These lecture notes focus on the mean field theory of spin glasses, with particular emphasis on the presence of a very large number of metastable states in these systems. This phenomenon, and some of its physical consequences, will be discussed in details for fully-connected models and for models defined on random lattices. This will be done using the replica and cavity methods.These notes have been prepared for a course of the PhD program in Statistical Mechanics at SISSA, Trieste and at the University of Rome "Sapienza". Part of the material is reprinted from other lecture notes, and when this is done a reference is obviously provided to the original. I would like to warmly thank all the students and colleagues who read these notes, gave me their feedback and sent me their corrections, that allowed to fix many errors on the original manuscript. I also would like to thank SISSA and the University of Rome "Sapienza" for inviting me to give these lectures.
Contents
I. IntroductionA. Why study spin glasses? B. Physical systems C. Optimization problems D. Models and universality classes E. Frustration and quenched disorder F. What is missing in these notes II. Fully connected models A. Free energy functional 1. The fully connected Ising ferromagnet 2. Metastable states in fully connected models 3. The general definition of the free energy functional 4. The Georges-Yedidia expansion 5. Free energy functional for a generic Ising model 6. Back to the fully connected ferromagnet 7. TAP equations for the SK model 8. Spherical p-spin model 9. Summary and remarks B. Metastable states and complexity 1. The simplest example: the spherical p-spin glass model 2. The partition function 3. A method to compute the complexity 4. Replicated free energy of the spherical p-spin model 5. 1-step replica symmetry breaking 6. The phase diagram of the spherical p-spin model 7. Spontaneous replica symmetry breaking: the order parameter C. The SK model: full replica symmetry breaking 1. The overlap distribution 2. The Parisi solution of the SK model D. Susceptibilities 1. The ferromagnet 2. Spin glasses: linear susceptibilities 3. The static spin glass susceptibility 4. The dynamic spin glass susceptibility E. Exercises III. Diluted models and optimization problems A. Definitions 1. Statistical mechanics formulation of optimization problems 2. Random optimization problems: random graphs and hypergraphs 3. Connectivity-temperature phase diagram B. XORSAT: clustering and SAT/UNSAT transition 1. Bounds from the first and second moments methods 2. The leaf-removal algorithm 3. Clustering and SAT/UNSAT transitions. 4. On backbones 5. Summary C. The replica symmetric cavity method 1. Recursions on a finite tree 2. From a tree to a random graph 3. Replica symmetric cavity equations in absence of local disorder 4. An alternative derivation 5. Fluctuations of the local environment: distributions of cavity probabilities 6. The zero temperature limit 7. On a factor graph 8. Summary D. 1-step replica symmetry breaking 1. The auxiliary model 2. RS cavit...