOR Y 7 By [16] the relative K -g ro ups fit in a long exact sequence (6) . . . K top n +1 ( A ) K rel n ( A ) K n ( A ) . . . K n − 1 ( A ) K rel n − 1 ( A ) K top n ( A ) ✲ i ✲ v ✲ θ ❄ i ✛ i ✛ θ ✛ v whic h terminates at K top 1 ( A ) . Here θ : K rel n ( A ) → K n ( A ) is induced b y the simplicial map θ : R n ( A ) → GL ( A ) n θ ( σ ) = ( σ ( 0 ) σ ( 1 ) − 1 , . . . , σ ( n − 1 ) σ ( n ) − 1 ) F urthermore, in [9] it is show n that the long exact sequence (6) is related to t he con tin uous S B I -sequence in homology by means of Chern c haracters . . . K top n +1 ( A ) K rel n ( A ) K n ( A ) K top n ( A ) . . . . . . H C n +1 ( A ) H C n − 1 ( A ) H H n ( A ) H C n ( A ) . . . ✲ i ✲ v ❄ c h top n +1 ✲ θ ❄ ( − 1) n ( n − 1)! c h rel n ✲ i ❄ 1 n ! D n ✲ v ❄ c h top n ✲ I ✲ S ✲ 1 n B ✲ I ✲ S W e will giv e a precise description of the r elative Chern c haracter c h rel n : K rel n ( A ) → H C n − 1 ( A ) as the comp osition of three homomorphisms. The first one is the Hurewicz homomorphism h n : K rel n ( A ) → H n ( | R ( A ) | + ) ∼ = H n ( R ( A )) The second one is the loga rithm L : H n ( R ( A )) → lim m →∞ H C n − 1 ( M m ( A )) whic h is induced b y (7) L : σ 7→ 1 n X s ∈ Σ n sgn ( s ) Z ∆ n ∂ σ ∂ t s (1) · σ − 1 ⊗ . . . ⊗ ∂ σ ∂ t s ( n ) · σ − 1 dt 1 ∧ . . . ∧ dt n for eac h smo o th map σ : ∆ n → GL m ( A ) . The last one is the generalized trace on con tin uous cyclic homolo g y TR : lim m →∞ H C n − 1 ( M m ( A )) → H C n − 1 ( A ) Definition 3.2. [9, 16] By the relativ e Chern c hara cter c h rel n : K rel n ( A ) → H C n − 1 ( A ) we wi l l understand the c omp osition (8) c h rel n = TR ◦ L ◦ h n of the Hur ewicz homomorphism , the lo garithm and the gener alize d tr ac e . 8 J. KAAD 3.2. The m ultiplicativ e c haracter. Let H be a separable Hilb ert space o f infinite dimension. F or eac h p ∈ N let M 2 p − 1 denote the unital C -subalgebra of L ( H ⊕ H ) consisting of op erators of the f orm  x 11 x 12 x 21 x 22  ∈ L ( H ⊕ H ) with x 12 , x 21 ∈ L 2 p ( H ) in the 2 p th Sc hatten ideal. The C -algebra M 2 p − 1 b ecomes a unital Banac h algebra when eq uipp ed with the norm k x k = k x k ∞ + k [ F U , x ] k 2 p F U =  1 0 0 − 1  Here the norms k · k ∞ and k · k 2 p are the op erator norm and the norm on the 2 p th Sc hatten ideal respectiv ely . F or details on the Sc hatten ideals w e refer to [28]. The contin uous linear map τ 2 p − 1 : C λ 2 p − 1 ( M 2 p − 1 ) → C ( x 0 , . . . , x 2 p − 1 ) 7→ − 1 2 2 p ( p − 1)! T r ( F U [ F U , x 0 ] · . . . · [ F U , x 2 p − 1 ]) determines a con tinuous cyclic co cycle and consequen tly a homomorphism τ 2 p − 1 : H C 2 p − 1 ( M 2 p − 1 ) → C F or details w e refer to [8]. Pre-comp osition with the relative Chern characte r th us yields a homomorphism τ 2 p − 1 ◦ c h rel 2 p : K rel 2 p ( M 2 p − 1 ) → C In [9] it is shown that K 2 p ( M 2 p − 1 ) is the cok ernel of v : K top 2 p +1 ( M 2 p − 1 ) → K rel 2 p ( M 2 p − 1 ) from the exact