On Virial Expansion in Hard Sphere Model

Kiyoharu Kawana · 2025

Read On Virial Expansion in Hard Sphere Model by Kiyoharu Kawana (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2501-15766.txt.

Introduction

Kiyoharu Kawana revisits the virial expansion, the classic series that writes a gas's pressure as powers of its density. For hard spheres, a model of particles that only collide and never attract, the point where the series stops converging could in principle signal a phase change. Kawana takes the known series coefficients up to twelfth order and studies their pattern.

The coefficients fit a clean power law, and the matching closed form shows no singularity anywhere in the physical range of packing, up to the densest possible value near 0.74. On that reading, the expansion hints at no phase transition at all. Yet simulations have firmly established that hard spheres do freeze. Kawana also finds that deeper cluster coefficients behave as if three dimensions already counted as a large-dimension limit. The note warns that naive extrapolation of the series leads to unphysical answers.

Catalog record

CREATORKiyoharu Kawana
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE24 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2501-15766

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

On Virial Expansion in Hard Sphere Model Kiyoharu Kawana Abstract Virial expansion is a traditional approach in statistical mechanics that expresses thermodynamic quantities, such as pressure p p , as power series of density or chemical potential. Its radius of convergence can serve as a potential indicator of phase transition. In this study, we investigate the virial expansion of the hard-sphere model, using the known dimensionless virial coefficients B ~ k ​ ( k = 1 , 2 , ⋯ ) \tilde{B}_{k}~(k=1,2,\cdots) up to the 12 12 th order. We find that it is well fitted by B ~ k = 1.28 × k 1.90 \tilde{B}_{k}=1.28\times k^{1.90} , corresponding to the analytic continuation of the virial expansion of the pressure as ∼ Li − 1.90 ​ ( η ) \sim\mathrm{Li}_{-1.90}(\eta) , where η \eta is the packing fraction and Li s ​ ( x ) \mathrm{Li}_{s}(x) is the polylogarithm function. This implies the absence of singular behavior in the physical parameter space η ≤ η max ≈ 0.74 \eta\leq\eta_{\mathrm{max}}\approx 0.74 and no indication of phase transition in the virial expansion approach. In addition, we calculate the cluster-integral coefficients { b l } l = 1 ∞ \{b_{l}\}_{l=1}^{\infty} and observe that their asymptotic behavior resembles the results obtained in the large dimension limit ( D → ∞ D\rightarrow\infty ), suggesting that D = 3 D=3 might be already regarded as large dimension. However, the existence of phase transition in the hard-sphere model has been confirmed by numerous simulations, which clearly indicates that a naive extrapolation of the virial series can lead to unphysical results. 1 Introduction Statistical mechanics of classical fluid systems has a long and rich history. One of the traditional approaches is the virial expansion [ 1 , 2 , 3 , 4 ] which represents thermodynamic quantities as power series of density or chemical potential around the ideal gas. However, the convergence of the virial series is often slow, and it can diverge at a finite density. Such a divergence can be interpreted as a potential indication of a phase transition although its occurrence is not necessarily guaranteed. One of the simplest yet non-trivial models in classical fluid systems is the hard-sphere model. Despite the absence of attractive forces, various