On the transfer between the Dickson algebras as modules over the Steenrod algebra (Q260076)

This paper is a contribution to the study of the Dickson-Mùi algebras and related problems. To state the main results of this article, recall some basic definitions. The letter \(p\) denotes a prime number and let \(\mathbb{F}_{p}\) be the finite field of \(p\) elements. Let \(W_{n}\) denote an elementary \(p\)-group of rank \(n\geq 1\), then the general linear group \(\mathrm{GL}(W_{n})\) acts naturally on \(W_{n}\) and hence on its cohomology \(\mathrm{H}^{*}(W_{n};\mathbb{F}_{p})\). The Dickson-Mùi algebra \(D_{n}\) is defined as the ring of invariants \(\mathrm{H}^{*}(W_{n};\mathbb{F}_{p})^{\mathrm{GL}(W_{n})}\). As the action of \(\mathrm{GL}(W_{n})\) commutes with the action of the Steenrod algebra \(\mathcal{A}_{p}\) on the cohomology of \(W_{n}\), we have that \(D_{n}\) is a graded module over \(\mathcal{A}_{p}\), which splits as the direct sum of its augmentation ideal \(\bar{D}_{n}\) and \(\mathbb{F}_{p}\), given by the unit of \(D_{n}\). In the paper [Math. Ann. 353, No. 3, 827--866 (2012; Zbl 1248.55007)], \textit{N. H. V. Hung} shows that there exist special \(\mathcal{A}_{p}\)-morphisms \[ \mathrm{res}_{l,m}:\bar{D}_{l}\longrightarrow \bar{D}_{m},\quad\mathrm{tr}_{m,n}:\bar{D}_{m}\longrightarrow \bar{D}_{n}, \] such that the set \[ \{ \mathrm{tr}_{m,n}\mathrm{res}_{l,m}\mid 1\leq m\leq \min\{l,n\}\} \] forms an \(\mathbb{F}_{p}\)-basis of \(\mathrm{Hom}_{\mathcal{A}_{p}}(\bar{D}_{l},\bar{D}_{n} )\). The present paper focuses on the study of the transfers \(\mathrm{tr}_{m,n}\) in characteristic \(2\). In this case, \[ D_{n}\cong\mathbb{F}_{2}[Q_{n,0},Q_{n,1},\dots,Q_{n,n-1}], \] where the symbols \(Q_{n,i}\) denote the Dickson invariants. In particular, \(D_{1}\) is the polynomial algebra on one variable of degree \(1\), which is denoted by \(x\) in this paper. The first result of this article is the complete determination of \(\mathrm{tr}_{1,n}\): \[ \mathrm{tr}_{1,n}(x^{k})=\sum_{\sum_{i=0}^{n-1}(2^{n}-2^{i})s_{i}=k}^{}\frac{(-1+\sum_{i=0}^{n-1}s_{i})!}{(s_{0}-1)!s_{1}!\dots s_{n-1}!}Q_{n,0}^{s_{0}}Q_{n,1}^{s_{1}}\dots Q_{n,n-1}^{s_{n-1}}, \] where \(s_{0}\geq 1\) and \(k>0\). This is obtained by induction, using the recursive formulation of the Dickson invariants: \[ Q_{n,i}=Q_{n-1,i-1}^{2}+V_{n}Q_{n-1,i}, \] where \(V_{n}\) is the Mùi invariant under the Sylow \(2\)-subgroup \(T_{n}\) of \(\mathrm{GL}(n,\mathbb{F}_{2})\) consisting of all upper triangular \(n\times n\)-matrices with \(1\) on the diagonal. In the second part of the paper, the authors introduce the notion of multilinear and alternating invariants of \(D_{m}\), for all \(m\geq 1\), which arises naturally from that of polynomials. Then, they show that \(\mathrm{tr}_{m,n}\) vanishes on all invariants of \(D_{m}\) that are multilinear and alternating. Other special types of invariants of \(D_{m}\) on which \(\mathrm{tr}_{m,n}\) vanishes are also studied. Moreover, the computations of \(\mathrm{tr}_{m,m+1}\) on the powers of multilinear and alternating invariants of \(D_{m}\) are described as well. Even though the transfer \(\mathrm{tr}_{m,n}\) is a morphism over the Steenrod algebra, all the computations in this paper are carried out using only standard arguments concerning the Dickson invariants.