Challenges
Datasets
Workspaces
Papers
Discussions
Leaderboard
Log inSign up
Challenges
Datasets
Workspaces
Papers
Discussions
Leaderboard
Blog
Job Board
Q3AS

© 2026 Aqora Quantum S.A.S.

TermsPrivacyLegal Notice

arXiv:2407.12603 · Physical review. E

Generalization of the central limit theorem to critical systems: Revisiting perturbation theory

Sankarshan Sahu, Bertrand Delamotte, Adam Rançon

Published Mar 26, 2025• 9 citations
arXivHTMLPDF

Abstract

The central limit theorem does not hold for strongly correlated stochastic variables, as is the case for statistical systems close to criticality. Recently, the calculation of the probability distribution function (PDF) of the magnetization mode has been performed with the functional renormalization group in the case of the three-dimensional Ising model [Balog et al., Phys. Rev. Lett. 129, 210602 (2022)10.1103/PhysRevLett.129.210602]. It has been shown in that article that there exists an entire family of universal PDFs parameterized by ζ=lim_{L,ξ_{∞}→∞}L/ξ_{∞} which is the ratio of the system size L to the bulk correlation length ξ_{∞} with both the thermodynamic limit and the critical limit being taken simultaneously. We show how these PDFs or, equivalently, the rate functions which are their logarithm, can be systematically computed perturbatively in the ε=4-d expansion. We determine the whole family of universal PDFs and show that they are in good qualitative agreement with Monte Carlo data. Finally, we conjecture on how to significantly improve the quantitative agreement between the one-loop and the numerical results.

DOI
10.1103/physreve.111.034128

Related papers

Critical probability distributions of the order parameter at two loops: I. Ising universality class

Sankarshan Sahu

Abstract There exists an entire family of universal probability distribution functions (PDFs) of the magnetization mode of the three dimensional Ising model parameterized by ζ = lim L , ξ ∞ L / ξ ∞ which is the ratio of the system size L to the bulk correlation length ξ ∞ with both the thermodynamic limit and the critical limit being taken simultaneously at fixed ζ . Recently, the PDFs of the magnetization mode of the three-dimensional Ising model has been computed at one-loop in the ϵ = 4 − d expansion (2024 arXiv preprint arXiv:2407.12603 ). 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.

arXiv:2501.18615•Journal of Statistical Mechanics Theory and Experiment•Dec 1, 2025•6 citations

Critical probability distributions of the order parameter at two loops: II. O ( n ) universality class

Sankarshan Sahu

Abstract We show how to compute the probability distributions of the order parameter of the O ( n ) model at two-loop order of perturbation theory generalizing the methods developed for computing the same in case of the Ising model (Sahu 2025 J. Stat. Mech. 2025 123202). We show that even for the O ( n ) model, there exists not one but a family of these probability distribution functions (PDFs) indexed by ζ which is the ratio of system size L to the bulk correlation length ξ ∞ . We also compare these PDFs to the Monte-Carlo simulations and the existing FRG results (Rançon et al 2025 arXiv:2501.04465 [cond-mat.statmech]) for the O (2) and O (3) models.

arXiv:2503.21799•Journal of Statistical Mechanics Theory and Experiment•Feb 1, 2026•1 citation

Critical Probability Distributions of the Order Parameter from the Functional Renormalization Group

Ivan Balog, Adam Rançon, Bertrand Delamotte

We show that the functional renormalization group (FRG) allows for the calculation of the probability distribution function of the sum of strongly correlated random variables. On the example of the three-dimensional Ising model at criticality and using the simplest implementation of the FRG, we compute the probability distribution functions of the order parameter or, equivalently, its logarithm, called the rate functions in large deviation theory. We compute the entire family of universal scaling functions, obtained in the limit where the system size L and the correlation length of the infinite system ξ_{∞} diverge, with the ratio ζ=L/ξ_{∞} held fixed. It compares very accurately with numerical simulations.

