Abstract. This is a semi-expository article concerning Langlands functoriality and Deligne's conjecture on the special values of L-functions. The emphasis is on symmetric power L-functions associated to a holomorphic cusp form, while appealing to a recent work of Mahnkopf on the special values of automorphic L-functions.
Contents
IntroductionLanglands functoriality principle reduces the study of automorphic forms on the adèlic points of a reductive algebraic group to those of an appropriate general linear group. In particular, every automorphic L-function on an arbitrary reductive group must be one for a suitable GL n . One should therefore be able to reduce the study of of special values of an automorphic L-function to those of a principal L-function of Godement and Jacquet on GL n .
A. RAGHURAM AND FREYDOON SHAHIDIWhile the integral representations of Godement and Jacquet do not seem to admit a cohomological interpretation, there is a recent work of J. Mahnkopf [27] [28] which provides us with such an interpretation for certain Rankin-Selberg type integrals. In particular, modulo a nonvanishing assumption on local archimedean Rankin-Selberg product L-functions for forms on GL n × GL n−1 , he defines a pair of periods, which seem to be in accordance with those of Deligne [9] and Shimura [43]. This work of Mahnkopf is quite remarkable and requires the use of both Rankin-Selberg and Langlands-Shahidi methods in studying the analytic (and arithmetic) properties of L-functions. His work therefore brings in the theory of Eisenstein series to play an important role. In §6 we briefly review this work of Mahnkopf.This article is an attempt to test the philosophy-to study the special values of L-functions while using functoriality-by means of recent cases of functoriality established for symmetric powers of automorphic forms on GL 2 [17] [21]. While a proof of the precise formulae in the conjectures of Deligne [9] still seem to be out of reach, we expect to be able to prove explicit connections between the special values of symmetric power L-functions twisted by Dirichlet characters and those of the original symmetric power L-functions using this work of Mahnkopf. These relations are formulated in this paper as Conjecture 7.1 which also follows from the more general conjectures of Blasius [3] and Panchiskin [33], although the heuristics underlying our conjecture are quite different.A standard assumption made in the study of special values of L-functions is that the representations (to which are attached the L-functions) are cohomological. This is the case in Mahnkopf's work. A global representation being cohomological is entirely determined by the archimedean components. For representations which are symmetric power lifts of a cusp form on GL 2 we have the following fact. Consider a holomorphic cusp form on the upper half plane of weight k. This corresponds to a cuspidal automorphic representation, which is cohomological if k ≥ 2, and any symmetric power lift, if cuspidal, is essentially cohomological. See Theorem 5...