On the noncommutative geometry of square superpotential algebras (Q1952136)

The author studies quiver algebras called \textit{superpotential algebras}. Let \(Q\) be a quiver and \(kQ\) its path algebra. The trace space \(\text{tr}(kQ)\) of \(kQ\) is the space spanned by all paths in \(Q\) up to cyclic equivalence, and its elements are called superpotentials. For each \(a\in Q\) a \(k\)-linear map \(\partial_a:\text{tr}(kQ)\rightarrow kQ\) is defined, relating paths to \(a\), and then \(\partial W=\langle\partial_aW|a\in Q_1\rangle\). The superpotential algebra with quiver \(Q\) and superpotential \(W\) is \(kQ/W\). A special class of superpotential algebras admits an embedding into a two-torus such that the image of its underlying graph is a square grid, possibly with diagonal edges in the unit squares. These are called \textit{square superpotential algebras}. To a square superpotential algebra, the author assigns a device called an \textit{impression}: Let \(A\) be a representable \(k\)-algebra with \(k\) algebraically closed and with center \(Z\). Suppose there exists a commutative, finitely generated \(k\)-algebra \(B\), an open dence subset \(U\subset\text{Max} B\), and an algebra monomorphism \(\tau:A\rightarrow\text{End}_B(B^d)\), \(d\leq\infty\), such that the composition with the evaluation map is a simple representation for each \(\mathfrak q\in U\). Then \(Z\cong R:=\{f\in B|f1_d\in\text{im}\tau\}\subset B\). If \(\text{Max}B\overset\phi\rightarrow\text{Max} R\), \(\mathfrak q\mapsto\mathfrak q\cap R\), is surjective then \((\tau,B)\) is called an impression of \(A\). It is called a pre-impression if \(\phi\) is not surjective. The main properties of impressions are established. An impression determines explicitly both the center of \(A\) and all simple \(A\)-module isoclasses of maximal \(k\)-dimension, called \textit{large modules}. In special nice cases, the large modules are parameterized by the smooth locus of the algebras center. It is proved that if \((\tau, B)\) is an impression of a finitely generated algebra \(A\), module-finite over its center, with \(B\) prime, then if \(V\) is a large module, then there is some \(\mathfrak q\in\text{Max}B\) such that \(V\cong (B/\mathfrak q)^d\), where \(av=\tau_q(a)v\). The next main result says that if \(A=kQ/I\) is a quiver algebra that admits a pre-impression \((\tau,B)\) such that \(\tau(e_i)=E_{ii}\) and \(\overline\tau(e_iAe_i)=\overline\tau(e_jAe_j)\subset B\) for each \(i,j\in Q_0\), then \(A\) and its center \(Z\) are noetherian rings, \(A\) is a finitely generated \(Z\)-module, and \[ Z=k[\sum\gamma_i\in\bigoplus e_iAe_i|\overline\tau(\gamma_j)=\overline\tau(\gamma_i)=\overline\tau(\gamma_j)\text{ for each }i,j\in Q_0]. \] In particular, the author determines an impression \((\tau,B=k[x_1,x_2,y_1,y_2])\) of a square superpotential algebra \(A\) with center \(Z\), and origin \(\mathfrak m\) of \(\text{Max} Z\). This is used to prove that - \(Z\) is a \(3\)-dimensional normal toric domain and the localization \(Z_{\mathfrak m}\) is Gorenstein. Also, \(A_{\mathfrak m} =Z_{\mathfrak m}\otimes_Z A\) is a noncommutative crepant resolution of \(Z_{\mathfrak m}\), and consequently a local Calabu-Yau algebra of dimension \(3\). - Let \(V\) be a simple \(A\)-module. Set \(\mathfrak p=\text{ann}_A V\) and \(\mathfrak m=\mathfrak p\cap Z\in\text{Max} Z\). Then \(\dim_k e_i V\leq 1\) for each \(i\in Q_0\). Also, \(V\) can be classified into three different cases. The \(Y^{p,q}\)-algebras are a special class of square superpotential algebras where the relations are related to Sasaki-Einstein manifold. They are treated in particular by proving that for a non-localized \(Y^{p,q}\)-algebra: (1) If \(p\neq q\) and \(V\) is a simple \(A\)-module, then \(V\) is either a vertex simple module or a large module, (2) The Azumaya locus of \(A\) coincides with the smooth locus of \(Z\), and (3) \(A\) is homogeneous of global dimension \(3\). The article is explicit with nice examples and good illustrations. The proofs are explicit and use the conversions from the geometry of the diagrams to the quiver algebras in a nice way. Furthermore, the article proves that the correct physical understanding depends on the representations of noncommutative algebras.