numerical simulations [ 5 , 6 , 7 , 8 , 9 ] have consistently demonstrated that this system exhibits a phase transition within physical density region \eta<\eta_{\mathrm{max}}\approx 0.74 , where \eta is the packing fraction. This intriguing result naturally motivates us to investigate the asymptotic behaviour of the virial series and its radius of convergence. Indeed, there have already been a number of studies [ 10 , 11 , 12 , 13 , 14 , 15 ] , employing various approximations and resummation techniques. In this paper, we present an updated analysis of the asymptotic behavior of the virial expansion in the hard-sphere model, using the known virial coefficients up to the 12 th order [ 16 ] . We find that these virial coefficients are well fitted by Eq. ( 48 ) and the estimated radius of convergence is \eta_{c}=1 , suggesting no phase transition within the physical density region \eta<\eta_{\mathrm{max}} . Furthermore, the estimated virial coefficients correspond to the analytic continuation of the pressure as Eq. ( 49 ), and \eta=\eta_{c}=1 is found to be a branch point of the polylogarithm function. We also calculate the cluster-integral coefficients b_{l} and observe that their asymptotic behavior resembles the results obtained in the large dimension limit ( D\rightarrow\infty ), suggesting that D=3 might be already regarded as large D . However, we should emphasize that it is still possible for the virial coefficients to become negative for k>12 , which then can lead to a finite radius of convergence below \eta_{\mathrm{max}}\approx 0.74 . This paper is organized as follows. In Section 2 , we provide a brief review the statistical mechanics of classical fluid system. We emphase that the existence of thermodynamic limit is guaranteed for a short-range potential that satisfies a couple of reasonable conditions. In Section 3 , we discuss the applicability of virial expansion and present one qualitative way of estimating the radius of convergence in the large l limit of the cluster integral \{b_{l}(V)\}_{l=1}^{\infty} . In Section 4 , we investigate the virial expansion in the hard-sphere model. For comparison, we first review D=1 and \infty limits, where the model is exactly solvable. We then examine the D=3 case, estimating the asymptotic behavior of the virial series using the known virial coefficients up to the 12 th order. Conclusion is given in Section 5 . 2 Classical fluid system and thermodynamic limit We consider thermodynamic equilibrium of classical particles interacting with a Hamiltonian in the D -dimensional Euclidean space: \displaystyle H_{N}=\sum_{i=1}^{N}\frac{p_{i}^{2}}{2m}+U_{N}(\{x_{i}\}_{i=1}^{N})~,\quad U_{N}(\{x_{i}\}_{i=1}^{N})=\sum_{i<j}^{N}v(x_{i},x_{j})~, (1) where x_{i}=(x_{i}^{1},x_{i}^{2},\cdots,x_{i}^{D}) , p_{i}^{2}=\sum_{M=1}^{D}(p_{i}^{M})^{2} , and v(x_{i},x_{j}) is a general two-body potential. If we impose the translation invariance, the pair potential becomes a mere function of relative coordinate as v(x_{1},x_{2})=v(x_{1}-x_{2}) , and we focus on this case in the following. The grand-canonical partition function is defined by \displaystyle\Xi[T,\mu,V] \displaystyle=\sum_{N=0}^{\infty}\frac{1}{N!}\left(\prod_{i=1}^{N}\int_{V}d^{D}x_{i}\int_{V}d^{D}p_{i}\right)e^{-\beta H_{N}+\beta\mu N} \displaystyle=\sum_{N=0}^{\infty}\frac{z^{N}}{N!