Critical Probability Distributions of the order parameter at two loops I: Ising universality class
Sankarshan Sahu · 2025
Read Critical Probability Distributions of the order parameter at two loops I: Ising universality class by Sankarshan Sahu (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-2501-18615.txt.
Introduction
Sankarshan Sahu computes the spread of possible magnetization values in a model magnet at its critical point in this 2025 physics paper. The model is the three-dimensional Ising model, a standard stand-in for magnets and fluids near a phase change. At criticality, the average magnetization follows a universal probability curve that depends on the ratio of system size to correlation length.
Earlier work had computed that curve to first order in a standard expansion scheme. Sahu pushes the calculation to second order for the whole family of curves. He shows the second-order curves match computer simulation data far better than the first-order ones did. The paper is the first in a planned series on these critical distributions. Sharper curves of this kind help physicists test theory against simulation and experiments on real materials near their critical points.
Catalog record
| CREATOR | Sankarshan Sahu |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 98 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2501-18615
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Critical Probability Distributions of the order parameter at two loops I: Ising universality class Sankarshan Sahu Abstract There exists an entire family of universal PDFs of the magnetization mode of the three dimensional Ising model parameterized by ζ = lim L , ξ ∞ L / ξ ∞ \zeta=\lim_{L,\xi_{\infty}}L/\xi_{\infty} which is the ratio of the system size L L to the bulk correlation length ξ ∞ \xi_{\infty} with both the thermodynamic limit and the critical limit being taken simultaneously at fixed ζ \zeta . Recently, the probability distribution functions (PDFs) of the magnetization mode of the three-dimensional Ising model has been computed at one-loop in the ϵ = 4 − d \epsilon=4-d expansion [ arXiv preprint arXiv:2407.12603 (2024) ]. We show how these PDFs or, equivalently, the rate functions which are their logarithm, can be systematically computed at second order of the perturbative expansion. We compute the whole family of universal rate-functions and show that their agreement with the Monte Carlo data improves significantly at this order when compared to their one-loop counterpart. I Introduction The speculation of the relationship between Renormalization Group and Probability Theory goes back at least as far as 1973 where Bleher and Sinai [ 1 ] hinted towards it in the context of hierarchical models which are simplified versions of the Ising and O (n) models. This idea was further extended by Jona-Lasinio in [ 2 ] , where he further established the connection between limiting theorems in probability theory and renormalization group. An example of such a limiting theorem is the Central Limit Theorem (CLT). The Central Limit Theorem (CLT) is of crucial importance in many key areas of research in physics, mathematics, and quantitative finance. According to this theorem, for a large number N of identically distributed independent random variables \hat{\sigma}_{i} , their sum S=\sum_{i}\hat{\sigma}_{i} has fluctuations of order \sqrt{N} and the asymptotic probability distribution of the properly normalized variable S/\sqrt{N} is a Gaussian law with finite variance if both the mean and the variance of the \hat{\sigma}_{i} ’s are finite. Most importantly, the resulting distribution (here, the Gaussian law) is universal, that is, independent of the probability distribution from which the variables \hat{\sigma}_{i} ’s are chosen [except for the mean and variance]. The CLT itself has been generalized to the cases where the variables \hat{\sigma}_{i} are weakly correlated (instead of being completely independent) [ 3 , 4 ] . Physically, the picture becomes that of finite clusters of \hat{\sigma}_{i} of size equal to their correlation length (say \xi ) which are strongly correlated within the cluster, but these clusters again being independent of each other. However, no such limiting theorem exists in case of strongly correlated systems, that is, for systems where all the \hat{\sigma}_{i} ’s are algebraically correlated. In our previous article [ 5 ] , we formulated a systematic perturbative approach to these limiting distributions for strongly correlated systems using the \epsilon=4-d expansion and computed these PDFs at one-loop order for the