Approximate Counting for Spin Systems in Sub-Quadratic Time

We present two approximate counting algorithms with $\widetilde{O}(n^{2-c}/\varepsilon^2)$ running time for some constant $c > 0$ and accuracy $\varepsilon$: (1) for the hard-core model when strong spatial mixing (SSM) is sufficiently fast; (2) for spin systems with SSM on planar graphs with quadratic growth, such as $\mathbb{Z}^2$. The latter algorithm also extends to (not necessarily planar) graphs with polynomial growth, such as $\mathbb{Z}^d$, albeit with a running time of the form $\widetilde{O}(n^2\varepsilon^{-2}/2^{c(\log n)^{1/d}})$ for some constant $c > 0$ and $d$ being the exponent of the polynomial growth. Our technique utilizes Weitz's self-avoiding walk tree (STOC, 2006) and the recent marginal sampler of Anand and Jerrum (SIAM J. Comput., 2022).