Reconstructing the Probability Measure of a Multi-group Curie-Weiss Model with Interacting Groups

arXiv:2607.21662 · physics.soc-ph, math.PR, math.ST, stat.TH · Submitted 2026-07-23 · Read on arXiv

Listen

Radio episode about this paper

Transcript

Introduction to the show: ident: Physics. The day in general physics: instruments, methods and results across the field.

Kai: We're looking at the paper: "Reconstructing the Probability Measure of a Multi-group Curie-Weiss Model with Interacting Groups".

Title and authors: Kai: Now, let's look at what this paper actually sets out to do based on its title and authors. It’s about reconstructing the probability measure of this multi-group Curie-Weiss model with interacting groups.

Mira: The authors are Miguel Ballesteros, Gerardo Franco Cordova, Fedro Guillen, and Gabor Toth. They're bringing together statistical mechanics and statistical inference to look at how group behavior is shaped by these interactions.

Lev: So they’re moving beyond the basic mean-field idea to handle situations where you have distinct subpopulations that aren't just randomly mixed.

Kai: Right, and what they are doing is focusing on using sample voting behavior—the real data we observe—to infer those underlying interaction structures, like the coupling matrices.

Mira: The authors are trying to provide some rigorous methodological tools for analyzing and designing voting systems that actually reflect how groups behave in the real world.

Lev: That sounds like a practical goal, taking these abstract statistical physics ideas and making them useful for things like policy design or social science modeling.

The paper's summary: Kai: The summary of the paper focuses on defining two main estimators for that coupling matrix J based on what we observe in the sample data.

Mira: They introduce an estimator called J hat N1 N2 n, which they set up based on a condition involving T of x being invertible and I minus N T of x inverse N being positive.

Lev: That’s where the complexity starts—you have to check if that matrix T is invertible and if that inequality holds for every possible sample configuration x.

Kai: And then they have another estimator, m hat N1 N2 n, which they use when the first condition doesn't hold at all, setting it to some scaled version of T of x minus one divided by n times N inverse.

Mira: So essentially, if the system is in a certain regime defined by those conditions, you use one estimator; otherwise, you switch to the other one.

Lev: It seems like they are trying to create a robust way to estimate the parameters even when the system isn't perfectly behaving according to their initial assumptions.

The paper's improvements: Kai: The main theoretical contribution is Theorem eleven which analyzes how these estimators behave in two very different regimes: high temperature and low temperature <ref:2607.21662#pg1>.

Mira: In the high temperature regime, they show that the estimator J hat N1 N2 n converges to the true coupling matrix J as n goes to infinity.

Lev: That convergence is tied to a continuous mapping theorem they use, which shows that if you look at the system in a certain way, you can map it onto something simpler and analyze its behavior there.

Kai: But then for the low temperature regime, things change; instead of converging directly to J hat N1 N2, we get convergence to a concentration point called m hat N1 N2.

Mira: That concentration point m hat converges to something specific based on the group sizes, where m one and m two depend on the group sizes N one and N two in a specific way <ref:2607.21662#pg1>.

Lev: What this means for real-world data is that when interactions are very strong, like in a low temperature setting, we don't just get a direct estimate of J; we get this average behavior defined by those concentration points.

Conclusion: Kai: So to wrap up on the main findings of "Reconstructing the Probability Measure of a Multi-group Curie-Weiss Model with Interacting Groups," they’ve shown how these estimators work in both the high and low temperature regimes.

Mira: The simulation results back up the theory, showing that their methods correctly identify which regime is active about seventy-three percent of the time for a specific coupling matrix.

Lev: And for the low temperature case, they found that m hat converges to a specific point in one hundred percent of the cases when you look at groups where N one and N two are large <ref:2607.21662#pg1>.

Kai: It gives us concrete tools to move from just theory to actually designing voting systems that reflect how groups really interact under different levels of social cohesion.

