Partition function of 2D Coulomb gases with radially symmetric potentials and a hard wall
Matthias Allard, Peter J. Forrester, Sampad Lahiry, Bojian Shen · 2025
Read Partition function of 2D Coulomb gases with radially symmetric potentials and a hard wall by Matthias Allard, Peter J. Forrester, Sampad Lahiry, Bojian Shen (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-2506-14738.txt.
Introduction
Matthias Allard, Peter Forrester, Sampad Lahiry, and Bojian Shen study a classic math physics model in a 2025 paper. The model treats many charged particles in a flat disk, held in by a round wall and a smooth pull toward the center. A key quantity, the partition function, sums the weights of all states and reveals the system's broad behavior.
The authors work out how that sum grows when the number of particles is large. They treat disks and rings, and walls set at the edge of the droplet or inside it. One term, tied to the log of the particle count, takes a different value when the wall sits inside. They also find fixed numbers, given by integrals, that do not depend on the pull used. The work sharpens exact results for this model.
Catalog record
| CREATOR | Matthias Allard, Peter J. Forrester, Sampad Lahiry, Bojian Shen |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 85 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2506-14738
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Partition function of 2D Coulomb gases with radially symmetric potentials and a hard wall Matthias Allard, Peter J. Forrester, Sampad Lahiry, Bojian Shen Abstract The large N N asymptotic expansion of the partition function for the normal matrix model is predicted to have special features inherited from its interpretation as a two-dimensional Coulomb gas. However for the latter, it is most natural to include a hard wall at the boundary of the droplet. We probe how this affects the asymptotic expansion in the solvable case that the potential is radially symmetric and the droplet is a disk or an annulus. We allow too for the hard wall to be strictly inside the boundary of the droplet. It is observed the term of order log N \log N , has then a different rational number prefactor to that when the hard wall is at the droplet boundary. Also found are certain universal (potential independent) numerical constants given by definite integrals, both at order N \sqrt{N} , and in the constant term. MSC Classification]60B20,82D05,41A60,60G55 keywords Normal matrix model; two-dimensional Coulomb gas; large N limit; asymptotic expansion; hard wall pacs [ 1 Introduction 1.1 Confined plasmas and constrained droplets At the beginning of the 1980’s Alastuey and Jancovici Alastuey and Jancovici 1981 introduced an exactly solvable two-dimensional one-component plasma model on a disk, albeit at just one value of the dimensionless inverse temperature \beta=2 ; this is detailed too in the text ( Forrester 2010 , §15.3.1) and the review Forrester 1998 . The model is defined within a disk of radius R . Inside the disk is a smeared out uniform background of charge density \rho_{\rm b}:=-N/\pi R^{2} and N positive unit charged particles — in particular the total charge is zero. The pair potential used to compute the energies associated with background and the particles is \Psi(\vec{r},\vec{r}^{\prime})=-\log|\vec{r}-\vec{r}\,^{\prime}| as required for a Coulomb system in two-dimensions. For general dimensionless inverse temperature \beta , the Boltzmann factor for this system, defined in terms of the energies U of the particle-particle, particle-background and background-background interactions is (see e.g. ( Forrester 2010 , Eq. (1.72) with \Gamma=\beta ) ) \displaystyle e^{-\beta U}:=e^{-\beta N^{2}((1/2)\log(R)-3/8)}e^{-\pi\beta\rho_{\rm b}\sum_{j=1}^{N}|z_{j}|^{2}/2}\prod_{1\leq j<k\leq N}|z_{k}-z_{j}|^{\beta}, (1.1) with z_{j}=x_{j}+iy_{j} , (x_{j},y_{j}) being the coordinates in the plane of particle j . In the case where \rho_{\rm b} is fixed and \beta=2 , it was shown that the thermodynamic limit N,R\to\infty of the dimensionless free energy \displaystyle\beta F|_{\beta=2}:=-\log\frac{1}{N!}\int_{\Omega}d^{2}z_{1}\cdots\int_{\Omega}d^{2}z_{N}e^{-\beta U}\Big|_{\beta=2}=N^{2}\log(R)-\frac{3}{4}N^{2}-\log\frac{\mathcal{Q}_{N}}{N!}, (1.2) can be obtained exactly. Here, \Omega=\{z:|z|<R\} (i.e. the disk of radius R ), and \mathcal{Q}_{N} denotes the configuration integral \displaystyle\mathcal{Q}_{N}:=\int_{\Omega}d^{2}z_{1}\cdots\int_{\Omega}d^{2}z_{N}\,e^{-\pi\rho_{\rm b}\sum_{j=1}^{N}|z_{j}|^{2}}\prod_{1\leq j<k\leq N}|z_{k}-z_{j}|^{2}. (1.3) Thus, the results of Alastuey and Jancovici 1981 (at leading order in N ), supplemented by results from Jancovici et al. 1994 (next two terms) give the asymptotic expansion \displaystyle\beta F|_{\beta=2}\sim\pi R^{2}f_{\Omega}+(2\pi R)\gamma_{\partial\Omega}+\frac{1}{12}\log(N)+{\rm O}(1), (1.4) where f_{\Omega} (the free energy per volume) and \gamma_{\partial\Omega} (the surface tension per boundary length) are given by \displaystyle