Chaos, concentration, and multiple valleys

Abstract

Disordered systems are an important class of models in statistical mechanics, having the defining characteristic that the energy landscape is a fixed realization of a random field. Examples include various models of glasses and polymers. They also arise in other areas, like fitness models in evolutionary biology. The ground state of a disordered system is the state with minimum energy. The system is said to be chaotic if a small perturbation of the energy landscape causes a drastic shift of the ground state. We present a rigorous theory of chaos in disordered systems that confirms long-standing physics intuition about connections between chaos, anomalous fluctuations of the ground state energy, and the existence of multiple valleys in the energy landscape. Combining these results with mathematical tools like hypercontractivity, we establish the existence of the above phenomena in eigenvectors of GUE matrices, the Kauffman-Levin model of evolutionary biology, directed polymers in random environment, a subclass of the generalized Sherrington-Kirkpatrick model of spin glasses, the discrete Gaussian free field, and continuous Gaussian fields on Euclidean spaces. We also list several open questions.

63 pages. Proof of Theorem 3.2 simplified. Mistake corrected in proof of Theorem 8.1. Several minor changes

References in corpus (8)

• Level-Spacing Distributions and the Airy Kernel
• Broken Replica Symmetry Bounds in the Mean Field Spin Glass Model
• Fredholm Determinants, Differential Equations and Matrix Models
• Anomalous Fluctuations of Directed Polymers in Random Media
• Second class particles and cube root asymptotics for Hammersley's process
• On the overlap in the multiple spherical SK models
• Temperature and Disorder Chaos in Low Dimensional Directed Paths
• Density estimates and concentration inequalities with Malliavin calculus

Cited by in corpus (19)

• Disorder chaos and multiple valleys in spin glasses
• Asymptotics of cover times via Gaussian free fields: Bounded-degree graphs and general trees
• Noise-stability and central limit theorems for effective resistance of random electric networks
• Field-measure correspondence in Liouville quantum gravity almost surely commutes with all conformal maps simultaneously
• High-dimensional Central Limit Theorems by Stein's Method
• Geodesics of Random Riemannian Metrics
• Oded Schramm's contributions to noise sensitivity
• Geometric Influences II: Correlation Inequalities and Noise Sensitivity
• Recursions and tightness for the maximum of the discrete, two dimensional Gaussian Free Field
• Fluctuations of the free energy in the mixed $p$-spin models with external field
• Phase transition for accessibility percolation on hypercubes
• An approach to chaos in some mixed p-spin models
• Disorder chaos in some diluted spin glass models
• Geodesics of Random Riemannian Metrics: Supplementary Material
• Dissertation: Geodesics of Random Riemannian Metrics
• Chaos in the mixed even-spin models
• Universality of chaos and ultrametricity in mixed p-spin models
• Sublinear variance for directed last-passage percolation
• Multi-dimensional Gaussian fluctuations on the Poisson space