Algebraic \(K\)-theory of number fields and rings of integers and the Stickelberger ideal (Q1185528)

Let \(F\) be an abelian extension of \(\mathbb{Q}\). The Stickelberger theorem constructs an ideal \(S_ 0\) in \(\mathbb{Z}[G(F/\mathbb{Q})]\) that annihilates the class group of \(F\). The ideal \(S_ 0\) is generated by elements of the form \[ \Theta_ 0(b)=(b-(b,F))\sum_{(a;f)=1;1\leq a<f}\zeta_ f(a,0)(a,F)^{-1}. \] \(f\) is the conductor of \(F\), \((a,F)\) is the restriction to \(F\) of the automorphism of \(\mathbb{Q}(\xi_ f)\), sending \(\xi_ f\) to \(\xi_ f^ a\). \(\zeta_ f(a,s)\) is the partial zeta function and \((b;2f)=1\). Coates and Sinnott defined analogous Stickelberger elements: \[ \Theta_ n(b)=(b^{n+1}- (b,F))\sum_{(a;f)=1;1\leq a<f}\zeta_ f(a,-n)(a,F)^{-1}. \] Let \(S_ n\) denote the ideal in \(\mathbb{Z}[G(F/\mathbb{Q})]\) generated by such elements. Inspired by partial results and the Lichtenbaum conjecture, they conjectured that \(S_ n\) annihilates \(K_{2n}(O_ F)\). This paper studies the relationship between algebraic \(K\)-groups and the Stickelberger elements to give some evidence for this conjecture. The following results are shown. Theorem A: Let \(D=\bigcap_{r\geq1}K_{2n}(F)^ r\) denote the subgroup of divisible elements of \(K_{2n}(F)\). Then \(nS_ n\) annihilates \(D_ \ell\) for each odd prime \(\ell\). Theorem B: If \(F\) is totally real abelian, \(n\geq 1\) an odd integer and \(\ell>n\) an odd prime, then there is a surjective map \[ \tilde c_{n+1,2}: K_{2n}(F)_ \ell \to H_{cts}^ 2(F,\mathbb{Z}_ \ell(n+1))_ \ell. \] Here \(H_{cts}\) is the continuous cohomology. Theorem C: For an odd, positive integer \(n\) and \(\ell>n\): \[ \left|{{w_{n+1}(F)\zeta_ F(-n)} \over {\prod_{v\mid\ell}w_ n(F_ v)}}\right|_ \ell^{-1}\quad\text{divides}\quad \#(\bigcap_{r\geq 1} K_{2n}(F)^ r)_ \ell. \] Using this result and its consequences, it is possible to determine for each odd prime number \(\ell\), except for irregular prime numbers such that \(\ell\leq n\) and \(\ell\mid nw_{n+1}(\mathbb{Q})\zeta(-n)\), whether or not the short exact sequence \[ 0\to K_{2n}(\mathbb{Z})\to K_{2n}(\mathbb{Q})\to\oplus_ v K_{2n- 1}(k_ v)\to 0 \] splits for the \(\ell\)-torsion part. Theorems A and C give the evidence for the conjecture of Coates and Sinnott.