Iwasawa theory and explicit reciprocity law. A remake of an article of P. Colmez (Q1291961): Difference between revisions

Explicit reciprocity laws in local class field theory are usually meant as explicit formulae for the Hilbert symbol. One trend of development of these laws has been following Iwasawa theory over a local field \(K\) (unramified over \(\mathbb{Q}_p)\), building on the study of inverse limits of local units in the cyclotomic extension \(K(\mu_{p^\infty})\) by means of the exponential map (Iwasawa, Coleman, Coates-Wiles, \dots). A deep generalization to all \(p\)-adic crystalline representations \(V\) of the absolute Galois group \(G_K\) -- the classical case corresponding to \(V=\mathbb{Q}_p(1)\) -- has been proposed by \textit{B. Perrin-Riou} and proved by her for all twists \(\mathbb{Q}_p(r)\) [Invent. Math. 115, 81-149 (1994; Zbl 0838.11071)]. Roughly speaking, Perrin-Riou ``interpolates'' \(p\)-adically the exponential maps of Bloch-Kato to construct an exponential map \(\Omega_V\) between two certain Iwasawa modules, and her reciprocity law, \(\text{Rec}(V)\), essentially states that under the natural dualities, \(\Omega_V\) and \(\Omega_{V^*(1)}\) are mutually adjoint. The reciprocity law \(\text{Rec}(V)\) has been recently proved by P. Colmez, by Kato-Kurihara-Tsuji, and by D. Benois independently. Whereas Kato-Kurihara-Tsuji (resp. Benois) use syntomic cohomology (resp. Galois cohomology via \((\varphi, \Gamma)\)-modules), the method of Colmez is ``\(p\)-adic analytical'' in the following sense: For simplicity, assume that \(K=\mathbb{Q}_p\) and \(V\) is crystalline. Note that \textit{P. Colmez}'s results are available for de Rham representations [Ann. Math. (2) 148, 485-571 (1998; Zbl 0928.11045)]. For a topological generator \(u\) of \(1+p \mathbb{Z}_p\), the main game consists in computing \(f(u^k-1)\), where \(f\) is certain analytical fonction on the unit disk of \(\mathbb{C}_p\), obtained by interpolation. In particular, for any \(n\geq 1\), \(i\geq 0\), there exists a unique polynomial \(R_{n, i} (f)\) of degree \(<p^n\) such that \[ f\equiv R_{n,i}(f)\bigl(u^{-i}(1+T)-1\bigr) \bmod\bigl(u^{-ip^n} (1+T)^{p^n}- 1\bigr) \] With the usual notations of Iwasawa theory \((\gamma_n= \gamma^{p^{n-1}}, u=\chi(\gamma)\), etc.\dots), \(R_{n,i}(f) (u^{k-i}-1)\) is related to the value at \(\gamma_n\) of a certain cocycle \(\Gamma_n \to V\), and this cocycle becomes a coboundary \((\gamma_n-1)c_n\) when the coefficients are extended to \(B_{\text{cris}}^{G_{\mathbb{Q} (\mu_p\infty)}}\). Thus the value \(R_{n,i}(f) (u^{k-i}-1)/(1-u^{(k-i) p^n})\) is related to the image of \(c_n\) in \(\mathbb{Q}_n (\mu_{p^n})\) by the Tate map \[ \lambda_{k,n}: B_{dR}^{ G_{\mathbb{Q} (\mu_p\infty)}} \to B_{dR}/ (\chi^{-k}(\gamma_n) \gamma_n-1)\to\mathbb{Q}_n (\mu_{p^n}), \] and this image is in turn related to the dual exponential map by a formula of Kato. In this paper, which she calls a ``remake'', the author gives a simplified version of Colmez's proof, along the lines explained above. The crystalline hypothesis allows to use less subtle arguments than the original ones in the ring \(B_{\text{crys}}\). The construction of Colmez's map \(\text{Log}_V^{(h)}\) appears as a consequence of the reciprocity law \(R_n(V)\). Formulae on special values of the logarithm map are also given (they have been used previously by the author in her theory of \(p\)-adic \(L\)-functions).