On nilpotent commuting varieties of \(r\)-tuples in the Witt algebra (Q2349941)

Let \({\mathfrak g}\) be a Lie algebra over an algebraically closed field \(k\) and \({\mathcal V}\) be a closed subvariety of \({\mathfrak g}\). Due in part to connections with the cohomology of Frobenius kernels, there has been recent interest in studying \(r\)-fold commuting varieties \({\mathcal C}_r({\mathcal V})\) for a positive integer \(r \geq 2\), where \({\mathcal C}_r({\mathcal V}) := \{(v_1,v_2,\dots, v_r) \in {\mathcal V}^r ~|~ [v_i,v_j] = 0 ~ \forall ~ 1 \leq i \leq j \leq r\}\). See for example work of \textit{N. Ngo} [J. Pure Appl. Algebra 218, No. 8, 1400--1417 (2014; Zbl 1298.20056)] and \textit{N. Ngo} and \textit{K. Šivic} [Linear Algebra Appl. 452, 237--262 (2014; Zbl 1291.15045)] for the special linear Lie algebra. Here the author focuses on the case that \({\mathfrak g}\) is the Witt Lie algebra \(W_1\), under the assumption that \(k\) has prime characteristic \(p > 3\). The main goal is to investigate the commuting variety \({\mathcal C}_r({\mathcal N})\), where \({\mathcal N}\) denotes the nilpotent cone of \({\mathfrak g}\), and extend earlier work of the author with \textit{H. Chang} [J. Pure Appl. Algebra 218, 1783--1791 (2014; Zbl 1364.17007)] for the case \(r = 2\). Note that in this situation the nilpotent cone \({\mathcal N}\) happens to agree with the restricted nilpotent cone \({\mathcal N}_p := \{x \in {\mathfrak g}~|~ x^{[p]} = 0\}\). The Witt algebra \({\mathfrak g}\) admits a \({\mathbb Z}\)-grading with \(p\) non-zero components in degrees \(-1\) through \(p-2\). There is also an associated filtration \({\mathfrak g} \supset {\mathfrak g}_0 \supset {\mathfrak g}_1 \supset \cdots \supset {\mathfrak g}_{p-2} \supset 0\). The author begins by determining the varieties \({\mathcal C}_r({\mathfrak g}_i)\) for each \(1 \leq i \leq \frac{p-3}{2}\). For \(i = \frac{p-3}{2}\), this variety is shown to be irreducible (and the dimension is determined). For smaller \(i\), the variety is reducible and the irreducible components are given. An interesting transition occurs for \(r > 2\). For \(1 \leq i \leq \frac{p-5}{2}\), \({\mathcal C}_2({\mathfrak g}_i)\) is equidimensional of dimension \(p\), but \({\mathcal C}_r({\mathfrak g}_i)\) is not equidimensional for \(r \geq 3\). It follows that the latter varieties are neither normal nor Cohen-Macaulay. A key tool in the computation is the classification of the nilpotent orbits in \({\mathfrak g}\) under the action of its automorphism group that was obtained in previous work of the the first author with \textit{B. Shu} [Commun. Algebra 39, No. 9, 3232--3241 (2011; Zbl 1256.17002)]. The author then makes use of the computations of the \({\mathcal C}_r({\mathfrak g}_i)\) to determine \({\mathcal C}_r({\mathcal N})\). When \(r = 2\), in the aforementioned work with Chang, the author had shown that the variety was equidimensional of dimension \(p\). For \(r \geq 3\), it is shown that \({\mathcal C}_r({\mathcal N})\) has \(\frac{p-1}{2}\) irreducible components whose dimensions do not agree. As before, it follows that \({\mathcal C}_r({\mathcal N})\) is neither normal nor Cohen-Macaulay.