A splitting for \(K_{1}\) of completed group rings (Q373485)

In classical (commutative) Iwasawa theory, one method for constructing \(p\)-adic \(L\)-functions uses the Coleman isomorphism \[ \begin{aligned}\bigl({\mathbb Z}_p [[T]]^\times \bigl)^{{\mathcal N} = \mathrm{id}}\displaystyle\buildrel\sim\over\to\displaystyle\mathop{\lim}_{\longleftarrow}{\mathbb Z}_p [\xi_{p^n}]^\times,~F \mapsto \bigl( F (\xi_{p^n} - 1) \bigl)_n,\end{aligned} \] where \({\mathcal N}\) is Coleman's ``norm'' operator. By the theory of the field of norms of Fontaine-Wintenberger, the inverse limit above can be identified with the group of units of a specific local field \(E\) of characteristic \(p\); the choice of the \(\xi_{p^n}\) gives a prime element of \(E\) which allows to identify \({\mathcal O}_E\) with the power series ring \({\mathbb F}_p [[T]]\). Modulo these identifications, the Coleman map is simply the natural projection \({\mathbb Z}_p [[T]] \longrightarrow {\mathbb F}_p [[T]],\) and Coleman's theorem provides a natural section for this projection. \noindent In non commutative Iwasawa theory one investigates towers of number fields whose Galois group \(G\) is a \(p\)-adic Lie group of a certain type. Let \(\Lambda (G)\) and \(\Omega (G)\) denote respectively the completed group rings of \(G\) over \({\mathbb Z}_p\) and \({\mathbb F}_p\). The problem of constructing \(p\)-adic \(L\)-functions in this context is closely related to the computation of the algebraic \(K\)-group \(K_1 (\Lambda (G))\) (see e.g. \textit{T. Fukaya} and \textit{K. Kato} [in: Proceedings of the St. Petersburg Mathematical Society. Vol. XII. Transl. from the Russian. Uraltseva, N.N.(ed.), Providence, RI: American Mathematical Society (AMS). Translations. Series 2. American Mathematical Society 219, 1--85 (2006; Zbl 1238.11105)], so that Coleman's theorem suggests to look for a section for the natural map \(K_1 (\Lambda (G)) \displaystyle\buildrel\pi\over{\rightarrow\!\!\!\!\rightarrow} K_1 (\Omega (G))\) which could be characterized by a certain ``norm type'' operator equation in \(K_1 (\Lambda (G))\). \noindent In the present paper, the authors start from a pro-\(p\) \(p\)-adic Lie group \(G (p \not= 2)\). The formalism of the exponential map, as developed in \textit{R. Oliver}'s book, [London Math. Soc. Lecture Notes Series 132, Cambridge (1988)], gives an isomorphism \(\exp (p_{\bullet}):\Lambda (G)^{\mathrm{ab}} := \Lambda(G) / \overline{[\Lambda(G), \Lambda(G)]} \displaystyle\buildrel\sim\over\to \ker\pi ,\) which is induced, via the classical isomorphism \(K_1 (\Lambda (G)) \simeq \Lambda (G)^\times / [\Lambda(G)^\times, \Lambda(G)^\times],\) by the map \(p \Lambda (G) \to 1 + p \Lambda (G)\) defined by the usual exponential power series. It is then natural to introduce the \(p\)-adic logarithm \(\Gamma:K_1(\Lambda (G)) \rightarrow \Lambda (G)^{\mathrm{ab}}\) of Oliver-Taylor. Assuming the hypothesis \((SK):\) ``\(S K_1 (\Lambda (G)) = 0\)'', it is known that \(\ker\Gamma = \mu_{p-1} \times G^{\mathrm{ab}}\) and \(\Gamma \circ \exp (p \ldotp) = p-\Phi,\) where \(\Phi\) is the additive endomorphism of \(\Lambda (G)^{\mathrm{ab}}\) induced by the map \(\varphi:g \in G \mapsto g^p \in G\). Assuming the hypothesis \((\Phi)\) : ``\(\varphi\) is injective and \(\varphi^n (G)\) is open in \(G\) for any \(n \geq 1\)'' (which holds for any uniform G), the authors can show that the inverse image by \(\Gamma,\) say \(K^\Phi_1 (\Lambda (G)),\) of a certain subgroup of \(\Lambda (G)^{\mathrm{ab}}\) provides a section for \(\pi,\) i.e. \(K^\Phi_1 (\Lambda (G))\displaystyle\buildrel\sim\over{\rightarrow} K_1(\Omega(G))\) via \(\pi\). To describe further the subgroup \(K^\Phi_1 (\Lambda (G)),\) they assume another hypothesis \((P):\) ``\(\varphi(G)\) is a subgroup of \(G\)'' (which also holds for any uniform \(G),\) in order to define an ``Adams operator'' \(\displaystyle\buildrel\sim\over\Phi:K_1(\Lambda (G)) \to K_1 (\Lambda (G)),\) \(x \mapsto \exp (p\Gamma (x))^{-1} x^p\) (in multiplicative notation) such that \(\Gamma \circ \displaystyle\buildrel\sim\over\Phi = \Phi \circ \Gamma\). The main result is that \(K_1^\Phi (\Lambda (G)) = K_1 (\Lambda (G))^{N_G(\ldotp) = \displaystyle\buildrel\sim\over\Phi (\ldotp)p^{d-1}},\) where \(d = [G:\varphi(G)]\) and the ``norm'' map \(N_G\) is the composite map : \(K_1 (\Lambda (G)) \displaystyle\buildrel{\text{\sevenrm norm}}\over\longrightarrow K_1 (\Lambda (\varphi(G)) \displaystyle\buildrel{\text{\sevenrm can}}\over\longrightarrow K_1(\Lambda (G))\). This gives the desired section. \noindent To get closer to the aspect of groups of local units in the original Coleman isomorphism, the authors express the hypothesis \((SK)\) as the injectivity of the natural map \(K_1 (\Lambda (G)) \to K_1 (\Lambda^\infty (G)),\) where \(\Lambda^\infty(G)\) denotes the complete group ring of \(G\) over \({\mathbb Q}_p\). In the style of Fröhlich's Hom-description, it can be shown that \(K_1 (\Lambda^\infty(G)) \displaystyle\buildrel\sim\over{\rightarrow} \Hom_{{\mathcal G}_p} (R_G, \overline{\mathbb Q}_p^\times),\) where \({\mathcal G}_p = \mathrm{Gal} (\overline{\mathbb Q}_p/{\mathbb Q}_p)\) and \(R_G\) denotes the free abelian group on the set of isomorphism classes of all \(\overline{\mathbb Q}_p\)-irreducible representations of \(G\) which are trivial on some open subgroup. The authors then show their generalization of the Coleman map : \(K^\Phi_1 (\Lambda(G)) \hookrightarrow \Hom_{{\mathcal G}_p} \bigl( R_G / \mathrm{Im} (\iota^p - p^{d-1} \psi^p), {\mathbb Q}^\times_p \bigl),\) where \(\psi^p\) is the usual Adams operator and and \(r^p\) is the induction operator on \(R_G,\) given by \(\iota^p ([V]) = [V \displaystyle\mathop\otimes_{{\mathbb Q}_p} {\mathbb Q}_p [G/\varphi(G)]]\). In the case of the group \(G = {\mathbb Z}_p,\) the cokernel is \({\mathbb Z}^\times_p,\) but unfortunately, in the general case, the cokernel can be very big. \noindent In a final section, the authors show that the group \(G\) of lower triangular unipotent matrices in \(\mathrm{GL}_n({\mathbb Z}_p)\) satisfies the hypothesis (SK).