Linear low energy excitations in fully-connected models of glasses

被引：6

作者：

Franz, Silvio ^{[1
]}

Nicoletti, Flavio ^{[2
]}

Ricci-Tersenghi, Federico ^{[2
,3
]}

机构：

[1] Univ Paris Saclay, Univ Paris Sud, LPTMS, CNRS,UMR 8626, F-105 Orsay, France

[2] Univ Roma La Sapienza, Dipartimento Fis, Ple A Moro 5, I-00185 Rome, Italy

[3] Ist Nazl Fis Nucl, Sez Roma1, CNR Nanotec, Rome Unit, Ple A Moro 5, I-00185 Rome, Italy

来源：

JOURNAL OF STATISTICAL MECHANICS-THEORY AND EXPERIMENT | 2022年 / 2022卷 / 05期

关键词：

random matrix theory and extensions; spin glasses; energy landscapes; mode coupling theory; INTERNAL FIELD DISTRIBUTIONS; SPIN-GLASSES; MATRIX; COMPLEXITY;

D O I：

10.1088/1742-5468/ac6518

中图分类号：

O3 [力学];

学科分类号：

08 ; 0801 ;

摘要：

We study the linear excitations around typical energy minima of a mean-field disordered model with continuous degrees of freedom undergoing a random first order transition. Contrary to naive expectations, the spectra of linear excitations are ungapped and we find the presence of a pseudogap corresponding to localized excitations with arbitrary low excitation energy. Moving to deeper minima in the landscape, the excitations appear increasingly localized while their abundance decreases. Beside typical minima, there also exist rare ultra-stable minima, with an energy gap and no localised excitations.

引用

页数：31

共 50 条

[1] Energy Complexity of Fully-Connected Layers
Sima, Jiri
Cabessa, Jeremie
ADVANCES IN COMPUTATIONAL INTELLIGENCE, IWANN 2023, PT I, 2023, 14134 : 3 - 15
[2] On energy complexity of fully-connected layers
Sima, Jiri
Cabessa, Jeremie
Vidnerova, Petra
NEURAL NETWORKS, 2024, 178
[3] An Energy Efficient MAC Protocol for Fully-Connected Wireless Networks
Malekshan, Kamal Rahimi
Zhuang, Weihua
2013 IEEE INTERNATIONAL CONFERENCE ON COMMUNICATIONS (ICC), 2013, : 1538 - 1542
[4] A Scalable Annealing Processing Architecture for Fully-Connected Ising Models
Jiang, Dong
Wang, Xiangrui
Huang, Zhanhong
Huang, Yukang
Yao, Enyi
2023 IEEE INTERNATIONAL SYMPOSIUM ON CIRCUITS AND SYSTEMS, ISCAS, 2023,
[5] Fully-connected networks with local connections
P. E. Kornilovitch
R. N. Bicknell
J. S. Yeo
Applied Physics A, 2009, 95 : 999 - 1004
[6] Fully-connected bond percolation on Zd
Dereudre, David
PROBABILITY THEORY AND RELATED FIELDS, 2022, 183 (1-2) : 547 - 579
[7] On the Learnability of Fully-connected Neural Networks
Zhang, Yuchen
Lee, Jason D.
Wainwright, Martin J.
Jordan, Michael I.
ARTIFICIAL INTELLIGENCE AND STATISTICS, VOL 54, 2017, 54 : 83 - 91
[8] The Unreasonable Effectiveness of Fully-Connected Layers for Low-Data Regimes
Kocsis, Peter
Sukenik, Peter
Braso, Guillem
Niessner, Matthias
Leal-Taixe, Laura
Elezi, Ismail
ADVANCES IN NEURAL INFORMATION PROCESSING SYSTEMS 35, NEURIPS 2022, 2022,
[9] Energy efficient, fully-connected mesh networks for high speed applications
Ekanayake, Samitha W.
Pathirana, Pubudu N.
Rolfe, Bernard F.
Palaniswami, Marimuthu
2008 IEEE 67TH VEHICULAR TECHNOLOGY CONFERENCE-SPRING, VOLS 1-7, 2008, : 2606 - +
[10] Fully-connected networks with local connections
Kornilovitch, P. E.
Bicknell, R. N.
Yeo, J. S.
APPLIED PHYSICS A-MATERIALS SCIENCE & PROCESSING, 2009, 95 (04): : 999 - 1004

← 1 2 3 4 5 →