On 81 symplectic resolutions of a 4-dimensional quotient by a group of order 32 (Q2359438)

This paper provides an explicit construction of \(81\) symplectic resolutions of a \(4\)-dimensional quotient singularity obtained by an action of a certain symplectic group \(G\) of order \(32\) and also a construction of a new Kummer-type symplectic \(4\)-fold. For a complex vector space \(V\) with a symplectic form \(\omega\), denote by \(\mathrm{Sp}(V)\) its group of symplectomorphisms. An element \(A\in\mathrm{Sp}(V)\) is called a symplectic reflection if its fixed point set is of codimension \(2\). The paper studies the case when \(V=\mathbb{C}^4\), \(\omega=dx_1\wedge dx_3+dx_2\wedge dx_4\) in canonical coordinates, and the group \(G\) is generated by \(5\) symplectic reflections given by explicit \(4\times 4\) matrices (see Section~2.C of the paper). This group \(G\) has order \(32\) and is conjugate in \(\mathrm{Sp}(V)\) to the group generated by Dirac gamma matrices. So, the quotient singularity in question is \(Y=V/G\). A resolution \(\varphi: X\to Y\) is called symplectic, if \(X\) admits a symplectic form. The fact that \(Y\) has symplectic resolutions and that there are exactly \(81\) of them was proven earlier by \textit{G. Bellamy} and \textit{T. Schedler} [Math. Z. 273, No. 3--4, 753--769 (2013; Zbl 1271.16029)]. The new contribution of the article under consideration is an explicit construction of these resolutions. The first main result of the paper reads as follows. Let \(\mathbb{T}\) be a \(5\)-dimensional algebraic torus with coordinates \(t_i\), \(i=0,\dots,4\) associated to five classes of symplectic reflections generating \(G\). Let \(\mathcal{R}\) be a \(\mathbb{C}\)-subalgebra generated in \(\mathbb{C}[V]\otimes\mathbb{C}[\mathbb{T}]\) by \(t_{i}^{-2}\) for \(i=0,\dots,4\) and \(\phi_{ij}t_it_j\) for \(0\leq i<j\leq 4\), where \(\phi_{ij}\) are certain eigenfunctions of the action of the abelianization \(\mathrm{Ab}(G)\) on \(\mathbb{C}[V]^{[G,G]}\). Note that \(\mathrm{Ab}(G)\) can be identified with the class group \(\mathrm{Cl}(Y)\) and \(\mathbb{C}[V]^{[G,G]}\) with the Cox (or total coordinate) ring of \(Y\). Then there exist \(81\) GIT quotients of \(\mathrm{Spec}\mathcal{R}\) with respect to the action of an algebraic torus associated to a certain grading of \(\mathcal{R}\), and these quotients yield all symplectic resolutions of \(Y\). There is one distinguished resolution, and all other are obtained from it by flops. An interesting feature of the approach undertaken by the authors is the extensive use of the Cox ring. For the exact relation between the Cox ring of resolution of \(Y\) and the algebra \(\mathcal{R}\) the reader should consult the paper itself. As an application of their technique, the authors describe an action of the group \(G\) on an abelian \(4\)-fold such that the resulting quotient admits a symplectic resolution and thus is a Kummer-type symplectic \(4\)-fold. Their second main result states that if \(\mathbb{E}\) is an elliptic curve with complex multiplication by \(\sqrt{-1}\), then there exists an embedding of the group \(G\) to the group \(\mathrm{Aut}(\mathbb{E}^4)\) such that the quotient \(\mathbb{E}^4/G\) has a resolution which is a Kummer symplectic \(4\)-fold \(X\) with \(b_2(X)=23\) and \(b_4(X)=276\). The article is generally well written and self-contained. Some places are, however, computationally involved and the authors rely there on computer algebra systems as Macaulay2 and Singular.