Enumerating submodules invariant under an endomorphism (Q2396228): Difference between revisions

For a commutative and unital ring \(R\) with polynomial submodule growth, finitely generated \(R\)-module \(M\) and \(A\in\mathrm{End}_R(M)\), define \[ \zeta_{A,R}(s)=\sum^{\infty}_{m=1}a_{m}(A,R)m^{-s}, \] where \(a_{m}(A,R)\) is the number of \(A\)-invariant \(R\)-submodules \(U\le M\) with \(|M:U|=m\). Let \(k\) be a number field with ring of integer \(\mathfrak{o}\). Denote \(\mathcal{V}_k\) by the set of non-Archimedean places of \(k\). For \(v\in \mathcal{V}_k\), let \(k_v\) be the \(v\)-adic completion of \(k\), \(\mathfrak{o}_v\) the valuation ring of \(k_v\), \(\mathfrak{p}_v\in\mathrm{Spec}(\mathfrak{o})\) the prime ideal corresponding to \(v\) and \(q_v=|\mathfrak{o}/\mathfrak{p}_v|\) the residue field size of \(k_v\). Moreover, for a finite set \(S\in\mathcal{V}_k\), let \(\mathfrak{o}_S\) be the ring of \(S\)-integers of \(k\). In the present paper, the author investigates analytic properties of \(\zeta_{A,R}(s)\) when \(R=\mathfrak{o}_S\), \(M=\mathfrak{o}_S^n\) and \(A\in \mathrm{M}_n(\mathfrak{o}_S)\) with \(\mathrm{M}_n(R)\) being the algebra of \(n\times n\) matrices over \(R\). To state the result, for a possibly non-associative \(R\)-algebra \(\mathsf{A}\) whose underlying \(R\)-module is finitely generated, define the ideal zeta function \[ \zeta_{\mathsf{A}}(s)=\sum_{\mathsf{I}\triangleleft_R \mathsf{A}\ |\mathsf{A}:\mathsf{I}|<\infty}|\mathsf{A}:\mathsf{I}|^{-s}, \] where \(\mathsf{I}\triangleleft_R \mathsf{A}\) means that \(\mathsf{I}\) is a two-sided ideal of \(\mathsf{A}\). Moreover, let \(((f_1,\boldsymbol{\lambda}_1),\ldots,(f_e,\boldsymbol{\lambda}_e))\) be the elementary divisor vector of \(A\) over \(k\). Namely, for \(1\le i\le e\), \(f_i\in k[X]\) is a monic irreducible polynomial and \(\boldsymbol{\lambda}_i=(\lambda_{i,1},\ldots,\lambda_{i,\mathrm{len}(\boldsymbol{\lambda}_i)})\) is a partition of \(n_i\) such that \(n=\deg(f_1)n_1+\cdots+\deg(f_e)n_e\) and \(A\) is similar to \[ \mathrm{diag}C(f^{\lambda_{1,1}}_1),\ldots,C(f^{\lambda_{1,\mathrm{len}\boldsymbol{\lambda}_1)}}_1),\ldots,C(f^{\lambda_{e,1}}_e),\ldots,C(f^{\lambda_{e,\mathrm{len}(\boldsymbol{\lambda}_e)}}_e)) \] over \(k\). Here, \(C(f)\in M_{m}(k)\) with \(m=\deg(f)\) is the companion matrix of \(f\in k[X]\). Write \(k_i=k[X]/(f_i)\) and let \(\mathfrak{o}_i\) be the ring of integers of \(k_i\). Put \(S_i=\{w\in\mathcal{V}_k\,:\,\text{ exists }v\in S.\,w\,|\,v\}\) and set \(\mathfrak{o}_{i,S_i}=(\mathfrak{o}_i)_{S_i}\). Then, it is shown that there exist finitely many places \(w_1,\ldots,w_{\ell}\in\mathcal{V}_{k}\setminus S\) and associated rational functions \(W_1,\ldots,W_{\ell}\in\mathbb{Q}(X)\) such that \[ \zeta_{A,\mathfrak{o}_S}(s)=\prod^{\ell}_{u=1}W_{u}(q^{-s}_{w_{u}})\times\prod^{e}_{i=1}\prod^{|\boldsymbol{\lambda}_i|}_{j=1} \zeta_{\mathfrak{o}_{i,S_i}}((\boldsymbol{\lambda}^{\ast}_i)^{-1}(j)\cdot s-j+1), \] where, for a partition \(\boldsymbol{\lambda}=(\lambda_1,\ldots,\lambda_{\mathrm{len}(\boldsymbol{\lambda})})\), \(|\boldsymbol{\lambda}|=\lambda_{1}+\cdots+\lambda_{\mathrm{len}(\boldsymbol{\lambda})}\), \(\boldsymbol{\lambda}^{\ast}\) the dual partition of \(\boldsymbol{\lambda}\) and \(\boldsymbol{\lambda}^{-1}(j)=\min\{i\in\{1,\ldots,\mathrm{len}(\boldsymbol{\lambda})\}:\,j\le \lambda_{1}+\cdots+\lambda_{i}\}\). This shows that \(\zeta_{A,\mathfrak{o}_S}(s)\) admits a meromorphic continuation to the complex plain. Moreover, using the detailed analysis of analytic properties of subobject zeta functions in [\textit{M. du Sautoy} and \textit{F. Grunewald}, Ann. Math. (2) 152, No. 3, 793--833 (2000; Zbl 1006.11051)], we see that the abscissa of convergence \(\alpha_{A,\mathfrak{o}_S}\) of \(\zeta_{A,\mathfrak{o}_S}(s)\) satisfies \(\alpha_{A,\mathfrak{o}_S}=\max_{1\le i\le e}\mathrm{len}(\boldsymbol{\lambda}_i)\) and the multiplicity \(\beta_{A,\mathfrak{o}_S}\) of the pole of \(\zeta_{A,\mathfrak{o}_S}(s)\) at \(\alpha_{A,\mathfrak{o}_S}\) satisfies \(\beta_{A,\mathfrak{o}_S}=\sum_{i\in I}\lambda_{i,\mathrm{len}(\boldsymbol{\lambda}_i)}\) with \(I=\{i\in\{1,\ldots,e\}:\,\mathrm{len}(\boldsymbol{\lambda}_i)=\alpha_{A,\mathfrak{o}_S}\}\). To obtain the explicit expression above, the author first establishes that for local zeta function \(\zeta_{A,\mathfrak{o}_v}(s)\), which has some applications else. For example, one can get a local functional equation for \(\zeta_{A,\mathfrak{o}_v}(s)\) with respect to \(q_v\Leftrightarrow q^{-1}_v\) for almost all \(v\in \mathcal{V}_k\), and show that \(\zeta_{A,\mathfrak{o}_v}(s)\) has a simple pole at zero for almost all \(v\in \mathcal{V}_k\) if and only if there exists \(a\in k\) such that \((A-a1_n)^n=0\). Moreover, for nilpotent \(A\in\mathrm{M}_n(k)\) and \(B\in\mathrm{M}_m(k)\), it is shown that \(\zeta_{A,\mathfrak{o}_v}(s)=\zeta_{B,\mathfrak{o}_v}(s)\) for almost all \(v\in \mathcal{V}_k\) if and only if \(n=m\) and \(A\) and \(B\) are similar.