Invariant rings and quasiaffine quotients (Q1395408)

Counterexamples to Hilbert's fourteenth problem show that for an arbitrary algebraic group action on an affine variety the ring of invariant functions need not be finitely generated (i.e. isomorphic to the ring of functions on an affine variety). In the paper under review it is proved that the ring of invariant functions is always isomorphic to the ring of functions on a quasi-affine variety. The proof is based on the theory of Krull rings and on the construction of the ideal transform. This result implies that for any irreducible, reduced, normal \(k\)-variety \(V\) and any subgroup \(G\subset \text{Aut}(V)\) there exists a quasi-affine \(k\)-variety \(Z\) and a rational map \(\pi: V\to Z\) such that: (1) \(\pi^*\) induces an inclusion \(k[Z]\subset k[V]\); (2) \(\pi^*(k[Z])=k[V]^G\); (3) for any affine \(k\)-variety \(W\) and any \(G\)-invariant morphism \(f: V\to W\) there exists a morphism \(F: Z\to W\) such that \(F\circ \pi\) is a morphism and \(f=F\circ \pi\). As an example, the corresponding variety \(Z\) is described for the action of the additive group \(G_a\) in the 5-dimensional vector space \(k^5\) with a non-finitely generated algebra of invariants, which was constructed by \textit{D. Daigle} and \textit{G. Freudenburg} [J. Algebra 221, 528-535 (1999; Zbl 0963.13024)].