On the motivic Donaldson-Thomas invariants of quivers with potentials (Q2434618): Difference between revisions

The paper under review studies motivic Donaldson-Thomas invariants for quivers \(Q\) (given by a set of vertices \(Q_{0}\) and a set of arrows \(Q_{1}\)) between them) together with a potential \(W=\sum a_{u}u\), where the sum extends over a finite number of cycles of the quiver (meaning compositions of arrows where the targer of the last one coincides with the source of the first one). Let \(\alpha\in\mathbb{N}^{Q_{0}}-\{0\}\) be a dimension vector and consider representations of \(Q\) over \(k\)-vector spaces \[ M=((M_{i})_{i\in Q_{0}}, (M_{a})_{a\in Q_{1}} , \] given by a \(k\)-vector space of dimension \(\alpha_{i}\) for each vertex and a linear map for each arrow. Let \(\theta\in\mathbb{R}^{Q_{0}}\) be some fixed vector, and define a representation \(M\) to be semistable (resp. stable) if for every proper non trivial subrepresentation \(N\) we have \(\mu_{\theta}(N)\leq \mu_{\theta}(M)\) (resp. \(\mu_{\theta}(N)<\mu_{\theta}(M)\)), where the slope is given by \[ \mu_{\theta}(N)=\frac{\theta_{i}\cdot \dim_{k}N_{i}}{\sum \dim_{k}N_{i}} . \] Let \(k=\mathbb{C}\). Denote by \(M_{\theta}^{\mathrm{sst}}(Q,\alpha)\) the moduli space of semistable representations of \(Q\) with dimension vector \(\alpha\). Given a weight function \(\mathrm{wt}:Q_{1}\rightarrow \mathbb{N}\), it defines an action of \(\mathbb{C}^{\ast}\) on the moduli space, by multiplying each linear map with the corresponding weight, \((tM)_{a}=t^{\mathrm{wt}(a)}M_{a}\), \(a\in Q_{1}\). Define the trace of the potential of a representation as \[ w(M)=\sum a_{u} tr(M_{u})\; , \] the sum taken over the cycles considered in the definition of the potential \(W\), and \(M_{u}=M_{a_{1}}\circ\cdots\circ M_{a_{n}}\), with \(u=a_{1}\cdots a_{n}\). It happens that, if \(\theta\) is \(\alpha\)-generic (meaning that semistable representations are actually stable), the function \[ w:M_{\theta}^{\mathrm{sst}}(Q,\alpha)\rightarrow \mathbb{C} \] is \(\mathbb{C}^{\ast}\)-equivariant, hence we can apply (Proposition 1.11 [\textit{K. Behrend} et al., Invent. Math. 192, No. 1, 111--160 (2013; Zbl 1267.14008)]) to obtain that the motivic vanishing cycle of \(w\) can be represented as \[ [\varphi_{w}]=[w^{-1}(1)]-[w^{-1}(0)]. \] By using this, it is defined the motivic Donaldson-Thomas series for each \(\mu\in\mathbb{R}\), as \[ A_{\mu}=\sum_{\mu_{\theta}(\alpha)=\mu}A_{\alpha}^{\theta}x^{\alpha}\; , \] where each \(A_{\alpha}^{\theta}\) is the virtual motive of the corresponding moduli stack. The paper shows that there exists a homomorphism \[ I_{\mathrm{eq}}:\hat{H}_{\mathrm{eq}}\rightarrow \mathbb{T} \] from the equivariant Hall subalgebra to the quantum torus of the quiver and, from this, the motivic Donaldson-Thomas invariants can be computed in the Hall algebra of the category of category of representations of the quiver of a finite field, in terms of the Harder-Narasimhan recursion (i.e., in terms of semistable or stable subrepresentations of given slopes). The paper finishes with an extension of the result for quivers with arbitrary potentials.