An integral transform for \(p\)-adic symmetric spaces (Q679039)

Let \(K\) be a local field of characteristic zero, and let \(\Omega^{(d+1)}\) be Drinfeld's \(p\)-adic symmetric space which is obtained from \(\mathbb{P}_K^d\) by removing all \(K\)-rational hypersurfaces. \textit{P. Schneider} and \textit{U. Stuhler} [Invent. Math. 105, No. 1, 47-122 (1991; Zbl 0751.14016)] have studied the cohomology of these spaces and in particular determined their de Rham cohomology in representation-theoretic terms. They obtained an (abstract) isomorphism between \(H_{DR}^d (\Omega^{(d+1)},K)\) and \(\Hom({\mathcal S}t,K)\), where \({\mathcal S}t\) is the Steinberg representation of \(G= \text{GL}_{d+1} (K)\). For \(d=1\), this is the space of locally constant functions on \(\mathbb{P}^1(K)\) modulo constants, and the isomorphism can be made explicit in one direction by \textit{P. Schneider}'s residue map [in: Number theory, Proc. Journ. arith., Noordwijkerhout 1983, Lect. Notes Math. 1068, 216-230 (1984; Zbl 0572.14014)]; with the help of Teitelbaum's Poisson kernel [\textit{J. T. Teitelbaum}, Invent. Math. 101, No. 2, 395-410 (1990; Zbl 0731.11065)], an inverse to this residue map can be constructed. The aim of the paper is to generalize these two constructions to \(d\)-forms on \(\Omega^{(d+1)}\). The first main result is the definition of an explicit residue map \(\text{Res}: \Omega_K^d\to \Hom({\mathcal S}t,K)\) which vanishes on exact forms. The authors use the reduction map \(r\) from \(\omega^{(d+1)}\) to the Bruhat-Tits building for \(\text{PGL}_{d+1}(K)\): on the inverse image chamber \(\Delta\) of this building, they associate to any \(d\)-form \(\eta\) an appropriate coefficient of its power series expansion and call it the residue \(\text{Res}_\Delta\eta\) of \(\eta\) on \(\Delta\). Then they show that \(\text{Res}_\eta\) is a harmonic function, i.e. \(\sum \text{Res}_\Delta\eta=0\) if the sum is taken over all chambers in the Bruhat-Tits building that contain a given codimension-one face. For this they need an explicit geometric description of \(r^{-1} (\Delta)\) which eventually allows them to reduce the problem to the one-dimensional case. In the second part of the paper the authors construct an inverse to the residue map. The essential ingredient here is a kernel function \(k(g,q)\) on \(G/P\times \Delta^{(d+1)}\), where \(P\) is a fixed Borel subgroup of \(G\). Elements \(\lambda\) of \(\Hom({\mathcal S}t,K)\) are interpreted as bounded \(p\)-adic distributions on \(G/P\), and the function \(F_\lambda(q):= \int_{G/P} k(g,q) d\lambda(g)\) is shown to be rigid analytic on \(\Omega^{(d+1)}\). For suitable coordinates \(x_0,\dots, x_{d-1}\) on \(\Omega^{(d+1)}\), the \(d\)-form \(\eta_\lambda:= F_\lambda dx_0\wedge\dots \wedge dx_{d-1}\) then satisfies \(\text{Res} (\eta_\lambda)= \lambda\). Technically the most subtle point is the proof that the map \(\lambda\mapsto \eta_\lambda\) is \(G\)-invariant (on the level of forms); for this the authors use results on finite arrangements of hyperplanes.