A universal construction for moduli spaces of decorated vector bundles over curves (Q1889880): Difference between revisions

This paper gives a moduli space construction which can be adapted to cover a wide variety of situations in which the objects to be parameterized consist of vector bundles together with some extra data. In Schmitt's general setup, the objects are defined by the following data: (i) a rank \(r\) holomorphic bundle \(E\) over an algebraic curve, \(X\), (ii) a representation \(\rho: \text{GL}(r)\rightarrow \text{GL}(V)\), where \(V\) is a fixed vector space, (iii) a line bundle \(M\)\ on \(X\), and (iv) a non-zero homomorphism \(\tau: E_{\rho}\rightarrow M\), where \(E_{\rho}\) is the associated vector bundle determined by \(\rho\). Fixing the topological type of the bundle \(E\), and the representation \(\rho\), the classification problem is then for triples \((E,M,\tau)\). This is obviously a very flexible framework. Specific choices for \(\rho\) produce familiar objects including Higgs bundles, vortex pairs, coherent systems, conic bundles, and framed modules. The construction in this paper uses Geometric Invariant Theory (GIT) and thus depends on an appropriately identified notion of semistability. The author defines semistability very carefully and explores the concept rather thoroughly. The definition is, in general, quite cumbersome; Schmitt identifies conditions under which it simplifies considerably and makes contact with the more familiar notions of slope-stability for vector bundles. The first parts of the paper give the general GIT construction of (coarse) moduli spaces. The paper ends with a study of special cases corresponding to known examples.