Twisted zeta functions of quaternion orders (Q897527)

This paper builds on work of \textit{C. J. Bushnell} and \textit{I. Reiner} [J. Reine Angew. Math. 327, 156--183 (1981; Zbl 0455.12010); ibid. 329, 88--124 (1981; Zbl 0488.12011)], in which they define \(L\)-functions for finite-dimensional semisimple algebras over \(\mathbb Q\), and discuss their analytic continuation, functional equations and behaviour at \(s=1\). This theory applies to quaternion division algebras over number fields, and the paper under review provides alternative proofs of the analytic continuation and functional equations for the \(L\)-functions of these algebras, and also presents additional analogues with abelian \(L\)-functions of number fields. Let \(F\) be a totally real algebraic number field, \(\mathcal O(F)\) its ring of integers, \(a,b\) totally positive elements of \(F\), \(A\) the quaternion algebra \((\frac{-a,-b}{F})\) and \(\Lambda=\Lambda(F)\) a maximal \(\mathcal O(F)\)-order in \(A\). Let \(K\) be a totally real abelian Galois extension of \(F\) with Galois group \(G\) and \(\hat{G}\) the character group of \(G\). The twist of the zeta function of \(\Lambda(F)\) by a character \(\chi\in\hat{G}\) is \(L_{\Lambda(F)}(s,\chi)=\sum_{I}\frac{\chi(\operatorname{nrd}(I))}{(\Lambda:I)^s}\), where \(I\) is a nonzero left ideal of \(\Lambda\) and \(\operatorname{nrd}(I)\) denotes the reduced norm of \(I\). (\(L_{\Lambda(F)}(s,\chi)\) depends on \(A\) and \(\chi\) but not on the choice of \(\Lambda\).) When \(\chi\) is the trivial character, this reduces to the zeta function of \(\Lambda\): \(\zeta_{\Lambda(F)}(s)=\sum_{I}\frac{1}{(\Lambda:I)^s}\). In Theorem 4.1, the author gives simplified proofs of, and more explicit formulas for, an Euler product expansion, analytic continuation and a functional equation for \(L_{\Lambda(F)}(s,\chi)\). The next results establish new connections and analogues with \(L\)-functions and zeta functions for number fields. Let \(\mathcal O(K)\) be the ring of integers of \(K\), with \(\Lambda(K)\) a maximal \(\mathcal O(K)\)-order in \(A(K)=(\frac{-a,-b}{K})\). Theorem 6.1 shows that if \(K/F\) is a Galois extension of totally real number fields of odd degree, then the quotient \(\zeta_{\Lambda(K)}(s)/\zeta_{\Lambda(F)}(s)\) is entire. (This result is an analogue of the Aramata-Brauer Theorem for number fields, which is used in the proof.) The author observes that this theorem holds for even degree provided that each prime ideal \(\mathfrak p\) of \(\mathcal O(F)\) which ramifies in \(\Lambda(F)\) has a prime ideal \(\mathfrak P\) of \(\mathcal O(K)\) above it that ramifies in \(\Lambda(K)\). He also provides an example which shows that the odd degree hypothesis cannot be entirely removed. Corollary 7.2 gives a factorization for \(\zeta_{\Lambda(K)}(s)\) when \(K/F\) is an abelian Galois extension of totally real number fields of odd degree: \(\zeta_{\Lambda(K)}(s)=\prod_{\chi\in\hat{G}}L_{\Lambda(F)}(s,\chi)\). This follows from a more general result for arbitrary degree, which includes a finite number of additional Euler factors which may appear in even degree (Theorem 7.1).