Complete intersection quiver settings with one dimensional vertices. (Q368646)

The author determines the quivers \(Q\) such that the corresponding affine toric quiver variety \(\mathrm{iss}(Q)\) is a complete intersection. Recall that \(\mathrm{iss}(Q)\) parameterizes the isomorphism classes of the \((1,\ldots,1)\)-dimensional semisimple representations of \(Q\). It is constructed as the affine quotient of the representation space by the action of the base change group (which is a torus for this dimension vector). \textit{R. Bocklandt} [J. Algebra 253, No. 2, 296-313 (2002; Zbl 1041.16010)] determined the quivers and dimension vectors \((Q,\alpha)\) for which the affine quotient \(\mathrm{iss}(Q,\alpha)\) of the representation space by the base change group is an affine space. The present author shows that for a strongly connected prime quiver \(Q\), the affine variety \(\mathrm{iss}(Q)\) is a complete intersection if and only if \(Q\) can be reduced to a one-vertex quiver by the reduction steps of Bocklandt and one more reduction step introduced here. This class of quivers is characterized also by some forbidden descendants (a notion similar to graph minors). Along the way it is shown that the ideal of relations of an affine toric quiver variety is minimally generated by a finite set of binomial relations, such that all but one of them is quadratic, and one of the terms of the remaining binomial is also quadratic.