Matrix orbit closures (Q1671270)

Consider the vector space \(V\) of \(r\times n\) matrices, acted upon by the group \(G=\mathrm{GL}_r(C) \times (C^{\times})^n\). Most points of \(V\) correspond to ordered \(n\)-tuples of points in the projective space \(P^{r-1}\). Such a tuple determines a matroid, a so-called representable matroid, via the dimensions of the spans of various subsets of the tuple. That matroid is invariant under the \(G\) action. The paper under review is one of the many works that explore the relation between the matroid and certain aspects of the geometry of the orbit. In particular, consider the class of the orbit closure in the \(G\)-equivariant \(K\)-theory of the vector space \(V\). The authors conjecture (Conjecture 5.1) that the matroid determines this class. Besides this conjecture the paper has two main results. In the first one the authors prove that certain coefficients of the \(K\)-class of the orbit closure are indeed determined by the matroid. In this result, the combinatorics of hook-shaped partitions/Schur functions play a role. The other main result is the description of the ideal of the orbit closure using only the matroid -- up to radical. If the ideal was reduced this would prove the conjecture, and it indeed does prove it in special cases when the ideal is reduced. For example, rank 2 or corank 2 uniform matroids satisfy this property.