A note on rigidity and triangulability of a derivation (Q2453703)

Let \(A\) be an integral domain containing \(\mathbb{Q}\) and \(A^{[n]}=A[X_1,\dots, X_n]=B\). An \(A\)-derivation \(D:B\to B\) is locally nilpotent if for each \(x\in B\), \(D^s(x)=0\) for some \(s>0\). For such a\(D\), its rank is defined as the smallest \(r\) such that \(A[X_1,\dots, X_{n-r}]\subset \mathrm{ker} D\) (after a change of variables). Such a derivation is triangulable if for a suitable change of variables, one has \(D(X_i)\in A[X_1,\dots, X_{i-1}]\) for \(1\leq i\leq n\). A derivation \(D\) of rank \(r\) is rigid, if for any two choices of variables \(X_i,X_i'\) with \(A_1=A[X_1,\dots, X_{n-r}]\) and \(A_2=A[X_1',\dots,X_{n-r}']\) both contained in \(\mathrm{ker} D\), then \(A_1=A_2\). \textit{D. Daigle} proved that when \(A\) is a field and \(n=3\), all locally nilpotent derivations are rigid [J. Pure Appl. Algebra 113, No. 3, 297--305 (1996; Zbl 0874.13021)]. The authors prove in the same vein for an arbitrary \(A\) as above, if \(D\) is such a derivation and \(D\) has the same rank when localized at the fraction field of \(A\) and rigid after localization, then \(D\) itself is rigid. A similar generalization of a result of Daigle on triangulability is also proved.