Mira: It’s a way to take concepts from statistical mechanics and apply them directly to questions about representation and fairness in social contexts, using these rigorous estimation techniques.

Lev: For the hardware side, this tells us what kind of convergence we can expect when dealing with real-world data that has those complex group interactions baked into it.

Miguel Ballesteros, Gerardo Franco Cordova, Fedro Guillen, Gabor Toth

IIMAS-UNAM

physics.soc-ph, math.PR, math.ST, stat.TH

Submitted: 2026-07-23

Updated: 2026-11-17

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

Importance score: 77/100

The gist: Reconstructing the probability measure of a multi-group Curie-Weiss model with interacting groups is studied to provide methodological tools for analyzing and designing voting systems that reflect…

Key concepts

Multi-group Curie-Weiss Model
This is a statistical model used to describe voting behavior across different groups. It involves partitioning the population into two or more groups, where voters within and between these groups interact based on a coupling matrix J. The goal is to understand how these interactions influence overall voting patterns.
Coupling Matrix (J)
The coupling matrix J defines the strength of interaction between different voters in the model. It dictates how the opinions or votes of one group affect another. Estimating this matrix is crucial for understanding the underlying social structure and decision-making processes.
Asymptotic Estimators
These are mathematical formulas used to estimate parameters when dealing with very large populations (large N). Instead of complex calculations, these estimators use sample observations to provide reliable approximations of the true interaction parameters, especially in regimes where exact calculations are too difficult.

Terminology

Summary

Reconstructing the probability measure of a multi-group Curie-Weiss model with interacting groups is studied to provide methodological tools for analyzing and designing voting systems that reflect empirically observed patterns of group behavior

Model Formulation

The paper introduces a multi-group Curie-Weiss model where the population is partitioned into two groups, each with size N1 and N2, and the interaction between voters is governed by a coupling matrix J The Hamiltonian H for the voting configuration is defined as H (x11,..., x1N1,..., x2N2):= - 1/2 X T J X The probability measure P is given by PJ(X1,i=1 = x1,i,. p. 0.., X2,N2 = x2,N2):= Z(-1/2) exp(-H(x11,. p. 2.., x2N2))

Estimation Statistics

The study defines two main estimators for the coupling matrix J based on sample observations

Definition 9

The estimator JˆN1,N2,n: Ω n N1+N2 → R(2×2) is defined as follows: for all x ∈ Ω n N1+N2 such that T(x) is invertible and I − NT(x)-1 N > 0 holds, we set JˆN1,N2,n (x) = I − NT(x)-1 N For all other x ∈ Ω n N1+N2, we set JˆN1,N2,n (x):= l

Definition 10

The estimator ˆmN1,N2,n: Ω n N1+N2 → R(2×2) is defined for all x ∈ Ω n N1+N2 such that T(x) is singular or I − NT(x)-1 N > 0 does not hold by mˆ N1,N2,n (x):= (N−1)(-1) T(x) N(-1) For all other x ∈ Ω n N1+N2, we set ˆmN1,N2,n (x):= 0

Asymptotic Results

Theorem 11 analyzes the asymptotic behavior of the estimators under two regimes:

High Temperature Regime (I − J > 0)

The estimator converges to the true coupling matrix in the limit n → ∞, specifically JˆN1,N2,n p−−−−→ n→∞ J˜N1,N2 and J˜N1,N2 −−−−−−−→ N1,N2→∞ J

Low Temperature Regime (I − J ≥ 0 not hold)

The estimator converges to a concentration point (m1, m2) in the limit N1,N2 → ∞, where m˜ N1,N2 −−−−−−−→ N1,N2→∞ m2 1 m1m2

Simulation and Validation

Numerical simulations were conducted to assess the likelihood of the estimator being in the wrong regime and its deviation from true parameter values