}\left(\prod_{i=1}^{N}\int_{V}d^{D}x_{i}\right)e^{-\beta\sum_{i<j}v(x_{i}-x_{j})}~, (2) where \beta=1/T is the inverse temperature, \mu is the chemical potential, and z=e^{\beta\mu}(2\pi mT)^{D/2} is the fugacity. Here, \sum_{i<j} means the summation over all different pairs (x_{i},x_{j}) among N particles. The integrand in Eq. ( 2 ) can be written as \displaystyle W_{N}(\{x_{i}\}_{i=1}^{N}) \displaystyle\coloneq e^{-\beta\sum_{i<j}v(x_{i},x_{j})}=\prod_{i<j}(1+f_{ij}) \displaystyle=1+\sum_{i<j}f_{ij}+\sum_{i<j}\underset{(k,l)\neq(i,j)}{\sum_{k<l}}f_{ij}f_{kl}+\cdots~, (3) where f_{ij}\coloneq e^{-\beta v(x_{i}-x_{j})}-1 is the Mayer’s f function. Then, the cluster expansion allows us to perform the summation of N in Eq. ( 2 ) as [ 1 , 2 ] \displaystyle\log\Xi[T,\mu,V]=V\sum_{l=1}^{\infty}b_{l}(V)z^{l}~, (4) where \displaystyle b_{k}(V)\coloneq\frac{1}{k!V}\left(\prod_{i=1}^{k}d^{D}x_{i}\right)C_{k}(\{x_{i}\}_{i=1}^{k})~ (5) is the cluster integral. Here, C_{k}(\{x_{i}\}_{i=1}^{k}) denotes the contribution in W_{k}(\{x_{i}\}_{i=1}^{k}) such that all the k -particles are connected each other by the Mayer’s function, i.e. connected diagrams. In addition, the density is similarly expressed as \displaystyle\frac{\langle N\rangle}{V}=\frac{\partial}{\partial(\beta\mu)}\frac{\log\Xi[T,\mu,V]}{V}=\sum_{l=1}^{\infty}lb_{l}(V)z^{l}~. (6) The existence of the thermodynamic limit \displaystyle\beta p(T,\mu) \displaystyle\coloneq\lim_{V\rightarrow\infty}\frac{\log\Xi[T,\mu,V]}{V}~ (7) is a nontrivial problem for a generic N -body potential; however, in classical fluid systems, it is rigorously proven for short-range N -body potentials that satisfy several reasonable conditions [ 17 ] . Since this is a crucial result, we briefly overview it below. We first define the mutual potential energy by \displaystyle M_{N_{1}N_{2}}(\{x_{i}\}_{i=1}^{N_{1}};\{x_{j}^{{}^{\prime}}\}_{j=1}^{N_{2}})\coloneq U_{N_{1}+N_{2}}(\{x_{i}\}_{i=1}^{N_{1}},\{x_{j}^{{}^{\prime}}\}_{j=1}^{N_{2}})-U_{N_{1}}(\{x_{i}\}_{i=1}^{N_{1}})-U_{N_{2}}(\{x_{j}^{{}^{\prime}}\}_{j=1}^{N_{2}})~. (8) Then, we say that U_{N}(\{x_{i}\}_{i=1}^{N}) is tempered if there exist {}^{\exists}\lambda>D and {}^{\exists}R_{0}>0 , {}^{\exists}A\geq 0 such that \displaystyle M_{N_{1}N_{2}}(\{x_{i}\}_{i=1}^{N_{1}};\{x_{j}^{{}^{\prime}}\}_{j=1}^{N_{2}})\leq AN_{1}N_{2}e^{-\lambda r} (9) whenever |x_{i}-x_{j}^{{}^{\prime}}|\geq r\geq R_{0} for all i=1,2,\cdots,N_{1} and j=1,2,\cdots,N_{2} . Roughly speaking, this temperedness implies the suppression of the positive part of U_{N} for the long-distance r\rightarrow\infty . For example, in the pair potential case ( 1 ), this implies (with N_{1}=N_{2}=1 ) \displaystyle v(x)\leq A|x|^{-\lambda}\quad\text{for}~|x|\geq R_{0}~, (10) which implies that positive (repulsive) part of v(x) should damp faster than |x|^{-\lambda}(<|x|^{-D}) for |x|\rightarrow\infty . In addition, we say that U_{N}(\{x_{i}\}_{i=1}^{N}) is stable if there exists {}^{\exists}B\geq 0 such that \displaystyle U_{N}(\{x_{i}\}_{i=1}^{N})\geq-NB (11) for all N=1,2,\cdots and {}^{\forall}(x_{1},\cdots,x_{N})\in(\mathbb{R}^{D})^{N} . The importance of this stability is that it implies the convergence of the grand-canonical partition function as \displaystyle\Xi[T,\mu,V] \displaystyle=1+\sum_{N=1}^{\infty}\frac{z^{N}}{N!}\left(\prod_{i=1}^{N}\int_{V}d^{D}x_{i}\right)e^{-\beta U_{N}(\{x_{i}\}_{i=1}^{N})} \displaystyle\leq 1+\sum_{N=1}^{\infty}\frac{z^{N}}{N!}\left(\prod_{i=1}^{N}\int_{V}d^{D}x_{i}\right)e^{-N\beta B} \displaystyle=\exp\left(Vze^{\beta B}\right)~ (12) for {}^{\forall}z\geq 0 and {}^{\forall}\beta\geq 0 . Here, we list typical pair potentials that satisfy temperedness and stability [ 17 ] : \bullet Hard sphere potential \displaystyle v_{\mathrm{HS}}(x)=\begin{cases}\infty&\text{for}|x|<2a\\ 0&\text{for}|x|\geq 2a\end{cases}~,

Read the complete text