Journal of Statistical Mechanics Theory and Experiment 2022

19 papers
The 2022 edition of the Journal of Statistical Mechanics: Theory and Experiment focused heavily on the intersection of statistical physics and high-dimensional machine learning. Research themes centered on the behavior of overparameterized models, including investigations into interpolation thresholds, Gaussian universality, and the geometry of benign overfitting. Contributions also explored message passing beyond tree topologies, pseudo-likelihood estimation for large graphical models, and optimal spectral shrinkage for matrix denoising. Prolific authors such as Lenka Zdeborova, Florent Krzakala, and Marc Mezard published work examining computational hardness in disordered systems and the proliferation of support vectors in high dimensions. Other studies addressed practical applications of statistical mechanics, such as crowd dynamic risk management and merging behaviors in large crowds. The year’s output also included theoretical advancements in overcomplete tensor decomposition and the role of locality in overcoming the curse of dimensionality within convolutional teacher-student scenarios. These papers reflect a trend toward using physical principles to understand the limits and performance of complex learning algorithms and high-dimensional data structures.

Top Research Directions

The 30 research directions with Journal of Statistical Mechanics Theory and Experiment 2022 papers, ranked by paper count, next to 2021. Share is the fraction of the edition’s papers in the direction.

#Direction20222021Share 2022Share 20212021 rank
1The Mechanics of Benign Overfitting in High Dimensions2010.5%0.0%—
2Algorithmic Recovery of Overcomplete Tensors105.3%0.0%—
3Asymptotic Decoupling in High Dimensional Estimation105.3%0.0%—
4Bridging the Linear-to-Nonlinear Theoretical Gap105.3%0.0%—
5Concentration Inequalities for High Dimensional Systems105.3%0.0%—
6Convergence of Message Passing on Loopy Graphs115.3%7.1%4
7Correcting for Phylogenetic Confounding in Statistical Inference105.3%0.0%—
8Credit Assignment via Direct Random Error Projection105.3%0.0%—
9Credit Assignment Without Weight Transport105.3%0.0%—
10Deep Networks as Unrolled Optimization Algorithms105.3%0.0%—
11Generalizing Neural Network Guarantees Beyond ReLU105.3%0.0%—
12Generalizing One-Dimensional Models to Higher Dimensions105.3%0.0%—
13How Neural Networks Bypass the Curse of Dimensionality105.3%0.0%—
14Linearizing Deep Learning via the Neural Tangent Kernel105.3%0.0%—
15Logarithmic Sample Complexity in High Dimensional Learning105.3%0.0%—
16Non-Equilibrium Thermodynamics of Self-Propelled Matter115.3%7.1%8
17Overcoming the Gaussian Assumption in High-Dimensional Learning105.3%0.0%—
18Pseudo-likelihood Estimation for High-Dimensional Graphical Models105.3%0.0%—
19Sample Complexity Scaling in Graphical Model Recovery105.3%0.0%—
20The Convergence Cost of Weak Regularization105.3%0.0%—
21The Duality Between Information and Estimation105.3%0.0%—
22The Empirical Validation Gap105.3%0.0%—
23The Gaussian Equivalence Principle in High Dimensional Learning105.3%0.0%—
24The Infinite Width Approximation Gap in Deep Learning105.3%0.0%—
25The Phase Transition Between Benign and Harmful Overfitting105.3%0.0%—

Download all directions as CSV

Institutions

Institutions credited by Journal of Statistical Mechanics Theory and Experiment 2022 papers: a paper credits every institution named in its affiliations, once. 34 institutions share 15 credited papers.

