Framed moduli spaces and tuples of operators (Q376162)

The paper under review studies the problem of classifying tuples of linear endomorphisms and linear functions on a finite dimensional vector space up to base change, which can be reinterpreted in the language of quivers, as classifying representations on finite dimensional vector spaces of the two-vertex quiver with as many loops on the first vertex as linear endomorphisms and as many arrows between the two vertices as the number of linear functions. Such representations can be reinterpreted as framed representations in the sense of [\textit{M. Reineke}, J. Algebra 320, No. 1, 94--115 (2008; Zbl 1153.14033)], where it was proved, for quivers without oriented loops, that the quotient of the stable representations up to isomorphism (i.e. up to base change of the linear maps) is isomorphic to a Grassmannian of subrepresentations of a certain injective representation of the quiver. The general case does not provide a projective quotient, and the author studies the fibers of this quotient in [\textit{S. N. Fedotov}, Trans. Am. Math. Soc. 365, No. 8, 4153--4179 (2013; Zbl 1277.14010)]. It remains to describe a trivializing covering for the quotient map, which is the problem adressed in this article. By generalizing Reineke's construction, the author associates a finite \textit{skeleton} to each stable representation, and each skeleton carries an open subset of the space parametrizing representations. As there are a finite number of possible skeletons, we find a finite open covering of the parametrizing space. The trivialization comes from the fact that each one of these open pieces defines a normal form of the representation, hence the open pieces are isomorphic to affine spaces. Besides, it is given a criterium to determine whether 2 representations are isomorphic, this is when they have a skeleton in common and the normal form associated to them coincides. The paper finishes with a collection of examples where all the computation are explicitly described.