Third homology of \({\mathrm{SL}}_2\) and the indecomposable \(K_3\) (Q897092)

For any infinite field \(F\), there is a natural map \(K^M_n(F) \to K_n(F)\) between the Milnor ring of \(F\) and the algebraic \(K\)-theory of \(F\), which is an isomorphism for \(0 \leq n \leq 2\). We define the indecomposable part \(K_n(F)^{\text{ind}}\) of \(K_n(F)\) to be the cokernel of this map. The Hurewicz homomorphism \(K_3(F)\to H_3(\mathrm{SL}(F))\) induces an isomorphism \(H_3(\mathrm{SL}(F))\cong K_3(F)/\{-1\} \cdot K_2(F)\), so that we have an induced surjection \(H_3(\mathrm{SL}(F)) \to K_3^{\text{ind}}(F)\). This surjection yields a map \(\alpha : H_3(\mathrm{SL}_2(F)) _{F^*}=H_0(F^*, H_3(\mathrm{SL}_2(F))) \to K_3^{\text{ind}}(F)\). The main theorem of this paper states that this \(\alpha\) is bijective if and only if both the natural homomorphisms \(H_3(\mathrm{GL}_2(F)) \to H_3(\mathrm{GL}_3(F))\) and \(H_3(\mathrm{SL}_2(F))_{F^*} \to H_3(\mathrm{GL}_2(F))\) are injective. If we focus on the question of bijectivity of \(\alpha\) due to Suslin, then the result in the theorem can be read as an effective criterion for the surjectivity of this map. For this reason, if trying to roughly sketch the proof in the \(\lq\lq\)if'' direction, it can be seen that this proof is based on the use of two short exact sequences with \(K_3(F)^{\text{ind}}\) and \(\tilde{H}_3(\mathrm{SL}_2(F))\) (details omitted) middle terms, which are due to \textit{A. A. Suslin} [Proc. Steklov Inst. Math. 183, 217--239 (1991; Zbl 0741.19005); translation from Tr. Mat. Inst. Steklova 183, 180--199 (1990)] and the author and Mokari (Preprint) respectively. In addition to these two exact sequences we are given a map \(\zeta : H_3(\mathrm{SL}_2(F)) _{F^*} \to \tilde{H}_3(SL_2(F))\) between the middle terms of them and so, if we suppose \(\alpha(x)=0\) for \(x \in H_3(SL_2(F)) _{F^*}\), then comparing these two exact sequences we obtain \(\zeta(x)=0\). In consequence, applying the injectivity assumption in the \(\lq\lq\)only if'' statement to this equality, we get \(x=0\).