Entropy of fully-packed rigid rods on generalized Husimi trees: a route to the square lattice limit

TL;DR

This study estimates the entropy of fully-packed rigid rods on generalized Husimi trees, providing new insights into their disordered phases and improving entropy estimates for small k-mer sizes, bridging towards the square lattice limit.

Contribution

It introduces a systematic Husimi lattice approach to approximate the entropy of fully-packed rods, extending known results and offering new estimates for larger k-mer sizes.

Findings

Entropy estimates for dimers and trimers closely match known results.

Evidence suggests the fully packed phase is disordered for k ≥ 4.

Husimi lattice calculations indicate a high-density nematic-disordered transition.

Abstract

Although hard rigid rods ($k$-mers) defined on the square lattice have been widely studied in the literature, their entropy per site, $s(k)$, in the full-packing limit is only known exactly for dimers ($k=2$) and numerically for trimers ($k=3$). Here, we investigate this entropy for rods with $k \le 7$, by defining and solving them on Husimi lattices built with diagonal and regular square lattice clusters of effective lateral size $L$, where $L$ defines the level of approximation to the square lattice. Due to an $L$-parity effect, by increasing $L$ we obtain two systematic sequences of values for the entropies $s_L(k)$ for each type of cluster, whose extrapolations to $L \rightarrow \infty$ provide estimates of these entropies for the square lattice. For dimers, our estimates for $s(2)$ differ from the exact result by only $0.03\%$, while that for $s(3)$ differs from best available estimates by $3\%$. In this paper, we also obtain a new estimate for $s(4)$. For larger $k$, we find that the extrapolated results from the Husimi tree calculations do not lie between the lower and upper bounds established in the literature for $s(k)$. In fact, we observe that, to obtain reliable estimates for these entropies, we should deal with levels $L$ that increase with $k$. However, it is very challenging computationally to advance to solve the problem for large values of $L$ and for large rods. In addition, the exact calculations on the generalized Husimi trees provide strong evidence for the fully packed phase to be disordered for $k\geq 4$, in contrast to the results for the Bethe lattice wherein it is nematic, thus providing evidence for a high density nematic-disordered transition in the system of $k$-mers with vacancies.