scholarly journals Mean-field theory for a spin-glass model of neural networks: TAP free energy and the paramagnetic to spin-glass transition

1997 ◽  
Vol 30 (23) ◽  
pp. 8085-8094 ◽  
Author(s):  
K Nakanishi ◽  
H Takayama
1996 ◽  
Vol 455 ◽  
Author(s):  
S. C. Glotzer ◽  
P. H. Poole ◽  
A. Coniglio ◽  
N. Jan

ABSTRACTThe temperature dependence of the microstructure and local dynamics in the paramagnetic phase of the d = 2 and d = 3 ± J Ising spin glass model is examined by comparing the equilibrium distributions of local flip-rates and local energies calculated in large-scale Monte Carlo simulations. The emergence in this model of fast processes as the glass transition is approached corresponds with recent experimental results.


1997 ◽  
Vol 30 (20) ◽  
pp. 7021-7038 ◽  
Author(s):  
Andrea Cavagna ◽  
Irene Giardina ◽  
Giorgio Parisi

2000 ◽  
Vol 13 (2) ◽  
pp. 147-160
Author(s):  
T. C. Dorlas ◽  
J. R. Wedagedera

A separable spin glass model whose exchange integral takes the form Jij=J(ξi1ξj2+ξi2ξj1) which was solved by van Hemmen et al. [12] using large deviation theory [14] is rigorously treated. The almost sure convergence criteria associated with the cumulant generating function C(t) with respect to the quenched random variables ξ is carefully investigated, and it is proved that the related excluded null set 𝒩 is independent of t. The free energy and hence the other thermodynamic quantities are rederived using Varadhan's Large Deviation Theorem. A simulation is also presented for the entropy when ξ assumes a Gaussian distribution.


Sign in / Sign up

Export Citation Format

Share Document