Papers
Topics
Authors
Recent
Detailed Answer
Quick Answer
Concise responses based on abstracts only
Detailed Answer
Well-researched responses based on abstracts and relevant paper content.
Custom Instructions Pro
Preferences or requirements that you'd like Emergent Mind to consider when generating responses
Gemini 2.5 Flash
Gemini 2.5 Flash 71 tok/s
Gemini 2.5 Pro 52 tok/s Pro
GPT-5 Medium 18 tok/s Pro
GPT-5 High 15 tok/s Pro
GPT-4o 101 tok/s Pro
Kimi K2 196 tok/s Pro
GPT OSS 120B 467 tok/s Pro
Claude Sonnet 4 37 tok/s Pro
2000 character limit reached

Derrida's random energy models. From spin glasses to the extremes of correlated random fields (1412.0958v1)

Published 2 Dec 2014 in math.PR

Abstract: These are notes for a mini-course on the extremes of correlated random fields which I gave in the spring 2013 in Marseille. The first chapter recalls the paradigmatic random energy models with finitely many scales introduced by B. Derrida in the context of mean field spin glasses. The second chapter presents a multi-scale refinement of the second moment method which is particularly efficient to analyze models with a growing number of scales, such as the Gaussian hierarchical field (the directed polymer on Cayley trees). In the third chapter, applications of the method to e.g. Gaussian free fields and issues of percolation in high dimensions are briefly touched upon. The last chapter deals with a procedure of "local projections" which allows, in a number of cases, to construct scales from first principles.

Citations (55)
List To Do Tasks Checklist Streamline Icon: https://streamlinehq.com

Collections

Sign up for free to add this paper to one or more collections.

Summary

Overview of Derrida's Random Energy Models

Nicola Kistler's notes on Derrida's random energy models (REM and GREM) provide an in-depth discussion on the relevance of these models beyond their initial context in mean-field spin glasses. Initially introduced by Bernard Derrida in the 1980s, these models have offered significant insights into certain phenomena associated with extreme values of correlated random fields. The purpose of Kistler's exposition is to illustrate the broader applicability of the REM-class across different fields using a simplified yet comprehensive approach, minimizing the intricacies typically involved in such treatments.

Key Concepts and Methodology

  1. Random Energy Model (REM) and Generalized REM (GREM): The REM and GREM are paradigms for understanding extreme behaviors in mean-field spin glasses. While the REM comprises independent Gaussian random variables, the GREM introduces a hierarchical structure, adding correlations between variables. This distinction forms the foundation for analyzing systems with more complex relationships between components.
  2. Hierarchical Field Models: Kistler discusses models such as the hierarchical Gaussian field, critical for studying extremes in certain random structures. Through a multiscale refinement of the second moment method, Kistler attempts to approximate these fields via simpler structured forms like GREM, illustrating their usage as approximative tools.
  3. Second Moment Method and Multiscale Refinement: The second moment method stands out as a classical technique used to evaluate the leading order behavior of such random fields. Kistler elaborates on its nuances and extensions which incorporate multiscale refinements – breaking down large combinatorial structures into manageable hierarchical levels.
  4. Limitations and Complexity in Models: While REM and GREM offer substantial insights, their treatment becomes intricate in structures that defy simple hierarchical partitioning, such as random fields with non-Gaussian correlations. Kistler's approach emphasizes identifying independent contributions at different scales as key to navigating these challenges.

Numerical and Theoretical Implications

Kistler's analysis points to significant implications for both theoretical exploration and practical application. The REM-class's ability to model non-trivial systems' freezing transitions despite its simplifications provides a bridge to investigating complex disordered states in statistical mechanics and beyond. This paper suggests a pathway to addressing models traditionally outside the REM-class through hierarchical approximations, thus broadening the scope of problems amenable to such treatment.

Future Directions

This research opens avenues for further exploration into the applications of these randomly hierarchical models in diverse areas such as number theory and random matrix theory. By formalizing the scales and employing multiscale approximation methods effectively, future studies could refine predictions and analyses of non-Gaussian correlated systems. Additionally, exploring the universality aspects within more models could offer profound insights into the mathematical structure governing extreme value statistics.

In conclusion, Kistler contributes significantly to the understanding of REM-class models, highlighting their potential to extend beyond traditional boundaries and influence multiple fields. His work prompts a reconsideration of hierarchical thinking and multiscale approaches, advocating for their central role in contemporary probability and statistical physics.

Dice Question Streamline Icon: https://streamlinehq.com

Follow-Up Questions

We haven't generated follow-up questions for this paper yet.

Authors (1)

X Twitter Logo Streamline Icon: https://streamlinehq.com