seq uence (6). F urthermore it is sho wn that the image of the homomorphism τ 2 p − 1 ◦ c h rel 2 p ◦ v : K top 2 p +1 ( M 2 p − 1 ) → C equals the additiv e subgroup (2 π i ) p Z of C . By consequen ce w e g et a homomorphism M U : K 2 p ( M 2 p − 1 ) ∼ = Cok er ( v ) → C / (2 π i ) p Z This is the o dd unive rsal multipli cativ e c haracter. No w, to eac h o dd unital 2 p - summable F redholm ( F , H ) o ver a unital C -algebra A w e can asso ciate a unital algebra homomorphism ρ F : A → M 2 p − 1 whic h b y functorialit y of algebraic K -theory yields a homomorphism ( ρ F ) ∗ : K 2 p ( A ) → K 2 p ( M 2 p − 1 ) The m ultiplicativ e ch aracter of t he F r edholm mo dule ( F , H ) o v er A is then defined as the comp osition M F = M ◦ ( ρ F ) ∗ : K 2 p ( A ) → C / (2 π i ) p Z COMP ARISON OF SECONDAR Y INV ARI ANTS OF A LGEBRAIC K -THEOR Y 9 4. Comp aring the determinant inv ariant and the mul t iplica tive chara cter 4.1. F actorization through r elativ e contin uous c yclic homology. In this section w e sho w that the homomorphism τ 1 : H C 1 ( M 1 ) → C factorizes thro ug h a relativ e con tin uous cyclic homology group, asso ciated with an extension of M 1 b y the op erators of trace class. This result can b e found in condensed form in the pro of of [9, Theorem 5 . 6 ]. Let H b e a separable Hilb ert space and let M 1 b e the Banac h algebra considered in Section 3. Let P b e the pro jection P =  1 0 0 0  ∈ L ( H ⊕ H ) and let q : L ( H ) → L ( H ) / L 1 ( H ) = C 1 ( H ) b e the quotien t map. W e then ha ve a unital algebra homomorphism R : M 1 → C 1 ( H ) R : x 7→ q ( P xP ) Let T 1 ⊆ L ( H ) × M 1 b e the unital C -subalgebra suc h that ( S, x ) ∈ T 1 precisely when S − P xP ∈ L 1 ( H ) . The diagram (9) T 1 M 1 L ( H ) L ( H ) / L 1 ( H ) ✲ π 2 ❄ π 1 ❄ R ✲ q is th us a comm utativ e diagram o f unital C -algebras. Here π 1 and π 2 are the pro jections giv en b y π 1 ( S, x ) = S and π 2 ( S, x ) = x . The C - algebra T 1 b ecomes a unital Banac h algebra when equipp ed with the norm k ( S, x ) k = k P xP − S k 1 + k x k ∞ + k [2 P − 1 , x ] k 2 It fits in the short exact sequence of Banach algebras (10) 0 − − − → L 1 ( H ) i − − − → T 1 π 2 − − − → M 1 − − − → 0 whic h has a con tin uous linear section s : M 1 → T 1 x 7→ ( P xP , x ) The existence of the con tin uous linear right in v erse to π 2 yields a n isomorphism H C ∗ ( π 2 ) ∼ = H C ∗ ( Ker ( π 2 )) b et w een the relative con tinuous cyclic homology of π 2 and the homology of the k ernel complex asso ciated with the chain map ( π 2 ) ∗ : C λ ∗ ( T 1 ) → C λ ∗ ( M 1 ) In particular, eac h ele men t in H C 0 ( π 2 ) can b e represen ted b y an op erat o r of trace class ( S, 0) ∈ T 1 S ∈ L 1 ( H ) 10 J. KAAD Lemma 4.1. T he tr ac e T r : K er ( π 2 ) → C ( S, 0) 7→ T r ( S ) induc es a map on r elative c ontinuous cyclic homolo gy T : H C 0 ( π 2 ) → C Pr o of. Let α : C λ 1 ( M 1 ) → L ( H ) denote the map induced by α : ( x 0 , x 