Completing the high-activity rod expansion of the hard-square lattice gas: the exact $O(z^{-2})$ three-rod term, convergence control, and the three-dimensional hard-cube extension

Ramola and Dhar (2012) reformulated the high-fugacity expansion of the hard-square lattice gas (nearest- and next-nearest-neighbor exclusion on the square lattice) as a Mayer-like series in interacting vertical rods in powers of $z^{-1/2}$, obtaining the free energy and density exactly through $O(z^{-3/2})$. Three parts of the program they posed remain open: (1) the exact $O(z^{-2})$ coefficient, which receives contributions from configurations with up to three rods; (2) rigorous control, in particular positive lower bounds, on the radii of convergence of these expansions — the two-dimensional $z^{-1/2}$ series and, once constructed, its three-dimensional analogue — which would also yield a rigorous upper bound on the critical fugacity of the fluid-columnar transition; and (3) the extension of the rod expansion to the three-dimensional analogue, $2\times2\times2$ hard cubes on the cubic lattice, with closed-form summation of the rodlike excitations and their interactions beyond the first-order order-by-disorder selection. Later work has extended the method to hard rectangles and $k\times k$ squares at two-defect-cluster order (Nath, Kundu and Rajesh 2015) and established the three-dimensional hard-cube phase diagram together with a first-order selection calculation (Vigneshwar et al. 2019), while rigorous convergence of inverse-fugacity expansions is proven only for non-sliding systems with finitely many perfect coverings (Jauslin and Lebowitz 2017), a class that excludes the sliding hard-square gas. The columnar ordering of hard squares itself has since been proven at large fugacity (Hadas and Peled 2025), but by reflection-positivity methods that yield no convergent expansion and no analyticity in $z^{-1/2}$.

Background

The hard-square lattice gas is a classical model of entropy-driven ordering. Particles occupy the sites of the square lattice with exclusion of both nearest and next-nearest neighbors, which is equivalent to placing $2\times2$ squares that cover four elementary plaquettes without overlap. Each particle carries a fugacity (activity) $z$: the grand-canonical partition function weights every allowed configuration by $z^{N}$, so the mean density increases monotonically with $z$. At small $z$ the system is a disordered fluid; at large $z$ it orders in a columnar fashion, with particles preferentially occupying every other row (or every other column), breaking translation symmetry in one direction only. This differs from the more familiar nearest-neighbor exclusion gas, which orders on one of the two checkerboard sublattices. The fluid-columnar transition of hard squares is continuous and occurs near $z\simeq97.5$; unlike hard hexagons, the model has no exact solution, so its dense phase is accessed by controlled expansions.

For lattice gases that order on a sublattice, the thermodynamics at large $z$ is captured by the Gaunt-Fisher expansion: an exact series for the pressure in powers of $1/z$ obtained by summing vacancy-cluster defects about the unique perfect sublattice covering (Gaunt and Fisher 1965). For hard squares the corresponding expansion is singular. At full packing the number of perfect coverings on an $L\times L$ torus is $4(2^{L/2}-1)$: exponentially large in the linear size, though subextensive in the area. The partition functions of individual columns and defects involve the largest root $\lambda=(1+\sqrt{1+4z})/2$ of $x^{2}-x-z=0$, which grows as $\sqrt{z}$, so the expansion naturally proceeds in powers of $z^{-1/2}$ rather than $z^{-1}$. Bellemans and Nigam (1967) computed the first correction, of order $1/z$, about a perfectly columnar state.

Ramola and Dhar (2012) turned this into a systematic singular perturbation theory. Introducing different fugacities $z_A$ and $z_B$ for particles on the two types of rows, expanding about the columnar state $z_B=0$, and then setting $z_A=z_B=z$, they showed that the resulting Mayer-like series regroups into a Mayer expansion of a polydisperse gas of interacting vertical rods, maximal vertical stacks of defect squares bridging the occupied rows, in which the $n$-th term is of order $z^{-(n+1)/2}$. The exponential correlation length between rods, $\xi\sim\sqrt{z}$, supplies compensating factors of $\sqrt{z}$ when the distances between rods are summed, which is why only configurations with at most $n$ rods contribute at order $z^{-(n+1)/2}$. Explicitly summing the one-rod and two-rod terms, they obtained for the free energy per site $f$ and the density $\rho$:

$$-f(z)=\frac{1}{4}\log z+\frac{1}{4z^{1/2}}+\frac{1}{4z}+\frac{3\log(9/8)+11/96}{z^{3/2}}+O(z^{-2}),$$ $$\rho(z)=\frac{1}{4}-\frac{1}{8z^{1/2}}-\frac{1}{4z}-\frac{(9/2)\log(9/8)+11/64}{z^{3/2}}+O(z^{-2}),$$

with an analogous series for the row order parameter, the observable measuring the excess of particles on, say, even over odd rows, which distinguishes the two columnar phases; the truncated series reproduces Monte Carlo densities for $z\gtrsim150$. The terms at orders $1/z$ and $1/z^{3/2}$ are the last ones known exactly.

