Defining the notion of "Galois class of L-functions" as a set of automorphic L-functions belonging to the Selberg class closed under the usual product and the Rankin-Selberg convolution and containing the constant map equal to $1$ and the Riemann zeta function, say $\mathcal{G}(F)$ is the Galois class of L-functions generated by $F$ if it is the smallest Galois class of L-functions containing $F$.

Do(es) the so far proven multiplicity one theorem(s) imply that an element of a Galois class of L-functions is entirely determined by its set of Satake parameters at all but finitely many places? If so, is the group of automorphisms of $\mathcal{G}(F)$ for a given $F$ (hence bijective maps from $\mathcal{G}(F)$ to itself preserving the usual product and RS convolution and sending $s\mapsto 1$ to itself and $\zeta$ to itself) naturally isomorphic to the Galois group $Gal(K_{F}/Q)$ where $K_{F}$ is the compositum of the number fields $K_{p}(F)$ obtained by adjoining to $Q$ the Satake parameters of $F$ at $p$ for all but finitely many $p$?

Would this imply that the group $Aut(\mathcal{M})$ where $\mathcal{M}$ is the maximal Galois class of L-functions is naturally isomorphic to the absolute Galois group of the rationals?

  • 2
    $\begingroup$ I assume that by "automorphic $L$-function" you mean "$L$-function of an automorphic form on $\mathrm{GL}_n$ over $\mathbb{Q}$ for some $n$". If $F$ is not a product of shifted Dirichlet $L$-functions, then I don't think it is known that $\mathcal{G}(F)$ exists. Assuming Langlands functoriality, $\mathcal{G}(F)$ exists of course, hence the multiplicity one theorem applies to its elements (as does to any principal automorphic $L$-function). Also, I don't think that $\mathcal{G}(F)$ has much to do with Galois groups of number fields (cf. my response to mathoverflow.net/questions/50581). $\endgroup$ – GH from MO Jul 11 at 22:52
  • $\begingroup$ But aren't Satake parameters algebraic numbers? $\endgroup$ – Sylvain JULIEN Jul 12 at 5:03
  • $\begingroup$ Also, if I'm not mistaken, for any pair $(k,m)$ of non-negative integers, the Rankin-Selberg convolution $\zeta^{k}\otimes\zeta^{m}$ exists and equals $\zeta^{km}$. So that there exists at least one Galois class of L-functions unconditionally, namely $\mathcal{G}(\zeta)$. $\endgroup$ – Sylvain JULIEN Jul 12 at 6:00
  • 3
    $\begingroup$ First, Satake parameters are not algebraic numbers in general. This is what I told you 8.5 years ago. Second, if $F$ is a product of shifted Dirichlet $L$-functions (e.g. when $F=\zeta$), then of course $\mathcal{G}(F)$ exists, but otherwise not, because Rankin-Selberg $L$-functions are only known to be automorphic for small degrees (cf. known cases of Langlands functoriality). $\endgroup$ – GH from MO Jul 12 at 7:41
  • $\begingroup$ Ok. Some Satake parameters are transcendental, so $Aut(\mathcal{M})$ can't be isomorphic to $G_{\mathbb{Q}}$. But don't the formal properties of the two operations and the fact that $\mathcal{M}$ is the maximal Galois class of L-functions imply that its automorphism group is isomorphic to the absolute Galois group of some field $\mathbb{K}$ such as $\mathbb{R}$? $\endgroup$ – Sylvain JULIEN Jul 14 at 10:02

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.