1 ) 7→ [ P x 0 P , P x 1 P ] Let β : C λ 1 ( T 1 ) → L 1 ( H ) b e the map induced b y β : (( S 0 , x 0 ) , ( S 1 , x 1 )) 7→ [ S 0 , S 1 ] − [ P x 0 P , P x 1 P ] W e then ha v e T r ◦ β = 0 F urthermore w e can express the comp osition of the Ho c hsc hild b oundary b : C λ 1 ( T 1 ) → C λ 0 ( T 1 ) = T 1 with t he pro jection π 1 : T 1 → L ( H ) as the sum π 1 ◦ b = β + α ◦ ( π 2 ) ∗ It follo ws that t he map T r ◦ b : Ker 1 ( π 2 ) → C v anishes, which in turn implies the desired result.  W e can no w pro v e that τ 1 : H C 1 ( M 1 ) → C factor izes through the relativ e con tin uous cyclic homology group H C 0 ( π 2 ) . Theorem 4.2. The cha r acter τ 1 : H C 1 ( M 1 ) → C factorizes as τ 1 = − T ◦ ∂ Her e ∂ : H C 1 ( M 1 ) → H C 0 ( π 2 ) is the b ound a ry map asso c iate d with the homomorph ism π 2 : T 1 → M 1 . Pr o of. W e start b y noting that the comp osition − T r ◦ ( b ◦ s ∗ − s ∗ ◦ b ) : C λ 1 ( M 1 ) → C coincides with the map τ 1 : C λ 1 ( M 1 ) → C Here s ∗ : C λ 1 ( M 1 ) → C λ 1 ( T 1 ) denotes the map induced b y t he con t inuous linear righ t inv erse s : M 1 → T 1 of π 2 : T 1 → M 1 . Since the b oundary map ∂ : H C 1 ( M 1 ) → H C 0 ( π 2 ) ∼ = H C 0 ( Ker ( π 2 )) is induced by t he map b ◦ s ∗ : Z λ 1 ( M 1 ) → Ker ( π 2 ) w e get the desired result.  COMP ARISON OF SECONDAR Y INV ARI ANTS OF A LGEBRAIC K -THEOR Y 11 As an immediate consequence of Theorem 4.2 w e get the follo wing factorization result for the comp osition of the r elative Chern c haracter and the homomorphism τ 1 : H C 1 ( M 1 ) → C . Corollary 4.3. The c omp osition τ 1 ◦ ch rel 2 : K rel 2 ( M 1 ) → C factorizes as τ 1 ◦ ch rel 2 = − T ◦ TR ◦ L ◦ ∂ ◦ h 2 Her e the map s TR : lim m →∞ H C 1 ( M m ( π 2 )) → H C 1 ( π 2 ) and L : H 2 ( R ( π 2 )) → lim m →∞ H C 1 ( M m ( π 2 )) ar e r elative ve rs i o ns of the g e ner alize d tr ac e a n d the lo garithm. Th e map ∂ : H 2 ( R ( T 1 )) → H 1 ( R ( π 2 )) is the b oundary map a sso ciate d with the simplicial map R ( π 2 ) : R ( T 1 ) → R ( M 1 ) . 4.2. Det er minan ts and the relative logarithm. In this part of the pap er w e start relating the mu ltiplicativ e c haracter with the F redholm determinan t. The main result of this section is th us Lemma 4.6, where w e show that the comp osition of the relativ e logar ithm L : H 1 ( R ( π 2 )) → lim m →∞ H C 0 ( M m ( π 2 )) with t he exp onential of the trace exp ◦ ( − T ◦ TR ) : lim m →∞ H C 0 ( M m ( π 2 )) → C ∗ is a F redholm determinan t. Definition 4.4. L et G den o te the op er ators of determinant class. By a smo oth map σ : ∆ n → G we wi l l understand an element in the gr oup σ ∈ GL  C ∞ (∆ n , L ( H ))  such that σ − 1 ∈ M ∞  C ∞ (∆ n , L 1 ( H ))  Define the map ˜ L : Z [ R ( T 1 ) 1 × R ( M 1 ) 1 R ( T 1 ) 1 ] = ˇ C 10 ( R ( π 2 )) → C b y t he a ssignmen t ˜ L : ( σ 1 , σ 2 ) 7→ − T r Z 1 0 d ( σ 1 σ − 1 2 ) dt · σ 2 σ − 1 1 dt Here the tra ce T r : M ∞ ( L 1 ( H )) → C is induced by T r : x 7→ P