Zero-temperature dynamics in the dilute Curie-Weiss model
Reza Gheissari, Charles M. Newman, and Daniel L. Stein

TL;DR
This paper studies zero-temperature dynamics in a disordered Curie-Weiss Ising model on dense random graphs, showing it typically reaches a uniform ground state and relating this to the local MINCUT problem, with differences observed in heavy-tailed coupling cases.
Contribution
It proves that zero-temperature dynamics in the disordered Curie-Weiss model typically avoids local minima and reaches a uniform ground state, linking this to the local MINCUT problem.
Findings
Dynamics avoids local minima in typical cases.
Greedy local MINCUT search fails to find non-trivial cuts.
Heavy-tailed couplings can lead to different local minima.
Abstract
We consider the Ising model on a dense Erd\H{o}s--R\'enyi random graph, , with fixed---equivalently, a disordered Curie--Weiss Ising model with couplings---at zero temperature. The disorder may induce local energy minima in addition to the two uniform ground states. In this paper we prove that, starting from a typical initial configuration, the zero-temperature dynamics avoids all such local minima and absorbs into a predetermined one of the two uniform ground states. We relate this to the local MINCUT problem on dense random graphs; namely with high probability, the greedy search for a local MINCUT of with fixed, started from a uniform random partition, fails to find a non-trivial cut. In contrast, in the disordered Curie--Weiss model with heavy-tailed couplings, we demonstrate that zero-temperature dynamics has positive…
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
