Quivers of type \(\mathbb D_n\), semi-invariants and complete intersections (Q448390)

From MaRDI portal
scientific article
Language Label Description Also known as
English
Quivers of type \(\mathbb D_n\), semi-invariants and complete intersections
scientific article

    Statements

    Quivers of type \(\mathbb D_n\), semi-invariants and complete intersections (English)
    0 references
    0 references
    6 September 2012
    0 references
    Let \(\mathbb{K}\) be an algebraically closed field, and let \(Q\) be a finite quiver with \(n\) vertices\(.\) It is well known that to each dimension vector \(\mathbf{d=}\left( d_{1},\dots,d_{n}\right) \mathbf{\in}\mathbb{N}^{n}\) one has the \(\mathbb{K}\)-vector space of representations rep\(\left( Q,\mathbf{d}\right) ,\) and GL\(\left( \mathbf{d}\right) :=\prod\)GL\(\left( d_{i},\mathbb{K}\right) \) acts on rep\(\left( Q,\mathbf{d}\right) \) in a natural way. If rep\(\left( Q,\mathbf{d}\right) \) contains an open orbit, say GL\(\left( \mathbf{d}\right) \cdot T,\) then rep\(\left( Q,\mathbf{d}\right) \) has only finitely many orbits, exactly one of which is open; furthermore there is a finite set \(f_{1},\dots,f_{s}\) of irreducible monic polynomials (in \(\mathbb{K}\left[ \text{rep}\left( Q,\mathbf{d}\right) \right] \)) whose zero sets are the irreducible components of codimension 1 of rep\(\left( Q,\mathbf{d}\right) \backslash\)GL\(\left( \mathbf{d}\right) \cdot T.\) Let \(\mathcal{Z}_{Q,\mathbf{d}}\) be the closed subvariety of rep\(\left( Q,\mathbf{d}\right) \) consisting the common zeros of \(f_{1},\dots,f_{s}.\) If \(\mathcal{Z}_{Q,\mathbf{d}}\) has codimension \(s\) then \(\mathcal{Z} _{Q,\mathbf{d}}\) is said to be a (set theoretic) complete intersection. Here, the focus is on finding necessary and sufficient conditions for \(\mathcal{Z}_{Q,\mathbf{d}}\) to be a complete intersection in the case where \(Q\) is a connected quiver of type \(\mathbb{D}_{n}\) (i.e., the underlying Dynkin diagram is of type \(\mathbb{D}_{n}\)). In this situation (as well as any other where \(Q\) is a disjoint union of Dynkin diagrams of type \(\mathbb{A} ,\mathbb{D}\), or \(\mathbb{E}),\) rep\(\left( Q,\mathbf{d}\right) \) does contain an open orbit GL\(\left( \mathbf{d}\right) \cdot T.\) For the case under consideration, \(\mathbf{d}\) is assumed to be sincere, i.e., \(T\left( k\right) \neq0\) for all vertices \(k.\) To describe the necessary and sufficient conditions required, the author introduces the notion of a folded rectangle. Loosely, a folded rectangle is a collection of vertices of the Auslander-Reiten quiver with a special vertex: a ``bent down corner''. A folded rectangle which satisfies extra conditions which depend on the representation \(T\) above is called a folded rectangle suitable for \(T\). The main result is that \(\mathcal{Z}_{Q,\mathbf{d}}\) is a complete intersection if and only if there is no folded rectangle suitable for \(T\), when \(T\) is the representative of the open orbit of a representation space given by a sincere dimension vector. As an application of the result above, a result of Reidtmann and Zwara is refined. We can decompose the \(T\) above as a direct sum of indecomposable representations \(T_{1}^{\lambda_{1}}\oplus\cdots\oplus T_{r}^{\lambda_{r}}.\) Previously, Reidtmann and Zwara proved that \(\mathcal{Z}_{Q,\mathbf{d}}\) is a complete intersection if (but not only if) \(\lambda_{i}\geq2\) for all \(i\). (Indeed, one of the conditions for \(T\) to have a suitable folded rectangle is for \(\lambda_{i}=1\) for some \(i\).) It is shown that if \(\mathcal{Z} _{Q,\mathbf{d}}\) is not a complete intersection, then there is a unique \(\lambda_{i}\) with \(\lambda_{i}=1.\) Also, \(\mathcal{Z}_{Q,\mathbf{d+e}_{i}}\) is a complete intersection but \(\mathcal{Z}_{Q,\mathbf{d+e}_{j}}\) is not, where \(\mathbf{e}_{i},\mathbf{e}_{j}\in\mathbb{N}^{n}\) are ``standard basis vectors'' and \(i\neq j.\)
    0 references
    0 references
    0 references
    quivers
    0 references
    quiver representations
    0 references
    complete intersections
    0 references
    invariants
    0 references
    0 references