Vector spaces of matrices of low rank (Q1110615)

The authors consider the problem of classifying the vector spaces of matrices, all of whose elements have rank at most a given number. This problem is roughly equivalent to the problem of classifying certain torsion-free sheaves on projective spaces. The authors solve this problem in case the sheaf in question has first Chern class equal to 1. In addition, the free resolutions are used to verify some properties of the low rank examples and to produce an abundance of examples of somewhat higher rank, connected with the theory of curves.