High Temperature Regime Simulation

For a high temperature coupling matrix J = [[0.6, 0.3], [0.3, 0.5]], the regime was correctly identified as high temperature from 73% of the samples The estimate JˆN1,N2,n (x) lay within a tolerance of 0. p. 51 of the true coupling matrix in (4) 20% of the time

Low Temperature Regime Simulation

For a low temperature coupling matrix J = [[2, 1], [1, 1.8]], the low temperature regime was correctly identified in all cases The estimator (mˆ) squared 1,(mˆ) squared 2 was realised within a tolerance of 0. p. 61 of the true value m2 1, m2 2 in 100% of the time

Proof Structure

The proof of Theorem 11 relies on Proposition 8 which relates expectations to the inverse matrix (I − J)-1 for large populations The continuous mapping theorem is applied using Theorem 14 to show that the estimators converge to the true parameters in the high temperature regime For the low temperature regime, a central limit theorem is used on the statistic T (X) to establish convergence of mˆ N1,N2,n p−−−−→ n→∞ m˜ N1,N2 The limiting covariance matrix C is derived using Theorem 14 and the asymptotic results for T (X) The proof also shows that the convergence of Jˆ to J is governed by the Jacobian matrix Delta(µ) and the asymptotic variance Σ

References

[1] Alexander P. Alodjants, A. p. 9 p. 9 Yu. Bazhenov, A. p. 9 p. 9 Yu. Khrennikov, and A. p. 9 V. Bukhanovsky Mean-field theory of social laser Sci Rep. p. 9 p. 9, 12(8566), 2022

[3] Miguel Ballesteros Rams´es H Mena Arno Siri-Jegousse and Gabor Toth Detection of an arbitrary number of communities in a block spin ising model PLoS One, 21(3), 2026 arXiv:2311.18112

[4] Miguel Ballesteros Rams´es H Mena Arno Siri-Jegousse and Gabor Toth Reconstruction of the probability measure and the coupling parameters in a curie-weiss model 2025 arXiv:2505.21778

[5] Miguel Ballesteros Ivan Naumkin and Gabor Toth Approximation techniques for the reconstruction of the probability measure and the coupling parameters in a Curie-Weiss model for large populations 2025 arXiv:2507.17073

[6] Miguel Ballesteros Ivan Naumkin and Gabor Toth Reconstructing the probability measure of a CurieWeiss model observing the realisations of a subset of spins 2025 arXiv:2508.03452

[7] Quentin Berthet Philippe Rigollet and Piyush Srivastava Exact recovery in the ising blockmodel Ann Statist., 47(4):1805–1834, 2019

[8] Jean-Yves Le Boudec David McDonald and Jochen Mundinger A generic mean field convergence result for systems of interacting objects In Fourth International Conference on the Quantitative Evaluation of Systems (QEST 2007), pages 3–18, 2007

[9] William A Brock and Steven N Durlauf Discrete choice with social interactions Rev. Econ. Stud., 68(2):235–260, 2001

[10] Pierluigi Contucci Stefano Ghirlanda Modelling society with statistical mechanics An application to cultural contact and immigration Qual. Quant., 41:569–578, 2007

[11] Hung T Diep Miron Kaufman and Sanda Kaufman An agent-based statistical physics model for political polarization A monte carlo study Entropy, 25(7), 2023

[12] Micaela Fedele Pierluigi Contucci Scaling limits for multi-species statistical mechanics mean-field models J. Stat. Phys., 144:1186–1205, 2011

[13] Micaela Fedele Cecilia Vernia and Pierluigi Contucci Inverse problem robustness for multi-species mean-field spin models J. Phys. A: Math. Theor., 46(6), 2013

[14] Ignacio Gallo Adriano Barra and Pierluigi Contucci Parameter evaluation of a simple mean-field model of social interaction Math Models Methods Appl. Sci., 19(supp01):1427–1439, 2009

[15] Ernst Ising Beitrag zur Theorie des Ferromagnetismus Zeitschrift fur Physik ¨, 31:253–258, 1925