m i =1 T r ( x ii ) for all x ∈ M m  L 1 ( H )  12 J. KAAD Lemma 4.5. T he map ˜ L : Z [ R ( T 1 ) 1 × R ( M 1 ) 1 R ( T 1 ) 1 ] → C p asses to a map on homolo gy ˜ L : H 1 ( Cok er ( δ R ( π 2 ) )) → C making the diagr am (11) H 1 ( R ( π 2 )) H 1 ( Cok er ( δ R ( π 2 ) )) C ❄ − T ◦ TR ◦ L ✛ ε ✟ ✟ ✟ ✟ ✟ ✟ ✟ ✙ ˜ L c ommute. Pr o of. W e show that ˜ L : Z [ R ( T 1 ) 1 × R ( M 1 ) 1 R ( T 1 ) 1 ] → C agrees with the map − T ◦ TR ◦ L ◦ ε : Z [ R ( T 1 ) 1 × R ( M 1 ) 1 R ( T 1 ) 1 ] → C See Section 2.1. Here TR ◦ L : Ker 1 ( R ( π 2 )) → Ker ( π 2 ) is induced b y the restriction of TR ◦ L : Z [ R ( T 1 ) 1 ] → T 1 to t he kern el of R ( π 2 ) : Z [ R ( T 1 ) 1 ] → Z [ R ( M 1 ) 1 ] . This essen tially amoun ts to a pro of of the equalit y ˜ L ( σ 1 , σ 2 ) = T r  Z 1 0 dσ 2 dt · σ − 1 2 dt − Z 1 0 dσ 1 dt · σ − 1 1 dt  for all ( σ 1 , σ 2 ) ∈ R ( T 1 ) 1 × R ( M 1 ) 1 R ( T 1 ) 1 . No w, since the trace T r : M m ( L 1 ( H )) → C is con tinuous and linear we ha v e ˜ L ( σ 1 , σ 2 ) = − T r Z 1 0 d ( σ 1 σ − 1 2 ) dt · σ 2 σ − 1 1 dt = − Z 1 0 T r  d ( σ 1 σ − 1 2 ) dt · σ 2 σ − 1 1  dt Using the Leibnitz rule for the differen tia l op erator d dt w e get d ( σ 1 σ − 1 2 ) dt = dσ 1 dt · σ − 1 2 − σ 1 σ − 1 2 · dσ 2 dt · σ − 1 2 COMP ARISON OF SECONDAR Y INV ARI ANTS OF A LGEBRAIC K -THEOR Y 13 Defining α ∈ M ∞  C ∞ (∆ 1 , L 1 ( H ))  b y α + 1 = σ 2 σ − 1 1 w e hav e − T r  d ( σ 1 σ − 1 2 ) dt · σ 2 σ − 1 1  = − T r  dσ 1 dt · σ − 1 1 − ( α + 1) − 1 dσ 2 dt · σ − 1 2 ( α + 1)  = T r  ( α + 1) − 1 dσ 2 dt · σ − 1 2 α  + T r  ( α + 1) − 1 dσ 2 dt · σ − 1 2 − dσ 1 dt · σ − 1 1  = T r  dσ 2 dt · σ − 1 2 − dσ 1 dt · σ − 1 1  pro ving the desired r esult.  In the next Lemma we will sho w that the comp osition of the map ˜ L : H 1 ( Cok er ( δ R ( π 2 ) )) → C with t he exp onential exp : C → C ∗ can b e expresse d using a F redholm determinan t. By [9, Prop osition 5 . 4 ] the first alg ebraic K -g roup of the Banac h algebra T 1 v anishes. In particular, it fo llo ws from the exact sequence (10) that the groups E ( M 1 ) and GL 0 ( M 1 ) coincide. Here GL 0 ( M 1 ) denotes the connected comp o nen t of the iden tit y . W e therefore ha v e a commutativ e diagram of simplicial maps R n ( T 1 ) E ( T 1 ) n R n ( M 1 ) E ( M 1 ) n ✲ θ ❄ R ( π 2 ) ❄ E ( π 2 ) ✲ θ and b y conseque nce an induced map θ : H 1 ( Cok er ( δ R ( π 2 ) )) → H 1 ( Cok er ( δ E ( π 2 ) )) F urthermore w e let D ∗ : H 1 ( Cok er ( δ E ( π 2 ) )) → H 1 ( Cok er ( δ E ( q ) )) denote the map induced by the comm utativ e diagra m E ( T 1 ) E ( L ( H )) E ( M 1 ) E ( C 1 ( H )) ✲ E ( π 1 ) ❄ E ( π 2 ) ❄ E ( q ) ✲ E ( R ) of group homomorphisms. 