Walking on Archimedean Lattices: Insights from Bloch Band Theory
Davidson Noby Joseph, Igor Boettcher
TL;DR
This work establishes a direct link between counting returning walks on Archimedean lattices and Bloch-band theory in tight-binding models. By introducing the Bloch adjacency matrix $A(\boldsymbol{k})$ and its spectrum, the authors derive $S_n$ from Brillouin-zone moments $\int_{\boldsymbol{k}}\varepsilon_\alpha(\boldsymbol{k})^n$, yielding closed forms or generating functions for all eleven two-dimensional Archimedean lattices. They also connect these results to the density of states and provide explicit asymptotics for large $n$, with analytical DOS expressions for several lattices and numerical validation via large flakes/clusters. The methods generalize to other periodic Euclidean lattices and dimensions, offering a powerful bridge between combinatorics, graph theory, and solid-state physics. Overall, the paper delivers a comprehensive, analytically tractable framework to count returning walks, compute DOS, and study asymptotics across a rich family of lattices.
Abstract
Returning walks on a lattice are sequences of moves that start at a given lattice site and return to the same site after $n$ steps. Determining the total number of returning walks of a given length $n$ is a typical graph-theoretical problem with connections to lattice models in statistical and condensed matter physics. We derive analytical expressions for the returning walk numbers on the eleven two-dimensional Archimedean lattices by developing a connection to the theory of Bloch energy bands. We benchmark our results through an alternative method that relies on computing the moments of adjacency matrices of large graphs, whose construction we explain explicitly. As condensed matter physics applications, we use our formulas to compute the density of states of tight-binding models on the Archimedean lattices and analytically determine the asymptotics of the return probability. While the Archimedean lattices provide a sufficiently rich structure and are chosen here for concreteness, our techniques can be generalized straightforwardly to other two- or higher-dimensional Euclidean lattices.
