Attractor invariants, brane tilings and crystals

Supersymmetric D-brane bound states on a Calabi-Yau threefold $X$ are counted by generalized Donaldsdon-Thomas invariants $\Omega_Z(\gamma)$, depending on a Chern character (or electromagnetic charge) $\gamma\in H^*(X)$ and a stability condition (or central charge) $Z$. Attractor invariants $\Omega_*(\gamma)$ are special instances of DT invariants, where $Z$ is the attractor stability condition $Z_\gamma$ (a generic perturbation of self-stability), from which DT invariants for any other stability condition can be deduced. While difficult to compute in general, these invariants become tractable when $X$ is a crepant resolution of a singular toric Calabi-Yau threefold associated to a brane tiling, and hence to a quiver with potential. We survey some known results and conjectures about framed and unframed refined DT invariants in this context, and compute attractor invariants explicitly for a variety of toric Calabi-Yau threefolds, in particular when $X$ is the total space of the canonical bundle of a smooth projective surface, or when $X$ is a crepant resolution of $C^3/G$. We check that in all these cases, $\Omega_*(\gamma)=0$ unless $\gamma$ is the dimension vector of a simple representation or belongs to the kernel of the skew-symmetrized Euler form. Based on computations in small dimensions, we predict the values of all attractor invariants, thus potentially solving the problem of counting DT invariants of these threefolds in all stability chambers. We also compute the non-commutative refined DT invariants and verify that they agree with the counting of molten crystals in the unrefined limit.