14 J. KAAD Lemma 4.6. T he diag r am (12) H 1 ( Cok er ( δ R ( π 2 ) )) H 1 ( Cok er ( δ E ( q ) )) C C ∗ ❄ ˜ L ✲ D ∗ ◦ θ ❄ det ✲ exp is c ommutative. Pr o of. W e show that for eac h generator ( σ 1 , σ 2 ) ∈ R ( T 1 ) 1 × R ( M 1 ) 1 R ( T 1 ) 1 w e ha ve (exp ◦ ˜ L )( σ 1 , σ 2 ) = det  E ( π 1 )( σ 1 ( 1 ) − 1 ) · E ( π 1 )( σ 2 ( 1 ))  Indeed the smooth map σ 1 σ − 1 2 : ∆ 1 → G L ( T 1 ) is of the form σ 1 σ − 1 2 = ( α, 1) with α : ∆ 1 → G b eing a smo oth map in the sense of De finition 4.4. It follow s that (exp ◦ ˜ L )( σ 1 , σ 2 ) = exp  − T r Z 1 0 dα dt · α − 1 dt  But this is precisely t he F redholm determinan t of the endp oint det ( α ( 1 ) − 1 ) = det  E ( π 1 )( σ 2 ( 1 )) · E ( π 1 )( σ 1 ( 1 ) − 1 )  pro ving the desired r esult.  4.3. P r o of of the main result. In this section we com bine the results of Lemma 4.2 and Lemma 4.6, in order to obtain a pro of of our main theorem: the determinan t in v a rian t equ als the m ultiplicativ e c haracter up to a canonical homomorphism on algebraic K -theory . W e start b y making a couple of useful o bserv ations. Let θ : H 2 ( R ( M 1 )) → H 2 ( E ( M 1 )) denote the homomorphism induced by the simplicial map θ : R n ( M 1 ) → E ( M 1 ) n The comp osition θ ◦ h 2 : K rel 2 ( M 1 ) → H 2 ( E ( M 1 )) then corr esp onds to the map θ : K rel 2 ( M 1 ) → K 2 ( M 1 ) under the identifi cation K 2 ( M 1 ) ∼ = H 2 ( E ( M 1 )) . F urthermore, the simplicial map R ( π 2 ) : R ( T 1 ) → R ( M 1 ) is surjectiv e in eac h degree. This follo ws from [1, Prop osition 3 . 4 . 3 ] b y the surjectivit y o f the map M m ( π 2 ) : M m ( C ∞ (∆ n , T 1 )) → M m ( C ∞ (∆ n , M 1 )) COMP ARISON OF SECONDAR Y INV ARI ANTS OF A LGEBRAIC K -THEOR Y 15 and since eac h ele men t σ ∈ R ( M 1 ) n can b e represen ted by a n elemen t σ ∈ GL 0  M m ( C ∞ (∆ n , M 1 ))  By Theorem 2.1 the map ε : H ∗ ( Cok er ( δ R ( π 2 ) )) → H ∗ ( R ( π 2 )) th us b ecomes an isomorphism. W e are no w ready to pro v e the main result of the pap er. Theorem 4.7. The c omp osition o f the universal m ultiplic ative c har acter M U : K 2 ( M 1 ) → C / (2 π i ) Z with the exp onential function exp : C / (2 π i ) Z → C ∗ c oincides with the c omp os ition of the ho momorphism on algebr aic K -the ory R ∗ : K 2 ( M 1 ) → K 2 ( C 1 ( H )) induc e d by the unital algebr a h o momorphism R : M 1 → C 1 ( H ) x 7→ q ( P xP ) and the universal determinant inva riant d : K 2 ( C 1 ( H )) → C ∗ That is exp ◦M U = d ◦ R ∗ Pr o of. F rom the long exact sequence (6) w e see that θ : K rel 2 ( M 1 ) → K 2 ( M 1 ) implemen ts the isomorphism Cok er ( v ) ∼ = K 2 ( M 1 ) By construction of the multiplic ativ e c haracter it is therefore enough to pro v e the equalit y exp ◦ τ 1 ◦ ch rel 2 = d ◦ R ∗ ◦ θ See Section 3. W e ev olv e on the left hand side. The iden tifications H 1 ( R ( π 2 )) ∼ = H 1  Cok er ( δ R ( π 2 ) )  and H 1 ( E ( q )) ∼ = H 1  Cok er ( δ E ( q ) )  as w ell as the iden tifications K 2 ( M 1 ) ∼ = H 2 ( E ( M 1 )) and K 2 ( C 1 ( H )) ∼ = H 2 ( E ( C 1 ( H ))) will b e suppressed. F urthermore w e will mak e use of the results in Corollary 4.3, Lemma 4.5 and Lemma 4.6 as w ell as of the observ ations preceeding the statemen t of the Theorem. W e 16 J. KAAD get exp ◦ τ 1 ◦ ch rel 2 = exp ◦ ( − T ◦ TR ◦ L ◦ ∂ ◦ h 2 ) = exp ◦ ˜ L ◦ ∂ ◦ h 2 = det ◦ D ∗ ◦ θ ◦ ∂ ◦ h 2 = det ◦ ∂ ◦ E ( R ) ∗ ◦ θ ◦ h 2 = det ◦ ∂ ◦ R ∗ ◦ θ But this is precisely t he comp osition d ◦ R ∗ ◦ θ : K rel 2 ( M 1 ) → C ∗ pro ving the theorem.  