Hamilton-Jacobi equations for finite-rank matrix inference
read the original abstract
We compute the large-scale limit of the free energy associated with the problem of inference of a finite-rank matrix. The method follows the principle put forward in arXiv:1811.01432 which consists in identifying a suitable Hamilton-Jacobi equation satisfied by the limit free energy. We simplify the approach of arXiv:1811.01432 using a notion of weak solution of the Hamilton-Jacobi equation which is more convenient to work with and is applicable whenever the non-linearity in the equation is convex.
This paper has not been read by Pith yet.
Forward citations
Cited by 2 Pith papers
-
Parisi's formula is a Hamilton-Jacobi equation in Wasserstein space
Parisi's formula is recast as the solution of a Hamilton-Jacobi equation in the Wasserstein space of probability measures on the positive half-line.
-
Concentration of the matrix-valued minimum mean-square error in optimal Bayesian inference
The matrix-valued MMSE concentrates in the large-system limit for optimal Bayesian inference of vector-valued signals.
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.