arXiv:2206.03769•Physical Review Letters•Nov 16, 2022•21 citations

Probability distributions of the order parameter of the O ( N ) model

Adam Rançon, Bertrand Delamotte, Lovro Šaravanja +1 more

We study the probability distribution function (PDF) of the order parameter of the three-dimensional O(N) model at criticality using the functional renormalization group. For this purpose, we generalize the method introduced by Balog et al. [Phys. Rev. Lett. 129, 210602 (2022)0031-900710.1103/PhysRevLett.129.210602] to the O(N) model. We study the large N limit, as well as the cases N=2 and N=3 at the level of the local potential approximation (LPA), and compare our results to Monte Carlo simulations. We compute the entire family of universal scaling functions, obtained in the limit where the system size L and the correlation length of the infinite system ξ_{∞} diverge, with the ratio ζ=L/ξ_{∞} constant. We also generalize our results to the approach of criticality from the low-temperature phase where another infinite family of universal PDF exists. We find that the LPA describes very well the functional form of the family of PDFs, once we correct for a global amplitude of the (logarithm of the) PDF and of ζ.

arXiv:2501.04465•Physical review. E•Mar 26, 2025•8 citations

Hidden universality of rare fluctuations and a central-limit structure at criticality

Nikolaos G. Fytas, Víctor Martín-Mayor, Attilio L. Stella +3 more

The central limit theorem is among the most fundamental principles in the natural sciences, governing the fluctuations of sums of weakly correlated random variables through Gaussian statistics. At critical points, however, long-range correlations invalidate this paradigm, raising a fundamental question: does any general limit theorem survive at criticality? Although scale invariance has long been recognised as the defining feature of typical critical fluctuations, the behaviour of rare fluctuations has remained largely unexplored, as their extraordinarily small probabilities place them beyond the reach of conventional numerical methods. Recent theoretical advances suggest that universal asymptotic laws {may govern large deviations in systems at criticality}, yet direct numerical evidence has been lacking. Here we provide the first such evidence in the two-dimensional Ising universality class. Using large-scale Tethered Monte Carlo simulations, we probe order-parameter fluctuations with probabilities as small as $10^{-200}$ and show that, once normalised by their characteristic width, the far tails collapse onto a single asymptotic law shared by distinct microscopic models in the same universality class. This collapse is governed by a universal amplitude, whose numerical extraction is compatible with the conjectured exact result we derive from conformal field theory, and points towards a generalised central limit theorem for critical fluctuations.

arXiv:2611.14621•Nov 10, 2026•0 citations

Are critical finite-size scaling functions calculable from knowledge of an appropriate critical exponent?

Rudolf Hilfer A, Nigel B. Wilding

Critical finite-size scaling functions for the order-parameter distribution of the two- and three-dimensional Ising model are investigated. Within a recently introduced classification theory of phase transitions the universal part of critical finite-size scaling functions has been derived by employing a scaling limit which differs from the traditional finite-size scaling limit. In this paper the analytical predictions are compared with Monte Carlo simulation results. We find good agreement between the analytical expression and the simulation results. The agreement is consistent with the possibility that the functional form of the critical finite-size scaling function for the order-parameter distribution is determined uniquely by only a few universal parameters, most notably the equation of state exponent.

