On the ring of constants of a diagonalizable higher derivation (Q1194298): Difference between revisions

Let \(k\) be a field of positive characteristic \(p>0\), let \(q=p^ e\), and let \(\mu_ q\) denote the group scheme of \(q\)-th roots of unity over \(k\). Suppose the \(\mu_ q\) acts linearly on the affine \(n\)-space \(\mathbb{A}^ n\) over \(k\). The authors compute the ring of invariants of such an action, show that it is always Cohen-Macaulay, compute its canonical module, and give a numerical criterion for it to be Gorenstein. Their work begins by observing that any such action is given by a diagonalizable higher derivation on the polynomial ring \(R=k[X_ 1,\dots,X_ n]\). A higher derivation is a sequence of \(k\)-linear endomorphisms \(D=\{d_ 0=1,\dots,d_{m-1}\}\) of \(R\) such that \(d_ j(ab)=\sum d_ i(a)d_{j-i}(b)\); \(D\) is diagonalizable if \(d_ i(X_ j)=\lambda_{ji}X_ j\) for all \(i\) and \(j\), with \(\lambda\) in \(k\). The ring of invariants above corresponds to \(S=\{a\in R| d_ i(a)=0\), \(i>0\}\). The authors give an explicit description of \(S\) in terms of the \(\lambda_{ij}\) and show that \(S\) is a direct summand of \(R\) as a graded \(S\)-module (and hence Cohen-Macaulay). They then give an explicit description of the canonical module for \(S\), also in terms of the \(\lambda\)'s, from which they derive an explicit condition for \(S\) to be Gorenstein.