Matching complexes, bounded degree graph complexes, and weight spaces of \(\mathrm{GL}_n\)-complexes (Q5942786): Difference between revisions

Let \(\gamma=(\gamma_1,\dots,\gamma_n)\) be a sequence of non-negative integers. \textit{V. Reiner} and \textit{J. Roberts} [J. Algebr. Comb. 11, No. 2, 135--154 (2000; Zbl 1012.13005)] introduced the bounded degree graph complex \(\Delta_\gamma\) as the abstract simplicial complex (on the set of possible edges) of all graphs on \(n\) distinguished nodes where the degree of the \(i\)th node is bounded by \(\gamma_i\) for all~\(i\). There is a natural action of the Young subgroup \(\Sigma(\gamma)\) on \(\Delta_\gamma\) which permutes the nodes of the same maximal degree. For any fixed field the homology vector spaces of~\(\Delta_\gamma\) become \(\Sigma(\gamma)\)-modules. As a special case one obtains the matching complex \(M_n=\Delta_{(1,1,\dots,1)}\). \textit{S. Bouc} [J. Algebra 150, No.~1, 158--186 (1992; Zbl 0786.55005)] computed the equivariant homology of~\(M_n\). Reiner and Roberts [loc. cit.] generalized Bouc's result to arbitrary~\(\Delta_\gamma\). \textit{T. Józefiak, P. Pragacz} and \textit{J. Weyman} [Astérisque 87/88, 109--189 (1981; Zbl 0488.14012)] computed the equivariant homology of a certain chain complex which is defined in entirely algebraic terms, more precisely, as the Koszul complex of a certain ideal in the symmetric algebra over a finite dimensional vector space. By means of explicit constructions the authors of the paper under review prove that the three results mentioned above are essentially the same. A key step in the argument is a result which shows how to ``lift'' Bouc's special result on the matching complex to the bounded degree graph complex. The authors apply this ``lifting lemma'' to obtain new results on the homology of bounded degree hypergraph complexes and higher-dimensional chessboard complexes.