An algorithm for computing invariants of linear actions of algebraic groups up to a given degree. (Q1401909)

One of the basic tasks in invariant theory is the computation of homogeneous invariants of some given degree. The paper under review proposes a new algorithm for this purpose. The input data of this algorithm consists of an algebraic group \(G\) (which need not be reductive), given by polynomial equations, a linear action of \(G\) by a morphism, and a degree \(d\). The output is a basis of the space of all homogeneous \(G\)-invariants of degree \(d\). There exists a standard algorithm for the same purpose, which is acknowledged in the present paper. The strength of the paper lies in the performance of the new algorithm: It is faster for all examples given in the paper. The reviewer has implemented both algorithms in Magma and tested a huge number of further examples. The result is that the new algorithm is always faster, usually significantly. However, it remains a mystery why this is true. The author cannot explain the better performance of his algorithm, and neither can the reviewer. Bayer's algorithm uses Gröbner basis methods and is not as straightforward as the standard algorithm.