Isolation of the cuspidal spectrum: the function field case (Q2154752)

The trace formula has a spectral decomposition with contributions from the continuous spectrum as well as the discrete spectrum. For applications, it is important to single out the contribution from the discrete spectrum, and further to isolate the contribution from a finite set of representations (if not a single representation). A recent breakthrough is a method by Beuzart-Plessis et al. that uses certain multipliers to isolate the cuspidal spectrum from the trace formula. This paper develops a similar method for trace formulas over a function field. The situation is slightly simpler over a function field. The authors can use a Hecke function as the multiplier so that the contribution from continuous spectrum is killed. By Harder's result on finiteness of the cuspidal spectrum [\textit{G. Harder}, Ann. Math. (2) 100, 249--306 (1974; Zbl 0309.14041)], it is possible to construct a further multiplier to isolate the contribution to the near equivalent class of a single cuspidal representation \(\pi\). Because the Hecke function used also kills off the contribution from the CAP representations, the final result is that: if \(\pi\) is a cuspidal representation that is not CAP, there is a Hecke function \(\mu\) such that the projection through \(\mu\) restricts the spectral contribution to those from the \(\pi\)-nearly isotropic subspace.