The main theorem can b e refined in the follow ing w ay . Let ρ F : A → M 1 b e the unital algebra homomorphism asso ciated with an o dd unital 2 - summable F redholm mo dule ( F , H ) ov er the C -algebra A . W e then define a C -subalgebra of L ( H ) b y E = { P ρ F ( a ) P + T | T ∈ L 1 ( H ) and a ∈ A } Letting B denote the quotien t o f E by L 1 ( H ) w e get an exact sequence X F : 0 − − − → L 1 ( H ) i − − − → E π − − − → B − − − → 0 W e then hav e the algebra homomorphism ι : B → C 1 ( H ) whic h is induced b y the inclusion ι : E → L ( H ) and the surjectiv e a lg ebra homomorphism R F : A → B a 7→ q ( P ρ F ( a ) P ) Here q : L ( H ) → C 1 ( H ) denotes the quotien t map. Corollary 4.8. The diagr am K 2 ( A ) K 2 ( B ) C / (2 π i ) Z C ∗ ✲ ( R F ) ∗ ❄ M F ❄ d ( X F ,ι ) ✲ exp is c ommutative. Pr o of. The result follows from Theorem 4.7 a nd the identit y R ◦ ρ F = ι ◦ R F : A → C 1 ( H ) using t he definition of the in v arian ts and the functorialit y of algebraic K -theory .  COMP ARISON OF SECONDAR Y INV ARI ANTS OF A LGEBRAIC K -THEOR Y 17 References [1] B. Bla ck adar, K-The ory for Op er ator Al gebr as , Mathematical Sciences Research Institute Publications , Springer-V erlag New Y ork, 19 86 . [2] K. S. Brown, Cohomolo gy of Gr oups , Gra dua te T exts in Math., v ol. 87 , Springer-V erlag, New Y or k, Heidelber g, Berlin, 19 8 2 . [3] L. G. B rown, The Determinant Invariant for Op er ators with T r ac e Class Self Commutators , Pro c. Co nf. on Ope r ator T heo ry , Lec ture Notes in Math., vol. 3 45 , Springer-V erlag, Berlin, Heidelb erg, and New Y o r k, 1973 , 210 - 228 . [4] L. G. Br own, Op er ator Algebr as and Algeb r aic K.the ory , B ull. Amer. Math. So c., 81 , 1973 , 1119 - 1121 . [5] R. W. Carey , J . D. Pincus. Pertu rb ation V e ctors , Int egra l Equations Op er a tor Theo ry , 35 , , 19 99 , 271 - 3 65 . [6] R. W. Ca rey , J. D. Pincus. Steinb er g Symb ols Mo dulo the T r ac e Class, Holonomy, and Limit The or ems for T o eplitz Determina nts , T ransactions of the American Ma thematical So ciety , vol. 358 , 200 4 , 5 09 - 55 1 . [7] H. Cartan and S. Eilen ber g, Homolo gic al Alg ebr a , Princeton Mathematica l Series, v ol. 19 , Princeton Univ. Press, Princeton, 19 56 . [8] A. Connes, Nonc ommutative Ge ometry , Academic Pre s s, 1994 . [9] A. Connes and M. Ka roubi, Car actèr e multiplic atif d’un mo dule de F r e dholm , K-theory 2 , 198 8 , 431 - 463 . [10] J. L. Dupont, Curvatur e and Char acteristic Classes , Lecture Notes in Mathematics, v ol. 6 40 , Springer V er lag, Berlin, Heidelb erg, New Y ork, 1978 . [11] A. Hatcher, Algebr aic T op olo gy , Cambridge Univ ersity Press, 20 0 2 . [12] J. W. Helton a nd