Geometrically nilpotent subvarieties (Q1696768)

Let \(X\) be a variety over a finite field \(F\) and \(T:X\to X\) a map defined over \(F.\) For a given subvariety of \(X\), we say that \(Y\) is nilpotent with respect to \(T\) if for some point \(P\in X\) and some \(k\geq1,\) we have \(T^{(k)}(Y)=\{P\}.\) On the other hand, we say that \(Y\) is geometrically nilpotent with respect to \(T\) if there is \(P\in X\) such that every geometric point of \(Y\) (i.e., a \(\text{Spec}(F)\)- morphism \(\text{Spec}(F^{\text{alg}})\to X\)) is mapped to \(P\) by a sufficiently high iteration of \(T.\) Clearly every nilpotent variety is geometrically nilpotent. The paper under review shows concrete examples where the converse does not hold, i.e., pairs \((Y,T)\) such that \(Y\) is geometrically nilpotent with respect to \(T\) but not nilpotent. These examples raise several intriguing questions which are posted at the end of the article.