Two further concepts delimit what is known. A hard-core lattice system is called non-sliding if it has only finitely many perfect coverings up to translation; for such systems Pirogov-Sinai theory, a rigorous cluster-expansion technique for phase coexistence among degenerate ground states, applies, and Jauslin and Lebowitz (2017) used it to prove that the Gaunt-Fisher expansion of non-sliding systems has a nonzero radius of convergence. The hard-square gas is sliding: it has an exponentially degenerate set of perfect coverings, it orders columnarly rather than on a sublattice, and its pressure need not be analytic in $1/z$ at all (the expansion variable is $z^{-1/2}$). This long placed it outside every rigorous scheme, and the situation has changed only in part: Hadas and Peled (2025) proved, via a novel extension of the chessboard estimate (a reflection-positivity technique that proves phase coexistence by averaging Gibbs states over periodic colorings of the lattice), that at large fugacity the model has four extremal periodic Gibbs states with columnar order, and quantified the directional correlation decay in them. Their result establishes the ordering that the rod series expands about, but it produces no convergent expansion, no analyticity of the pressure in $z^{-1/2}$, and no radius-of-convergence bound, so the series itself remains without rigorous control. Order by disorder denotes entropic selection within a degenerate manifold of reference states: the leading full-packing free energies are equal, but subleading terms differ, and the state with the largest fluctuation entropy wins in the thermodynamic limit. Both notions govern the unresolved questions of convergence and of the extension to three dimensions, where the reference states of the $2\times2\times2$ hard-cube gas on the cubic lattice are columnar states whose cross-sections are fully packed $2\times2$-square configurations and the excitations contributing at order $1/z$ are rigid rods along the $x$- or $y$-directions.

Problem Statement

The problem is to complete and rigorously control the high-activity rod expansion of the hard-square lattice gas, as posed by Ramola and Dhar (2012). It comprises three linked open problems posed by the source, all serving one objective: turning the two-rod, leading-order result into a complete and controlled series method for the high-density phases of sliding, exponentially degenerate hard-core lattice gases. All three parts are required for a complete resolution of the task.

1. Exact $O(z^{-2})$ term. Compute the exact closed-form contribution of order $z^{-2}$ to the free energy per site, the density, and the row order parameter of the two-dimensional hard-square gas. By the power counting of the rod expansion, this order receives contributions from configurations with up to three rods of arbitrary sizes, including the subleading pieces of the one- and two-rod sectors; the result must be consistent with the published coefficients at orders $z^{-1}$ and $z^{-3/2}$.

2. Convergence control. Establish a rigorous, strictly positive lower bound on the radius of convergence of the two-dimensional series in the variable $z^{-1/2}$. Establish also the analogous lower bound on the radius of convergence of the three-dimensional series of item 3, for that series when it is constructed. This is the sliding analogue of the Gaunt-Fisher convergence theorem, known only for non-sliding systems with finitely many perfect coverings, and such a lower bound would yield a rigorous upper bound on the critical fugacity of the fluid-columnar transition near $z\simeq97.5$ (equivalently, an upper bound on the critical density); the recent chessboard-estimate proof of columnar order for this model (Hadas and Peled 2025) does not address this question.

3. Three-dimensional extension. Construct the corresponding high-activity expansion for the three-dimensional analogue, $2\times2\times2$ hard cubes on the cubic lattice, in closed form through the two-rod order, at which interactions between rods first enter: the reference states are the columnar states whose cross-sections are fully packed $2\times2$-square configurations, and the excitations contributing at order $1/z$ are rigid rods along the $x$- or $y$-directions. The order-by-disorder selection among the reference states is established at first order (alignment of columnar order across planes, with free-energy gain $\sim L^{2}/4z$; Vigneshwar et al. 2019); the rod-rod summations in three dimensions are harder to carry out in closed form.