[16] Boyan Jovanovic and Robert W Rosenthal Anonymous sequential games J. Math. Econ., 17(1):77–87, 1988

[17] Pieter Willem Kasteleijn Statistical Problems in Ferromagnetism Antiferromagnetism and Adsorption 1956

[20] Werner Kirsch Gabor Toth Optimal weights in a two-tier voting system with mean-field voters Soc. Choice Welf. p. 3 p. 3, 2025

[21] Holger Knöpfel Matthias Löwe Kristina Schubert and Arthur Sinulis Fluctuation results for general block spin Ising models J. Stat. Phys., 178:1175–1200, 2020

[23] Lars Onsager Crystal statistics I A two-dimensional model with an order-disorder transition Phys. Rev., 65(3-4):117–149, 1944

[25] Adrianus Willem Van der Vaart Asymptotic Statistics Cambridge University Press, 1 edition, 1998

--- Page 6 ---

The paper is a scientific analysis of the statistical estimation of coupling parameters in a multi-group Curie-Weiss model to provide rigorous methodological tools for analyzing and designing voting systems that reflect empirically observed patterns of group behavior.

Pruning redundant reasoning tokens

The investigation focuses on reconstructing the probability measure of a multi-group Curie-Weiss model from sample voting behavior The complexity arises because the model has a larger number of coupling parameters compared to single-group models While maximum likelihood estimators have desirable theoretical properties, computational challenges necessitate an estimator based on asymptotic approximations valid for large populations

How the training signal is built

The core of the problem involves defining statistics T(x) which captures the voting configuration across groups Definition 6 defines this statistic as T (x):= 1/n X t=1 (PN1 i=1 x(t) 1,i2 PN1 i=1 x(t) 1,i PN2 j=1 x(t) 2,j PN2 j=1 x(t) 2,j2) The definitions of J˜N1,N2 and m˜ N1,N2 depend on the group sizes Nλ and the group voting margins Sλ

Where the gains come from

The theoretical contribution is Theorem 11 which analyzes the statistical properties of these estimators The results show that JˆN1,N2,n p−−−−→ n→∞ J˜N1,N2 and J˜N1,N2 −−−−−−−→ N1,N2→∞ J Furthermore, the simulation results validate the regime identification accuracy and parameter estimation performance across different sample sizes

Conclusion

The work contributes to an expanding body of research by extending inference methods to correlated voting across group boundaries using multi-group Curie-Weiss models It provides rigorous methodological tools for analyzing and designing voting systems that reflect empirically observed patterns of group behavior complemented by numerical simulations The presentation aims to remain accessible by combining mathematical precision with conceptual exposition

[1] Alexander P. Alodjants, A. Yu. Bazhenov, A. Yu. Khrennikov and A. p. 9 V. Bukhanovsky Mean-field theory of social laser Sci Rep., 12(8566), 2022

[2] Fran¸cois Baccelli Fridrikh I Karpelevich M Ya Kelbert Anatolii A Puhalskii A N Rybko and Yuri M Suhov A mean-field limit for a class of queueing networks J. Stat. Phys., 66:803–825, 1992

[3] Miguel Ballesteros Rams´es H Mena Jos´e Luis Perez and Gabor Toth Detection of an arbitrary number of communities in a block spin ising model PLoS One, 21(3), 2026 arXiv:2311.18112

[4] Miguel Ballesteros Rams´es H Mena Arno Siri-Jegousse and Gabor Toth Reconstruction of the probability measure and the coupling parameters in a curie-weiss model 2025 arXiv:2505.21778

[5] Miguel Ballesteros Ivan Naumkin and Gabor Toth Approximation techniques for the reconstruction of the probability measure and the coupling parameters in a Curie-Weiss model for large populations 2025 arXiv:2507.17073

