Cells and modular representations of Hecke algebras (Q2565057)

Let \({\mathcal A}=\mathbb{Z}[q^{1/2},q^{-1/2}]\) with \(q^{1/2}\) an indeterminate, and \(H\) the Hecke algebra of an irreducible Weyl group \(W\) over \(\mathcal A\). Let \(H({\mathcal B})=H\otimes_{\mathcal A}{\mathcal B}\) for any \(\mathcal A\)-algebra \(\mathcal B\). In the paper under review, the author first shows the following result: (A) Assume that \(W\) is not of type \(E_8\). If two left cells of \(W\) afford isomorphic \(H(\mathbb{Q}(q^{1/2}))\)-modules, then the \(W\)-graphs associated to these left cells are also isomorphic. Then the author uses this result to determine the decomposition of an irreducible \(H(\mathbb{Q}(q^{1/2}))\)-module into irreducible \(H(\mathbb{F}_p(q^{1/2}))\)-modules (in the Grothendieck group) under the \(p\)-reduction in the case where \(W\) is not of type \(E_8\), where \(\mathbb{F}_p\) is the prime field of characteristic \(p>0\). The author showed already in his previous paper [J. Algebra 174, No. 2, 553-572 (1995; Zbl 0833.20054)] that the decomposition matrix of \(H(\mathbb{Q}(q^{1/2}))\) with respect to the \(p\)-reduction is the identity matrix when \(p\) is not bad. Thus he needs only to deal with the cases where \(p\) is a bad prime in the present paper. Note that Lusztig has independently proved the result (A) without the restriction that \(W\) is not of type \(E_8\).