REVIEW 1 cited by
Large deviation theory for diluted Wishart random matrices
Not yet reviewed by Pith; the record is open.
This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.
SPECIMEN: schema-true, not a live event
T0 review · schema-true
One-sentence machine reading of the paper's core claim.
pith:XXXXXXXX · record.json · timestamp
Large deviation theory for diluted Wishart random matrices
read the original abstract
Wishart random matrices with a sparse or diluted structure are ubiquitous in the processing of large datasets, with applications in physics, biology and economy. In this work we develop a theory for the eigenvalue fluctuations of diluted Wishart random matrices, based on the replica approach of disordered systems. We derive an analytical expression for the cumulant generating function of the number of eigenvalues $\mathcal{I}_N(x)$ smaller than $x\in\mathbb{R}^{+}$, from which all cumulants of $\mathcal{I}_N(x)$ and the rate function $\Psi_{x}(k)$ controlling its large deviation probability $\text{Prob}[\mathcal{I}_N(x)=kN] \asymp e^{-N\Psi_{x}(k)}$ follow. Explicit results for the mean value and the variance of $\mathcal{I}_N(x)$, its rate function, and its third cumulant are discussed and thoroughly compared to numerical diagonalization, showing a very good agreement. The present work establishes the theoretical framework put forward in a recent letter [Phys. Rev. Lett. 117, 104101] as an exact and compelling approach to deal with eigenvalue fluctuations of sparse random matrices.
Forward citations
Cited by 1 Pith paper
-
Spectral Signatures of Replica Symmetry Breaking in Optimization-Induced Random Matrices
Difference spectra of two optimization-induced matrices from independent Gibbs samples form an explicit spectral transform of the Parisi overlap order parameter, while a single-matrix bulk is blind to replica symmetry...
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.