[6] Miguel Ballesteros Ivan Naumkin and Gabor Toth Reconstructing the probability measure of a CurieWeiss model observing the realisations of a subset of spins 2025 arXiv:2508.03452

[7] Quentin Berthet Philippe Rigollet and Piyush Srivastava Exact recovery in the ising blockmodel Ann Statist., 47(4):1805–1834, 2019

[8] Jean-Yves Le Boudec David McDonald and Jochen Mundinger A generic mean field convergence result for systems of interacting objects In Fourth International Conference on the Quantitative Evaluation of Systems (QEST 2007), pages 3–18, 2007

[9] William A Brock and Steven N Durlauf Discrete choice with social interactions Rev. Econ. Stud., 68(2):235–260, 2001

[10] Pierluigi Contucci Stefano Ghirlanda Modelling society with statistical mechanics An application to cultural contact and immigration Qual. Quant., 41:569–578, 2007

[11] Hung T Diep Miron Kaufman and Sanda Kaufman An agent-based statistical physics model for political polarization A monte carlo study Entropy, 25(7), 2023

[12] Micaela Fedele Pierluigi Contucci Scaling limits for multi-species statistical mechanics mean-field models J. Stat. Phys., 144:1186–1205, 2011

[13] Micaela Fedele Cecilia Vernia and Pierluigi Contucci Inverse problem robustness for multi-species mean-field spin models J. Phys. A: Math. Theor., 46(6), 2013

[14] Ignacio Gallo Adriano Barra and Pierluigi Contucci Parameter evaluation of a simple mean-field model of social interaction Math Models Methods Appl. Sci., 19(supp01):1427–1439, 2009

[15] Ernst Ising Beitrag zur Theorie des Ferromagnetismus Zeitschrift fur Physik ¨, 31:253–258, 1925

[16] Boyan Jovanovic and Robert W Rosenthal Anonymous sequential games J. Math. Econ., 17(1):77–87, 1988

[17] Pieter Willem Kasteleijn Statistical Problems in Ferromagnetism Antiferromagnetism and Adsorption 1956

[20] Werner Kirsch Gabor Toth Optimal weights in a two-tier voting system with mean-field voters Soc. Choice Welf., 2025

[21] Holger Kn¨opfel Matthias L¨owe Kristina Schubert and Arthur Sinulis Fluctuation results for general block spin Ising models J. Stat. Phys., 178:1175–1200, 2020

[23] Lars Onsager Crystal statistics I A two-dimensional model with an order-disorder transition Phys. Rev., 65(3-4):117–149, 1944

[25] Adrianus Willem Van der Vaart Asymptotic Statistics Cambridge University Press, 1 edition, 1998

--- Page 7 ---

The paper is a scientific analysis of the statistical estimation of coupling parameters in a multi-group Curie-Weiss model to provide rigorous methodological tools for analyzing and designing voting systems that reflect empirically observed patterns of group behavior.

Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.

Improvements for AI systems

  1. Enhanced parameter estimation for complex social systems by using asymptotic approximations to estimators based on large population limits, making applications to larger populations computationally feasible where maximum likelihood estimation is impractical due to exponential complexity of the partition function.

  2. Ability to accurately infer interaction structures, such as coupling matrices J or concentration points m1, m2, in multi-group models even in the low temperature regime where strong interactions occur between voters.

  3. Robust detection of the operating regime (high vs low temperature) for a given coupling matrix by assessing the probability that the estimator is realized in the wrong regime, improving reliability across different population sizes.

  4. Precise quantification of estimation error by providing asymptotic results for estimators, such as showing that sqrt(n)(Jˆ - J˜) converges to N(0, C) and sqrt(n)(mˆ - m˜) converges to N(0, D).

  5. Application of the Delta Method to provide the limiting covariance matrix C = Delta (mu) Sigma Delta (mu) T for estimating coupling parameters in the high temperature regime when I - J > 0.

Sources

Related papers