Explicit weight two motivic cohomology complexes and algebraic \(K\)- theory (Q1204329)

This paper presents most of the proofs of statements and results of a paper by \textit{A. A. Bejlinson}, \textit{A. B. Goncharov}, \textit{V. V. Schechtman} and \textit{A. N. Varchenko} [in The Grothendieck Festschrift, Collect. Artic. in Honor of the 60th Birthday of A. Grothendieck, Vol. I, Prog. Math. 86, 135-172 (1990; Zbl 0737.14003)]. Let \(F\) be a field with at least 4 elements. For every positive integer \(n\) one can define an abelian group \(A_ n(F)\) generated by suitable pairs of \(n\)-simplices in \(\mathbb{P}^ n(F)\) with some imposed relations. It turns out that \(A_ 1(F)\) is isomorphic to \(F^ \times\). \(A_ 2(F)\) is one of the main objects of study of the paper. One defines functions \(\delta:F\to A_ 2(F)\) (the dilogarithm), \(\mu:F\times F\to A_ 2(F)\) (the rectangle function), inducing a morphism of abelian groups \(\mu:F^ \times\otimes F^ \times\to A_ 2(F)\), and the half-square map \(\nu:F\to A_ 2(F)\). Several relations between \(\delta\), \(\mu\) and \(\nu\) are derived. Also, a comultiplication \(\nu_{11}:A_ 2(F)\to F^ \times\otimes F^ \times\) is constructed by geometric means, i.e. based on geometric configurations of lines and triangles in \(\mathbb{P}^ 2(F)\). Inspired by functional properties of the dilogarithm one is led to define a group \(Q(F)\) as the quotient of the free abelian group on generators \([x]\) \((x\in F)\) by a subgroup of very specific elements reflecting the functional properties alluded to above. One defines a homomorphism \(\lambda:Q(f)\to F^ \times\otimes F^ \times\) by \(\lambda[x]=x\otimes 1-x\) for \(x\neq 0\), and \(\lambda[0]=1\otimes 1\). An important morphism \(f:Q(F)\to A_ 2(F)\) is defined by \([x]\mapsto\delta(x)+\mu(x\otimes 1-x)-\delta(1)\). It bears some peculiar relation of \(\delta\), \(\mu\) and \(\nu\), and one has \(\nu_{11}\circ\lambda=f\). As a corollary one obtains that \(A_ 2(F)\) is generated by the dilogarithms. Next comes the definition of the `Bloch groups' \(B_ n(F)\), in particular in the situation discussed in the paper, the group \(B_ 2(F)\). Its definition parallels that of \(A_ 2(F)\), i.e. it is generated by pairs of triangles in \(\mathbb{P}^ 2(F)\) with suitable relations imposed. There is a natural map \(i:B_ 2(F)\to A_ 2(F)\) with nontrivial kernel, and one defines a comultiplication map \(\nu_ 2:B_ 2(F)\to F^ \times\otimes F^ \times\). One can prove the identity \(\nu_ 2=\nu_{11}\circ i\). Finally, one constructs a map \(\eta:B_ 2(F)\to Q(F)\) via an intricate geometric procedure using lots of identities between suitable pairs of triangles. The upshot is that all the maps \(f\), \(\lambda\), \(\nu_{11}\), \(\eta\), \(\nu_ 2\) and \(i\) fit into a commutative diagram, thus relating \(Q(F)\), \(A_ 2(F)\), \(B_ 2(F)\), and \(F^ \times\otimes F^ \times\). Putting everything together, one can state the main result: Theorem. There is a complex \(0\to A_ 2(F)@>\nu_{11}>>A_ 1(F)\otimes A_ 1(F)\to 0\) such that (1) \(A_ 1(F)\simeq F^ \times\); (2) \(\text{coker}(\nu_{11})\simeq K_ 2(F)\). (3) Assume \(\text{char}(F)\neq 2\). Up to 2-torsion, there is an exact sequence \[ 0\to C\to K_ 3(F)_{\text{ind}}/\text{Tor}(F^ \times,F^ \times)\to\ker(\nu_{11})\to 0 \] with \(3C=0\).