f_{\Omega}=\frac{\rho_{\rm b}}{2}\log(\frac{\rho_{\rm b}}{2 \pi^2}),\qquad\gamma_{\partial\Omega}=-\Big(\frac{\rho_{\rm b}}{2\pi}\Big)^{1/2}\int_{0}^{\infty}\log\Big( \frac{1 }{ 2} \left(1 + {\rm erf} (y)\right) \Big)\,dy. (1.5) The term \frac{1}{12}\log(N) in ( 1.4 ) is of much significance in the theoretical understanding of the two-dimensional one-component plasma. In fact for general \beta>0 and general two-dimensional geometries, the prediction of Jancovici et al. 1994 —based on a field theory viewpoint of the electric potential and electric field—is that, for large N , \displaystyle\beta F\sim AN+BN^{1/2}+\frac{\chi}{12}\log(N)+\cdots (1.6) where A=\frac{\beta}{4}\log(\pi\rho_{\rm b})+f(\beta) for some f(\beta) , B is proportional to the surface tension \gamma_{\partial\Omega}(\beta) , and \chi the Euler characteristic of the two-dimensional domain (e.g. \chi=1 for a disk, \chi=2 for a sphere, \chi=0 for an annulus). The striking feature of ( 1.6 ) is the predicted universality of the \log(N) term, being independent of \beta but rather distinguishing the underlying topology. The expansion ( 1.6 ) is consistent with the exact result ( 1.5 ), as well as with other known exact results for the sphere Caillol 1981 and annulus Fischmann and Forrester 2011 ; Fischmann 2013 domains in the case \beta=2 . It also agrees with numerical results for the disk and sphere domains for \beta=4,6 , and 8 Téllez and Forrester 1999 ; Téllez and Forrester 2012 . In the case of the disk and annulus, it is expected that the structure of the expansion holds true upon the removal of the confining hard wall implied by \Omega , with the background uniform charge density creating an effective disk or annulus — see the text in the following paragraph. This scenario was, in fact, the one tested numerically for the disk in Téllez and Forrester 1999 ; Téllez and Forrester 2012 . A feature of the Boltzmann factor ( 1.1 ) is that, up to a renormalisation of \rho_{\rm b} , its dependence on the particle coordinates is unchanged by the scalings z_{l}\mapsto cz_{l} ( l=1,\dots,N ) for any c>0 . Making this change of variable (with c=\sqrt{N/\pi\rho_{\rm b}} ) allows ( 1.1 ), in the case \beta=2 , to be identified as an example of the functional form \displaystyle e^{-N\sum_{j=1}^{N}Q(z_{j})}\prod_{1\leq j<k\leq N}|z_{k}-z_{j}|^{2}, (1.7) specifically with Q(z)=|z|^{2} . In fact, in this case and up to a normalisation, ( 1.7 ) corresponds to the eigenvalue probability density function for N\times N standard complex Gaussian matrices (also referred to as matrices from the Ginibre unitary ensemble, GinUE Byun and Forrester 2024 ), scaled by the factor 1/\sqrt{N} . For more general potentials Q , the density ( 1.7 ) can be normalised and interpreted as the eigenvalue probability density function for the class of complex normal random matrices MM^{\dagger}=M^{\dagger}M distributed according to the probability density e^{-N{\rm Tr}\,Q(M)}/Z_{N} , where Z_{N} is the corresponding normalisation constant; see the review Zabrodin 2006 . Note, however, that while the eigenvalues \{z_{j}\}_{j=1}^{N} of the aforementioned normal matrix model are supported on the whole complex plane \mathbb{C} , in the model ( 1.1 ), they are constrained to a finite disk by a hard wall boundary condition. Another remarkable feature of the density ( 1.7 ) with the choice Q(z)=|z|^{2} is that the marginal probability density on one eigenvalue (referred to as 1 -point function or level density) converges, as N\to\infty , to the uniform distribution \frac{1}{\pi}\mathds{1}_{|z|<1} . This is known in the random matrix literature as the circular law and the support of the limiting probability measure—here the unit disk—is referred to as the droplet; see the review Bordenave and Chafaï 2012 . In the rescaled variables, the (abrupt) boundary of the support |z|=1 is precisely the boundary of the confining disk relating to the constrained model ( 1.1 ). Focusing on ( 1.7 ) with each z_{l}\in\mathbb{C} , one can raise it to the power \beta/2 to obtain a generalisation for any inverse temperature \beta>0 . To render it a probability density, one needs to introduce the corresponding normalisation constant—partition function— {Z}_{N,\beta}^{\rm s} (here the superscript “s” stands for soft indicating the absence of a hard wall container, and thus implying that particles are free to occupy regions outside of the droplet, albeit with low probability), explicitly given by \displaystyle{Z}_{N,\beta}^{\rm s}:=\frac{1}{N!}\int_{\mathbb{C}}d^{2}z_{1}\cdots\int_{\mathbb{C}}d^{2}z_{N}\,e^{-\beta N\sum_{j=1}^{N}Q(z_{j})/2}\prod_{1\leq j<k\leq N}|z_{k}-z_{j}|^{\beta}. (1.8) It is a relatively recent result Leblé and Serfaty 2017 (see too Sandier and Serfaty 2015 ) that this partition function, has the large N asymptotic expansion \displaystyle-\log{Z}_{N,\beta}^{\rm s}\sim\frac{\beta}{2}N^{2}I_{\Omega}[\mu_{Q}]+\Big(1-\frac{\beta}{4}\Big)N\log(N)+\bigg(C(\beta)+\Big(1-\frac{\beta}{4}\Big)E_{\Omega}[\mu_{Q}]\bigg)N+{\rm