Casimir versus Helmholtz forces: Exact results

TL;DR

This paper provides exact analytical results comparing Casimir and Helmholtz fluctuation-induced forces in the one-dimensional Ising model under different ensembles, highlighting ensemble dependence and finite-size effects.

Contribution

It derives exact partition functions and forces for the 1D Ising model with fixed magnetization using transfer matrix methods, extending previous results and analyzing ensemble differences.

Findings

Helmholtz and Casimir forces differ significantly for finite systems.

Finite-size corrections are more pronounced in the canonical ensemble.

Legendre transformation relates free energies in different ensembles in the thermodynamic limit.

Abstract

Recently, attention has turned to the issue of the ensemble dependence of fluctuation induced forces. As a noteworthy example, in $O(n)$ systems the statistical mechanics underlying such forces can be shown to differ in the constant $\vec{M}$ magnetic canonical ensemble (CE) from those in the widely-studied constant $\vec{h}$ grand canonical ensemble (GCE). Here, the counterpart of the Casimir force in the GCE is the \textit{Helmholtz} force in the CE. Given the difference between the two ensembles for finite systems, it is reasonable to anticipate that these forces will have, in general, different behavior for the same geometry and boundary conditions. Here we present some exact results for both the Casimir and the Helmholtz force in the case of the one-dimensional Ising model subject to periodic and antiperiodic boundary conditions and compare their behavior. We note that the Ising model has recently being solved in Phys.Rev. E {\bf 106} L042103(2022), using a combinatorial approach, for the case of fixed value $M$ of its order parameter. Here we derive exact result for the partition function of the one-dimensional Ising model of $N$ spins and fixed value $M$ using the transfer matrix method (TMM); earlier results obtained via the TMM were limited to $M=0$ and $N$ even. As a byproduct, we derive several specific integral representations of the hypergeometric function of Gauss. Using those results, we rigorously derive that the free energies of the CE and grand GCE are related to each other via Legendre transformation in the thermodynamic limit, and establish the leading finite-size corrections for the canonical case, which turn out to be much more pronounced than the corresponding ones in the case of the GCE.