Normal elements in the mod-\(p\) Iwasawa algebra over \(\mathrm{SL}_n(\mathbb{Z}_p)\): a computational approach (Q2279815)

Given a compact $p$-adic analytic group $G$, the so-called \textit{mod-$p$ Iwasawa algebra $\Omega_G$} is defined as the projective limit of the group algebras $\mathbb{F}_p[G/N]$ when $N$ runs through the normal subgroups of $G$. First studied by \textit{M. Lazard} [Publ. Math., Inst. Hautes Étud. Sci. 26, 389--603 (1965; Zbl 0139.02302)], these complete semilocal noetherian algebras have known a revival of interest with developments in the Langlands program (on $p$-adic representations) or in Iwasawa theory (on arithmetic algebraic geometry of elliptic curves). The present article is the last in a series where the authors are concerned with normal elements of noncommutative Iwasawa algebras over $\mathrm{SL}_n(\mathbb{Z}_p)$. Here $a\in\Omega_G$ is called normal if $a\Omega_G=\Omega_G a$ in which case $I = a\Omega_G$ is a reflexive ideal, meaning that the canonical map $I\to \Hom_{\Omega_G}(\Hom_{\Omega_G}(I,\Omega_G), \Omega_G)$ is an isomorphism. The natural question is then the existence of nontrivial two-sided reflexive ideals in $\Omega_G$. A negative answer was given by \textit{K. Ardakov} et al. [J. Algebra 320, No. 1, 259--275 (2008; Zbl 1162.16013)] using derivation hypothesis, powerful Lie algebras, and a condition on the prime $p$ called ``nice''. More precisely, let $G$ be a compact $p$-adic analytic group whose $\mathbb{Q}_p$-Lie algebra $\mathcal{L}(G)$ is split semisimple over $\mathbb{Q}_p$. If $p$ is a nice prime for the root system of $\mathcal{L}(G)$ then $\Omega_G$ has no non trivial two-sided reflexive ideals; in particular, every nonzero normal element of $\Omega_G$ is a unit. It was however believed that the ``nice'' hypothesis was superfluous. Here the authors provide an entirely different proof, using partial formal differentiation of non commutative polynomials to show that an element $W\in\Omega_G$ is normal if and only if, as a non commutative formal series, it contains non trivial constant terms (in which case $W$ is a unit). This approach allows to get rid of the ``nice'' condition on $p$, and it also has potential application to the study of the mod-$p$ Iwasawa algebra over the first congruence subgroup $\Gamma_1(\mathrm{SL}_n(\mathbb{Z}_p))$. The drawback -- acknowledged by the authors themselves -- is a ``formidable'' system of \textit{fourteen} partial differential equations, whose ``tedious and complicated'' treatment must be relegated into two devoted appendices in order to facilitate the reading of the main argument.