The spin-glass phase-transition in the Hopfield model with p-spin interactions

DOI10.4310/ATMP.2001.V5.N6.A2MaRDI QIDQ1860784zbMATH OpenOpenAlexFDO

Authors Anton Bovier, Beat M. Niederhauser

Publication date 9 June 2003

Published in Advances in Theoretical and Mathematical Physics (Search for Journal in Brave)

Full work available at URL https://arxiv.org/abs/cond-mat/0108235

zbMATH Keywords

phase transitionspin glassesoverlap distributionHopfield models

Mathematics Subject Classification ID

Interacting random processes; statistical mechanics type models; percolation theory (60K35) Critical phenomena in equilibrium statistical mechanics (82B27) Statistical mechanics of random media, disordered materials (including liquid crystals and spin glasses) (82D30)

Abstract: We study the Hopfield model with pure $p$-spin interactions with even $pgeq4$, and a number of patterns, M(N) growing with the system size, $N$, as $M(N)=a{N}^{p-1}$. We prove the existence of a critical temperature $$ characterized as the first time quenched and annealed free energy differ. We prove that as $puparrowinfty$, $$. Moreover, we show that for any $a>0$ and for all inverse temperatures $$, the free energy converges to that of the REM at inverse temperature $$. Moreover, above the critical temperature the distribution of the replica overlap is concentrated at zero. We show that for large enough $a$, there exists a non-empty interval of in the low temperature regime where the distribution has mass both near zero and near $pm1$. As was first shown by M. Talagrand in the case of the $p$-spin SK model, this implies the the Gibbs measure at low temperatures is concentrated, asymptotically for large $N$, on a countable union of disjoint sets, no finite subset of which has full mass. Finally, we show that there is $a_{p}sim1/p!$ such that for $a>a_p$ the set carrying almost all mass does not contain the original patterns. In this sense we describe a genuine spin glass transition. Our approach follows that of Talagrand's analysis of the $p$-spin SK-model. The more complex structure of the random interactions necessitates, however, considerable technical modifications. In particular, various results that follow easily in the Gaussian case from integration by parts fromulas have to be derived by expansion techniques.

Recommendations

• scientific article; zbMATH DE number 2177568
• On the Hopfield model at the critical temperature
• Rigorous low-temperature results for the mean field p-spins interaction model
• On the p-spin interaction model at low temperature
• Rigorous results for the Hopfield model with many patterns

Cited in

(25)

• Interpolating between boolean and extremely high noisy patterns through minimal dense associative memories
• About the ergodic regime in the analogical Hopfield neural networks: Moments of the partition function
• A \(p\)-spin interaction Ashkin-Teller spin-glass model
• scientific article; zbMATH DE number 5243961 (Why is no real title available?)
• Nonlinear PDEs approach to statistical mechanics of dense associative memories
• Exact solution of the gauge symmetric \(p\)-spin glass model on a complete graph
• THE INVESTIGATION OF THE SPIN GLASS PROPERTIES OF THE HOPFIELD NEURAL NETWORK MODEL
• A spectral approach to Hebbian-like neural networks
• Notes on ferromagnetic diluted p-spin model
• Hebbian learning from first principles
• scientific article; zbMATH DE number 2177568 (Why is no real title available?)
• Dense Hebbian neural networks: a replica symmetric picture of unsupervised learning
• Dense Hebbian neural networks: a replica symmetric picture of supervised learning
• scientific article; zbMATH DE number 1394783 (Why is no real title available?)
• Rigorous low-temperature results for the mean field p-spins interaction model
• On a model of associative memory with huge storage capacity
• Spontaneous replica symmetry breaking and interpolation methods for complex statistical mechanics systems
• The overlap gap property and approximate message passing algorithms for \(p\)-spin models
• Notes on the \(p\)-spin glass studied via Hamilton-Jacobi and smooth-cavity techniques
• Replica symmetry breaking in dense Hebbian neural networks
• Pattern reconstruction with restricted Boltzmann machines
• An interacting diffusion model and SK spin glass equation
• Binary associative memory networks: a review of mathematical framework and capacity analysis
• Concentration of the complexity of spherical pure \(p\)-spin models at arbitrary energies
• Mean-field theory for a spin-glass model of neural networks: TAP free energy and the paramagnetic to spin-glass transition

This page was built for publication: The spin-glass phase-transition in the Hopfield model with \(p\)-spin interactions

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q1860784)