#InstitutionPapersShare20212021 rank
1École Polytechnique Fédérale de Lausanne 960.0%42
2Centre National de la Recherche Scientifique 533.3%41
3École Normale Supérieure - PSL 533.3%0—
4Université Paris Sciences et Lettres 426.7%0—
5Laboratoire de Physique de l'Ecole Normale Supérieure 320.0%0—
6Sorbonne Université 320.0%0—
7Scuola Internazionale Superiore di Studi Avanzati 213.3%0—
8The University of Tokyo 213.3%0—
9Université Paris Cité 213.3%0—
100073twa91 16.7%0—
1104zaaa143 16.7%0—
1205y6rqs46 16.7%0—
13Bocconi University 16.7%0—
14Columbia University 16.7%0—
15Courant Institute of Mathematical Sciences 16.7%0—
16ETH Zurich 16.7%0—
17Forschungszentrum Jülich 16.7%0—
18Georgia Institute of Technology 16.7%0—
19Institute for Advanced Study 16.7%0—
20King's College London 16.7%0—
21Kyoto University 16.7%113
22Kyushu Institute of Technology 16.7%0—
23Lomonosov Moscow State University 16.7%0—
24Massachusetts Institute of Technology 16.7%0—
25Meta (United States) 16.7%0—
26National Academy of Sciences of Ukraine 16.7%0—
27New York University 16.7%0—
28RIKEN Center for Advanced Intelligence Project 16.7%0—
29RWTH Aachen University 16.7%0—
30Santa Fe Institute 16.7%0—

Download all institutions as CSV

About this data

Papers. Journal of Statistical Mechanics Theory and Experiment 2022 lists 19 papers on Lacuna, duplicate versions counted once; rejected and withdrawn submissions are left out.

Directions. Directions are Lacuna’s clustering of the concept elements extracted from each paper, so a paper can belong to several and direction counts add up to more than the paper count. A direction’s share is its papers divided by the edition’s papers. Comparisons with 2021 use shares, since the edition went from 14 to 19 papers.

Institutions. A paper credits each institution named in its affiliations, resolved to a ROR record, once. 15 of 19 papers (78.9%) credit at least one institution and are the base for institution shares; the rest have no resolved affiliation yet.

Data. Counts come from Lacuna’s data build of 2026-11-20 and change as papers are added. Tables as CSV: directions · institutions. Cite as: Lacuna, “Journal of Statistical Mechanics Theory and Experiment 2022”, https://lacuna.tiptreesystems.com/venue/journal-of-statistical-mechanics-theory-and-experiment/2022.

Top Authors

AuthorPapers
Lenka Zdeborova4
Florent Krzakala3
Alessandro Favero2
Bruno Loureiro2
Marc Mezard2
Matthieu Wyart2
Sebastian Goldt2
Alexander van Meegen1
Antoine Maillard1
Daniel Hsu1
David Dahmen1
Francesco Cagnetta1
Gabriele Sicuro1
Mario Geiger1
Mikhail Tamm1
Moritz Helias1
Ruben Ohana1
Stéphane d'Ascoli1
Yoshiyuki Kabashima1

Impactful Papers

Papers from Journal of Statistical Mechanics Theory and Experiment 2022 that shaped research clusters.

Influential

Papers

Title
Aligning random graphs with a sub-tree similarity message-passing algorithm
Align, then memorise: the dynamics of learning with feedback alignment*
Characteristics of merging behavior in large crowds
Coherence-resonance chimeras in coupled HR neurons with alpha-stable Lévy noise
Disordered systems insights on computational hardness
From steady-state TASEP model with open boundaries to 1D Ising model at negative fugacity
Generalization error rates in kernel regression: the crossover from the noiseless to noisy regime*
Information theoretic limits of learning a sparse rule
Ising model selection using ℓ 1-regularized linear regression: a statistical mechanics analysis*
Learning curves of generic features maps for realistic datasets with a teacher-student model*
Locality defeats the curse of dimensionality in convolutional teacher–student scenarios*
Nonlinear analysis of the internal resonance response of an L-shaped beam structure considering quadratic and cubic nonlinearity
On the proliferation of support vectors in high dimensions*
Optimal regularizations for data generation with probabilistic graphical models
Particle dual averaging: optimization of mean field neural network with global convergence rate analysis*
Perturbative construction of mean-field equations in extensive-rank matrix factorization and denoising
Relative stability toward diffeomorphisms indicates performance in deep nets*
Research on crowd dynamic risk management based on the psychological stress perception function
Unified field theoretical approach to deep and recurrent neuronal networks