English

High-dimensional Asymptotics of Langevin Dynamics in Spiked Matrix Models

Statistics Theory 2024-03-01 v1 Machine Learning Probability Machine Learning Statistics Theory

Abstract

We study Langevin dynamics for recovering the planted signal in the spiked matrix model. We provide a "path-wise" characterization of the overlap between the output of the Langevin algorithm and the planted signal. This overlap is characterized in terms of a self-consistent system of integro-differential equations, usually referred to as the Crisanti-Horner-Sommers-Cugliandolo-Kurchan (CHSCK) equations in the spin glass literature. As a second contribution, we derive an explicit formula for the limiting overlap in terms of the signal-to-noise ratio and the injected noise in the diffusion. This uncovers a sharp phase transition -- in one regime, the limiting overlap is strictly positive, while in the other, the injected noise overcomes the signal, and the limiting overlap is zero.

Keywords

Cite

@article{arxiv.2204.04476,
  title  = {High-dimensional Asymptotics of Langevin Dynamics in Spiked Matrix Models},
  author = {Tengyuan Liang and Subhabrata Sen and Pragya Sur},
  journal= {arXiv preprint arXiv:2204.04476},
  year   = {2024}
}

Comments

26 pages