Fractal geometry of spin-glass models

Abstract

Stability and diversity are two key properties that living entities share with spin glasses, where they are manifested through the breaking of the phase space into many valleys or local minima connected by saddle points. The topology of the phase space can be conveniently condensed into a tree structure, akin to the biological phylogenetic trees, whose tips are the local minima and internal nodes are the lowest-energy saddles connecting those minima. For the infinite-range Ising spin glass with p-spin interactions, we show that the average size-frequency distribution of saddles obeys a power law $<\psi (w)>\sim w^{-D}$, where w=w(s) is the number of minima that can be connected through saddle s, and D is the fractal dimension of the phase space.

Cited by in corpus (6)

• Random Walk over Basins of Attraction to Construct Ising Energy Landscapes
• Shapes of tree representations of spin-glass landscapes
• Phase transition and landscape statistics of the number partitioning problem
• The Statistics of the Number of Minima in a Random Energy Landscape
• Universality of REM-like aging in mean field spin glasses
• Quality-diversity in dissimilarity spaces