O(n) and Ising models. A calculation of these PDFs already exists in a non-perturbative setting or more precisely at the leading order of the Functional Renormalization Group FRG calculations (LPA approximation) [ 6 ] . The goal of the current article is to go to the next order of the \epsilon=4-d expansion, thereby performing a two-loop calculation of the PDFs. We do this once again in the context of the well known Ising model in three dimensions. The generalization to the O(n) case is not as trivial as in the one-loop calculation and is beyond the scope of the current work. Before going further we must clarify what is to be gained from studying these probability distributions beyond mere academic curiosity. One of the greatest successes of Renormalization Group was the computation of universal critical exponents and critical amplitude ratios to striking accuracy in both the perturbative as well as non-perturbative setting. Most of the time, therefore, the characterization of the different universality classes is carried out using the values of the critical exponents, that is, by two numbers since only two of them are independent, at least as far as the leading scaling behavior is concerned. However, for comparison with numerical simulations and experiments, it is much better to have universal functions rather than a small amount of numbers. There are two kinds of universal functions close to criticality, those that depend on wavenumbers, this is the case of the universal critical structure factor, and those that are field dependent such as universal finite-size scaling functions routinely computed in numerical simulations. We compute below one such function, namely the critical probability distribution function. Notice that both the computation and the measure of the critical structure factors are delicate whereas the critical PDFs are not as shown below. One must keep in mind that although universal, these critical PDFs have a dependence on boundary conditions 1 1 1 Here only regular isotropic lattices have been considered, or more precisely all isotropic lattices with same boundary conditions will yield the same PDF. . In this paper, we only consider periodic boundary conditions. In our previous work [ 5 ] and in [ 6 , 7 ] , it has already been shown that taking the critical limit, T\to T_{c} , and the thermodynamic limit simultaneously, one does not end up with just one PDF but a family of PDFs indexed by \zeta=L/\xi_{\infty} , with L being the system size and \xi_{\infty} being the correlation length of the system depending on the way, the critical limit, and the thermodynamic limit is approached. Although the one-loop results are qualitatively good, they are very far from being quantitatively accurate. In the current article we show how going to two loops actually brings us to the level of quantitative accuracy when compared with the Monte-Carlo simulations, thereby tremendously improving the one-loop results. II Field Theoretic Formalism In the following, we consider the Ising model described by the Hamiltonian: \mathcal{H}=-J\sum_{<ij>}\hat{\sigma}_{i}\hat{\sigma}_{j}, (1) with coupling constant J>0 , and spins \hat{\sigma}_{i}=\pm 1 where \langle ij\rangle represent the nearest-neighbor interactions on a hyper cubic d dimensional lattice with length L and periodic boundary conditions. When L\to\infty , a second order phase transition occurs in the Ising model for some critical temperature T_{c} and is characterized by the divergence of the correlation length \xi_{\infty} , i.e. at a temperature T=T_{c} , \xi_{\infty}\rightarrow\infty , which is the hallmark of strong correlations. What we are actually interested in is the PDF of the total normalized spin defined as \hat{s}=L^{-d}\sum_{i}\hat{\sigma}_{i} , the average of which is the magnetization m=\langle\hat{s}\rangle . We represent this quantity by \hat{P}(\hat{s}=s) . The fluctuations of \hat{s} are measured by the quantity \langle\hat{s}^{2}\rangle=L^{-d}\chi with \chi the magnetic susceptibility. For a weakly correlated system, i.e. T\neq T_{c} and in the limit of infinite volume ( L\rightarrow\infty ), one recovers (a generalization of) the Central Limit Theorem, i.e. we find {\hat{P}(\hat{s}=s)}\propto\exp{\left(-\frac{L^{d}s^2}{2\chi}\right)} . However, the same fails to hold in case of strongly correlated systems where \chi scales with L as \chi\propto L^{2-\eta} (with \eta the anomalous dimension). This ultimately leads to the failure of CLT in the limit L\rightarrow\infty and at T=T_{c} . Thus what we will be mostly interested in is not just any PDF but the critical