Quadratic base change of \(\theta_{10}\) (Q5939289)

The authors show that quadratic base change is not always well behaved. Specifically, suppose \(k/k_0\) is a quadratic extension of \(p\)-adic fields having odd residue characteristic. The authors define a class of representations of \(\text{GSp}_4(k)\) that are analogous to the representation \(\theta_{10}\) of \(\text{Sp}_4(k)\). Some of these representations are invariant under the action of \(\text{Gal}(k/k_0)\), so it is reasonable to ask whether or not they arise as Shintani lifts of representations of \(\text{GSp}_4(k_0)\). It turns out that they do not, a fact that is proved by comparing character formulas near the identity. Some of the representations of type \(\theta_{10}\) are supercuspidal, so one might expect their \(L\)-functions to be trivial. However, at least some of these \(L\)-functions have two poles. Finally, the authors look at the situation in terms of Langlands parameters, using the functoriality associated to an embedding of \(\text{GSp}_4(\mathbf C)\) into \(\text{GL}_4(\mathbf C)\). Restricting parameters from the Weil group \(W_{k_0}\) of \(k_0\) to \(W_k\) corresponds to a Langlands lift from representations of \(\text{GSp}_4(k_0)\) to representations of \(\text{GSp}_4(k)\). The authors show that some of the representations of type \(\theta_{10}\) arise as Langlands lifts in this sense, and some do not.