Abstract:Approximate message passing (AMP) has emerged both as a popular class of iterative algorithms and as a powerful analytic tool in a wide range of statistical estimation problems and statistical physics models. A well established line of AMP theory proves Gaussian approximations for the empirical distributions of the AMP iterate in the high dimensional limit, under the GOE random matrix model and its variants. This paper provides a non-asymptotic, leave-one-out representation for the AMP iterate that holds under a broad class of Gaussian random matrix models with general variance profiles. In contrast to the typical AMP theory that describes the empirical distributions of the AMP iterate via a low dimensional state evolution, our leave-one-out representation yields an intrinsically high dimensional state evolution formula which provides non-asymptotic characterizations for the possibly heterogeneous, entrywise behavior of the AMP iterate under the prescribed random matrix models. To exemplify some distinct features of our AMP theory in applications, we analyze, in the context of regularized linear estimation, the precise stochastic behavior of the Ridge estimator for independent and non-identically distributed observations whose covariates exhibit general variance profiles. We find that its finite-sample distribution is characterized via a weighted Ridge estimator in a heterogeneous Gaussian sequence model. Notably, in contrast to the i.i.d. sampling scenario, the effective noise and regularization are now full dimensional vectors determined via a high dimensional system of equations. Our leave-one-out method of proof differs significantly from the widely adopted conditioning approach for rotational invariant ensembles, and relies instead on an inductive method that utilizes almost solely integration-by-parts and concentration techniques.

Equivalence of Approximate Message Passing and Low-Degree Polynomials in Rank-One Matrix Estimation

Equivalence of approximate message passing and low-degree polynomials in rank-one matrix estimation

A Unitary Transform Based Generalized Approximate Message Passing

Optimality of Approximate Message Passing Algorithms for Spiked Matrix Models with Rotationally Invariant Noise

Approximate Message Passing for orthogonally invariant ensembles: Multivariate non-linearities and spectral initialization

A Non-asymptotic Analysis of Generalized Approximate Message Passing Algorithms with Right Rotationally Invariant Designs

A Concise Tutorial on Approximate Message Passing

Rank-one matrix estimation: analysis of algorithmic and information theoretic limits by the spatial coupling method

Approximate Message Passing for the Matrix Tensor Product Model

Linear Operator Approximate Message Passing (OpAMP)

A Non-Asymptotic Framework for Approximate Message Passing in Spiked Models

Matrix denoising: Bayes-optimal estimators via low-degree polynomials

Universality of Approximate Message Passing algorithms and tensor networks

A leave-one-out approach to approximate message passing

A non-asymptotic distributional theory of approximate message passing for sparse and robust regression

Fundamental limits of Non-Linear Low-Rank Matrix Estimation

Analysis of approximate message passing with a class of non-separable denoisers

Optimal low-rank approximations of posteriors for linear Gaussian inverse problems on Hilbert spaces

Approximate Survey Propagation for Statistical Inference

Estimation of high-dimensional low-rank matrices