A formula for the cohomology and \(K\)-class of a regular Hessenberg variety (Q2301887): Difference between revisions

Let $G=\mathrm{GL}_n(\mathbb{K})$ be the general linear group over an algebraically closed field $\mathbb{K}$. Let $G/B$ be the full flag variety consisting of complete flags \[ F_\bullet:\quad \{0\}= F_0\subset F_1\subset \cdots \subset F_n=\mathbb{K}^n \] of subspaces of $\mathbb{K}^n$ so that $\dim F_i=i$. Fix a Hessenberg function $h:\{1, \dots, n\} \to \{1, \dots, n\}$ which is, by definition, a non-decreasing function such that $h(i) \geq i$ for all $i$. Fix a linear operator $X:\mathbb{K}^n \to \mathbb{K}^n$. Recall that the corresponding Hessenberg variety is defined by: \[ \mathcal{Y}_{X, h}:= \{F_\bullet: X(F_i)\subset F_{h(i)} \text{ for all } i\}\subset G/B. \] In the paper under review, the authors give an explicit expression for the class of $ \mathcal{Y}_{X, h}$ in the $K$-theory $K(G/B)$ as well as in the cohomology $H^*(G/B)$ in terms of certain Grothendieck (resp. Schubert) polynomial if $X$ is a regular operator. In fact, they generalize the result for any $X$ such that $ \mathcal{Y}_{X, h}$ has the expected dimension $\sum_i(h(i)-i)$. We recall that a different formula for the cohomology class of $ \mathcal{Y}_{X, h}$ was given by \textit{D. Anderson} and the second author [J. Algebra 323, No. 10, 2605--2623 (2010; Zbl 1218.14041)] using degeneracy arguments. Also, a different formula for the $K$-theory class of $ \mathcal{Y}_{X, h}$ was given by \textit{H. Abe}, \textit{N. Fujita} and \textit{H. Zeng} [``Geometry of regular Hessenberg varieties'', Transform. Groups (to appear; \url{doi:10.1007/s00031-020-09554-8})]. But the formulae in this paper differ from the ones given by Abe-Fujita-Zeng and Anderson-the second author viewed as polynomials (but of course not as $K$-theory or cohomology classes). The authors also reprove that $ \mathcal{Y}_{X, h}$ is Cohen-Macaulay (hence equi-dimensional).