References

1. Kabir Ramola and Deepak Dhar, 'High-activity perturbation expansion for the hard square lattice gas', Physical Review E 86, 031135 (2012). DOI: 10.1103/PhysRevE.86.031135; arXiv:1204.4484. https://doi.org/10.1103/PhysRevE.86.031135
2. David S. Gaunt and Michael E. Fisher, 'Hard-Sphere Lattice Gases. I. Plane-Square Lattice', The Journal of Chemical Physics 43, 2840 (1965). DOI: 10.1063/1.1697217. https://doi.org/10.1063/1.1697217
3. A. Bellemans and R. K. Nigam, 'Phase Transitions in Two-Dimensional Lattice Gases of Hard-Square Molecules', The Journal of Chemical Physics 46, 2922 (1967). DOI: 10.1063/1.1841157. https://doi.org/10.1063/1.1841157
4. Trisha Nath, Joyjit Kundu, and R. Rajesh, 'High-Activity Expansion for the Columnar Phase of the Hard Rectangle Gas', Journal of Statistical Physics 160, 1173-1197 (2015). DOI: 10.1007/s10955-015-1285-y; arXiv:1411.7831. https://doi.org/10.1007/s10955-015-1285-y
5. Ian Jauslin and Joel L. Lebowitz, 'Crystalline Ordering and Large Fugacity Expansion for Hard-Core Lattice Particles', The Journal of Physical Chemistry B 122, 3266-3271 (2018; published online 2017). DOI: 10.1021/acs.jpcb.7b08977; arXiv:1705.02032. https://doi.org/10.1021/acs.jpcb.7b08977
6. Ian Jauslin and Joel L. Lebowitz, 'High-Fugacity Expansion, Lee-Yang Zeros, and Order-Disorder Transitions in Hard-Core Lattice Systems', Communications in Mathematical Physics 364, 655-682 (2018). DOI: 10.1007/s00220-018-3269-7; arXiv:1708.01912. https://doi.org/10.1007/s00220-018-3269-7
7. N. Vigneshwar, Dipanjan Mandal, Kedar Damle, Deepak Dhar, and R. Rajesh, 'Phase diagram of a system of hard cubes on the cubic lattice', Physical Review E 99, 052129 (2019). DOI: 10.1103/PhysRevE.99.052129; arXiv:1902.06408. https://doi.org/10.1103/PhysRevE.99.052129
8. Trisha Nath, Deepak Dhar, and R. Rajesh, 'Stability of columnar order in assemblies of hard rectangles or squares', EPL 114, 10003 (2016). DOI: 10.1209/0295-5075/114/10003; arXiv:1601.02198. https://doi.org/10.1209/0295-5075/114/10003
9. Kabir Ramola, Kedar Damle, and Deepak Dhar, 'Columnar Order and Ashkin-Teller Criticality in Mixtures of Hard Squares and Dimers', Physical Review Letters 114, 190601 (2015). DOI: 10.1103/PhysRevLett.114.190601; arXiv:1408.4943. https://doi.org/10.1103/PhysRevLett.114.190601
10. Dipanjan Mandal, Trisha Nath, and R. Rajesh, 'Estimating the critical parameters of the hard square lattice gas model', Journal of Statistical Mechanics: Theory and Experiment 2017, 043201 (2017). DOI: 10.1088/1742-5468/aa650e. https://doi.org/10.1088/1742-5468/aa650e
11. A. Mazel, I. Stuhl, and Y. Suhov, 'Kepler's conjecture and phase transitions in the high-density hard-core model on Z^3' (2021). arXiv:2112.14250. https://arxiv.org/abs/2112.14250
12. Xinliang Lyu, 'Lattice and PT symmetries in tensor-network renormalization group: Case study of a hard-square lattice gas model', Physical Review B 114, 034111 (2026). DOI: 10.1103/vl23-xsyr. https://doi.org/10.1103/vl23-xsyr
13. Daniel Hadas and Ron Peled, 'Columnar order in random packings of $2\times2$ squares on the square lattice', Journal of the European Mathematical Society, published online (2025). DOI: 10.4171/jems/1680; arXiv:2206.01276. https://doi.org/10.4171/jems/1680
14. Qidong He and Ian Jauslin, 'High-Fugacity Expansion and Crystallization in Non-sliding Hard-Core Lattice Particle Models Without a Tiling Constraint', Journal of Statistical Physics 191, 135 (2024). DOI: 10.1007/s10955-024-03349-x. https://doi.org/10.1007/s10955-024-03349-x