arXiv:cond-mat/9503157•Journal of Physics A Mathematical and General•May 21, 1995•62 citations
The central limit theorem does not hold for strongly correlated stochastic variables, as is the case for statistical systems close to criticality. Recently, the calculation of the probability distribution function (PDF) of the magnetization mode has been performed with the functional renormalization group in the case of the three-dimensional Ising model [Balog et al., Phys. Rev. Lett. 129, 210602 (2022)10.1103/PhysRevLett.129.210602]. It has been shown in that article that there exists an entire family of universal PDFs parameterized by ζ=lim_{L,ξ_{∞}→∞}L/ξ_{∞} which is the ratio of the system size L to the bulk correlation length ξ_{∞} with both the thermodynamic limit and the critical limit being taken simultaneously. We show how these PDFs or, equivalently, the rate functions which are their logarithm, can be systematically computed perturbatively in the ε=4-d expansion. We determine the whole family of universal PDFs and show that they are in good qualitative agreement with Monte Carlo data. Finally, we conjecture on how to significantly improve the quantitative agreement between the one-loop and the numerical results.
Abstract There exists an entire family of universal probability distribution functions (PDFs) of the magnetization mode of the three dimensional Ising model parameterized by ζ = lim L , ξ ∞ L / ξ ∞ which is the ratio of the system size L to the bulk correlation length ξ ∞ with both the thermodynamic limit and the critical limit being taken simultaneously at fixed ζ . Recently, the PDFs of the magnetization mode of the three-dimensional Ising model has been computed at one-loop in the ϵ = 4 − d expansion (2024 arXiv preprint arXiv:2407.12603 ). 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.
Abstract We show how to compute the probability distributions of the order parameter of the O ( n ) model at two-loop order of perturbation theory generalizing the methods developed for computing the same in case of the Ising model (Sahu 2025 J. Stat. Mech. 2025 123202). We show that even for the O ( n ) model, there exists not one but a family of these probability distribution functions (PDFs) indexed by ζ which is the ratio of system size L to the bulk correlation length ξ ∞ . We also compare these PDFs to the Monte-Carlo simulations and the existing FRG results (Rançon et al 2025 arXiv:2501.04465 [cond-mat.statmech]) for the O (2) and O (3) models.
We show that the functional renormalization group (FRG) allows for the calculation of the probability distribution function of the sum of strongly correlated random variables. On the example of the three-dimensional Ising model at criticality and using the simplest implementation of the FRG, we compute the probability distribution functions of the order parameter or, equivalently, its logarithm, called the rate functions in large deviation theory. We compute the entire family of universal scaling functions, obtained in the limit where the system size L and the correlation length of the infinite system ξ_{∞} diverge, with the ratio ζ=L/ξ_{∞} held fixed. It compares very accurately with numerical simulations.
We study the probability distribution function (PDF) of the order parameter of the three-dimensional O(N) model at criticality using the functional renormalization group. For this purpose, we generalize the method introduced by Balog et al. [Phys. Rev. Lett. 129, 210602 (2022)0031-900710.1103/PhysRevLett.129.210602] to the O(N) model. We study the large N limit, as well as the cases N=2 and N=3 at the level of the local potential approximation (LPA), and compare our results to Monte Carlo simulations. We compute the entire family of universal scaling functions, obtained in the limit where the system size L and the correlation length of the infinite system ξ_{∞} diverge, with the ratio ζ=L/ξ_{∞} constant. We also generalize our results to the approach of criticality from the low-temperature phase where another infinite family of universal PDF exists. We find that the LPA describes very well the functional form of the family of PDFs, once we correct for a global amplitude of the (logarithm of the) PDF and of ζ.
The central limit theorem is among the most fundamental principles in the natural sciences, governing the fluctuations of sums of weakly correlated random variables through Gaussian statistics. At critical points, however, long-range correlations invalidate this paradigm, raising a fundamental question: does any general limit theorem survive at criticality? Although scale invariance has long been recognised as the defining feature of typical critical fluctuations, the behaviour of rare fluctuations has remained largely unexplored, as their extraordinarily small probabilities place them beyond the reach of conventional numerical methods. Recent theoretical advances suggest that universal asymptotic laws {may govern large deviations in systems at criticality}, yet direct numerical evidence has been lacking. Here we provide the first such evidence in the two-dimensional Ising universality class. Using large-scale Tethered Monte Carlo simulations, we probe order-parameter fluctuations with probabilities as small as 10−20010^{-200}10−200 and show that, once normalised by their characteristic width, the far tails collapse onto a single asymptotic law shared by distinct microscopic models in the same universality class. This collapse is governed by a universal amplitude, whose numerical extraction is compatible with the conjectured exact result we derive from conformal field theory, and points towards a generalised central limit theorem for critical fluctuations.
Critical finite-size scaling functions for the order-parameter distribution of the two- and three-dimensional Ising model are investigated. Within a recently introduced classification theory of phase transitions the universal part of critical finite-size scaling functions has been derived by employing a scaling limit which differs from the traditional finite-size scaling limit. In this paper the analytical predictions are compared with Monte Carlo simulation results. We find good agreement between the analytical expression and the simulation results. The agreement is consistent with the possibility that the functional form of the critical finite-size scaling function for the order-parameter distribution is determined uniquely by only a few universal parameters, most notably the equation of state exponent.