A path model for geodesics in Euclidean buildings and its applications to representation theory (Q936142)

The authors give a combinatorial characterization of projections of geodesics in Euclidean buildings to Weyl chambers. The authors apply these results to the representation theory of complex reductive Lie groups and to spherical Hecke rings associated with split nonarchimedean reductive Lie groups. The main application is a generalization of the saturation theorem of \textit{A. Knutson} and \textit{T. Tao} for \(\text{SL}_n\) [J. Am. Math. Soc. 12, No. 4, 1055--1090 (1999; Zbl 0944.05097)] to other complex semisimple Lie groups. Let \(\underline{G}\) be a \(\mathbb{Q}\)-split reductive algebraic group defined over \(\mathbb{Z}\) and \(\underline{G}^\vee\) be its Langlands dual. In the present paper the authors continue their study [Mem. Am. Math. Soc. 896, 83 p. (2008; Zbl 1140.22009)] of the interaction between the representation theory of the group \(G^\vee:=\underline{G}^\vee(\mathbb{C})\) and the geometry of the Bruhat-Tits building associated with the nonarchimedean group \(G=\underline{G}(\mathbb{K})\), where \(\mathbb{K}\) is a complete field with discrete valuation. The authors restrict to the case when \(K\) is a local field, in which case, algebraically speaking, the authors study the relation between the representation ring of the group \(G^\vee\) and the spherical Hecke algebra \(\mathcal{H}_G\) associated with \(G\). In his papers [Invent. Math. 116, No. 1--3, 329--346 (1994; Zbl 0805.17019)], [Ann. Math. (2), 142, No. 3, 499--525 (1995; Zbl 0858.17023)], \textit{P. Littelmann} introduced a path model for representations of complex reductive Lie groups \(G^\vee\). The Littelmann path model gives a method to compute the structure constants of the representation ring of \(G^\vee\) by counting certain piecewise linear paths, called LS paths. In the present paper the authors define a class of piecewise-linear paths in \(\Delta\), a Weyl chamber of the Weyl group \(W\) of \(G^\vee\). These paths are called Hecke paths. The authors prove that a path \(p\) in \(\Delta\) is a Hecke path if and only if \(p\) is the projection into \(\Delta\) of a geodesic segment in the Euclidean (Bruhat-Tits) building \(X\) associated with \(G\). Thus, unlike LS paths which had to be invented, the Hecke paths appear very naturally as projections of geodesic segments. Hecke paths are defined by eliminating one of the axioms for LS paths, therefore each LS path for \(G^\vee\) is a Hecke path for \(G\). Denote by \(R\) the root system of the group \(\underline{G}\). Let \(\alpha_1,\dots,\alpha_l\in R\) be simple roots (corresponding to \(\Delta\) ). Let \(\theta\) be the highest root and define positive integers \(m_1,\dots,m_l\) by \(\theta=\sum_{i=1}^l m_i \alpha_i\). Then the \textsl{saturation factor} \(k_R\) is the least common multiple of the numbers \(m_i,i=1,\dots,l\). Below, \(L\) is the character lattice of a maximal torus in \(G^\vee\) (so that \(\Delta \subset L\otimes \mathbb{R}\)), \(Q(R^\vee)\) is the root lattice of \(R^\vee\). The main result is the following theorem, which, as the authors say, in a weaker form, has been conjectured by S. Kumar: Theorem 1.1 (Saturation theorem). Let \(G^\vee\) and \(L\) be as above. Suppose that \(\alpha,\beta,\gamma\in L\) are dominant characters such that \(\alpha+\beta+\gamma\in Q(R^\vee)\) and there exists \(N\in \mathbb{N}\) so that \((V_{N\alpha}\otimes V_{N\beta}\otimes V_{N\gamma})^{G^\vee}\neq 0\). Then for \(k=k^2_R\) we have: \((V_{k\alpha}\otimes V_{k\beta}\otimes V_{k\gamma})^{G^\vee}\neq 0\). (Here \(V_\lambda\) is the irreducible representation of \(G^\vee\) associated with the dominant weight \(\lambda\) of \(G^\vee\).) As an immediate corollary of Theorem 1.1 the authors obtain a new proof of the saturation theorem of A. Knutson and T. Tao (op. cit.). Corollary 1.2. Suppose that \(R=A_l\), i.e., the semisimple part of \(G^\vee\) is locally isomorphic to \(SL_{l+1}\). Suppose further that \(\alpha,\beta,\gamma\) are dominant characters such that \(\alpha+\beta+\gamma\in Q(R^\vee)\) and that there exists \(N\in \mathbb{N}\) so that \((V_{N\alpha}\otimes V_{N\beta}\otimes V_{N\gamma})^{G^\vee}\neq 0\). Then \((V_\alpha\otimes V_\beta\otimes V_\gamma)^{G^\vee}\neq 0\). The paper also contains many other interesting results concerning: absolute and relative chains, Hecke paths, a compactness theorem, folding via retraction, converting folded triangles in spherical buildings into chains, folding polygons in Euclidean buildings, Littelmann polygons, generalized LS paths, path model for the representation theory of Lie groups, unfolding Littelmann triangles and characterization of folded triangles.