R. E. Howe, Inte gr al O p er ators: Commutators, T r ac es, Index, and Homolo gy , Pro c. Conf. on O per ator Theory , Lecture Notes in Math., vol. 34 5 , Springer-V erlag, Berlin, Heidelb erg, New Y o r k, 1973 , 141 - 209 . [13] G. Donadze, N. Ina ssaridze, T. Porter, N-fold ˇ Ce ch D erive d F unctors and Gener alize d Hopf T yp e F or- mulas K -theory preprint archiv es, 624 , 2 003 . [14] G. Ginot, F ormules explicites p our le c ar actèr e de Chern en K-thé orie algébrique , Ann. Inst. F ourier, Grenoble 54 , 200 4 , 23 27 - 23 5 5 . [15] A. Grothendieck, Pr o duits T ensoriels T op olo giques , Mem. AMS., No. 16 , 195 5 . [16] M. Karo ubi, Homolo gie Cyclique et K -thé orie , Astérisque No. 1 49 , 1987 . [17] M. Levine, R elative Mil nor K-The ory , K-theor y 6 , 1992 , 11 3 - 175 . [18] J. -L. Lo day , K- thé orie algébrique et re pr ésent ations de gr oup es , Ann. Scient. Ecole Nor m. Sup. 9 , 19 76 , 309 - 3 77 . [19] J. -L. Lo day , Cyclic Homolo gy , Grundlehren der mathematischen Wissenschaften, vol. 30 1 , Spr ing er- V er lag, New Y o rk, Berlin, Heidelbe r g, 199 2 . [20] J. -L. Lo day and D. Quillen, Cyclic Homolo gy and The Lie Algebr a Homolo gy of Matric es , Comment. Math. Helv. 59 , 19 84 , 56 5 - 591 . [21] D. Mac duff and P . de la Ha r pe, A cyclic gr oups of automorphisms , Commen t. Math. Helv. 58 , 1 983 , 48 - 71 . [22] J. P . May , Simplicia l Obje cts in Algebr aic T op olo gy , C hica go Lectures in Mathematics, 199 3 . [23] G. K. Pedersen, C ∗ -Algeb r as and Their Automorphism Gr oups , Academic Press, London, New Y ork, San F rancisco, 1979. [24] D. Quillen, Higher Algebr aic K-the ory I , Algebra ic K - theory I (Battelle, 1972 ) (H. Bass, ed.), Lecture Notes in Math., vol. 3 41 , Springer-V erlag, New Y ork, Berlin, Heidelb erg, 197 3 , 85 - 147 . [25] J. Rose n b erg, A lgebr aic K-the ory and Its Appli c ations , Graduate T exts in Math., vol. 14 7 , Springer- V er lag, New Y o rk, Berlin, Heidelbe r g, 199 4 . [26] J. Rosenberg, Comp arison Betwe en Algebr aic and T op olo gic al K-The ory for Banach Algebr as and C ∗ - Alg ebr as , Handbo ok o f Algebraic K -theor y , Springer, 20 04 , 843 - 87 4 . [27] A. Presley and G. Segal, L o op Gr oups , Clare ndon Press, Oxfo rd, 1986 . [28] B. Simon T r ac e Ide als and Their Applic ations , Cambridge Universit y Press, 19 79 . [29] U. Tillmann, F actori zation of the (R elative) C hern Char acter thr ough Lie A lgebr a Homolo gy , K -theory 6 , 199 2 , 457 - 463 . 18 J. KAAD [30] J. B. W agoner, Delo oping Classifyi ng Sp ac es in Algebr aic K -t he ory , T opolo gy , V o l. 1 1 , 1972 , 349 - 370 . [31] A. W a ssermann, Ana lysis of Op er ators , Michaelmas, 2006 . [32] G. W. Whitehead, Elements of Homotopy The ory , Graduate T exts in Math., vol. 61 , Springer-V erlag, New Y ork, Heidelberg , Berlin, 1978 .

Original Paper

Loading high-quality paper...

Comments & Academic Discussion

Loading comments...

Leave a Comment