• Corpus ID: 4565761

Asymptotic normality and analysis of variance of log-likelihood ratios in spiked random matrix models

  title={Asymptotic normality and analysis of variance of log-likelihood ratios in spiked random matrix models},
  author={Debapratim Banerjee and Zongming Ma},
The present manuscript studies signal detection by likelihood ratio tests in a number of spiked random matrix models, including but not limited to Gaussian mixtures and spiked Wishart covariance matrices. We work directly with multi-spiked cases in these models and with flexible priors on the signal component that allow dependence across spikes. We derive asymptotic normality for the log-likelihood ratios when the signal-to- noise ratios are below certain thresholds. In addition, we show that… 

Fundamental limits of detection in the spiked Wigner model

The fundamental limits of detecting the presence of an additive rank-one perturbation, or spike, to a Wigner matrix are studied and it is proved that the log-likelihood ratio of the spiked model against the non-spiked one is asymptotically normal below a certain reconstruction threshold.

Detection limits and fluctuation results in some spiked random matrix models and pooling of discrete data

The fundamental limits of detecting and recovering a weak signal hidden in a large amount of noise are examined, and sharp upper and lower bounds on the minimal number of measurements are determined for the signal to be essentially unique, thus in principle recoverable from the observed data.


The fundamental limits of detecting the presence of an additive rank-one perturbation to a Wigner matrix are studied and it is proved that the log-likelihood ratio of the spiked model against the nonspiked one is asymptotically normal below a certain reconstruction threshold, and that it is degenerate above.

Contextual Stochastic Block Model: Sharp Thresholds and Contiguity

It is expected that the conjecture holds as soon as the average degree exceeds one, so that the graph has a giant component, and the sharp threshold for detection and weak recovery is characterized.

Fluctuations of the free energy of the mixed $p$-spin mean field spin glass model

The Sherrington-Kirkpatrick [SK75] model and its variants [Der80; GM84; Tal00] are important models of disordered magnetic systems and paradigmatic examples in the theory of complex systems [MPV87;

Clustering Network Vertices in Sparse Contextual Multilayer Networks

This work establishes that the detection threshold coincides with the threshold for weak recovery of the common community structure using multiple correlated networks and co-variate matrices, and provides a quasi-polynomial time algorithm to estimate the latent communities in the recovery regime.



Detection limits in the high-dimensional spiked rectangular model

A probabilistic approach capable of treating generic product priors, and in particular, sparsity in the spike is allowed, based on Talagrand's interpretation of the cavity method from spin-glass theory.

Testing in high-dimensional spiked models

We consider the five classes of multivariate statistical problems identified by James (1964), which together cover much of classical multivariate analysis, plus a simpler limiting case, symmetric

Optimality and Sub-optimality of PCA for Spiked Random Matrices and Synchronization

The fundamental limitations of statistical methods are studied, including non-spectral ones, and it is shown that inefficient procedures can work below the threshold where PCA succeeds, whereas no known efficient algorithm achieves this.

Optimal hypothesis testing for stochastic block models with growing degrees

A class of adaptive tests that are computationally tractable and completely data-driven that achieve nontrivial powers in the contiguous regime and consistency in the singular regime whenever $n p_{n,av} \to\infty$ is the average connection probability.

Signal detection in high dimension: The multispiked case

This paper considers the problem of testing the null hypothesis of sphericity of a high-dimensional covariance matrix against an alternative of (unspecified) multiple symmetry-breaking directions (multispiked alternatives) and derives simple analytical expressions for theasymptotic power envelope and the asymptotic powers of previously proposed tests are derived.

Information-Theoretic Bounds and Phase Transitions in Clustering, Sparse PCA, and Submatrix Localization

The upper bounds show that for each of these problems there is a significant regime where reliable detection is information-theoretically possible but where known algorithms such as PCA fail completely, since the spectrum of the observed matrix is uninformative.

Asymptotic power of sphericity tests for high-dimensional data

This paper studies the asymptotic power of tests of sphericity against perturbations in a single unknown direction as both the dimensionality of the data and the number of observations go to

A CLT for a band matrix model

A law of large numbers and a central limit theorem are derived for linear statistics of random symmetric matrices whose on-or-above diagonal entries are independent, but neither necessarily

Random Regular Graphs: Asymptotic Distributions and Contiguity

  • S. Janson
  • Mathematics
    Combinatorics, Probability and Computing
  • 1995
The asymptotic distribution of the number of Hamilton cycles in a random regular graph is determined, and results imply that some different models of random regular graphs are contiguous, which means that they are qualitatively asymPTotically equivalent.

Introduction to Random Matrices

Here I = S j (a2j 1,a2j) andI(y) is the characteristic function of the set I. In the Gaussian Unitary Ensemble (GUE) the probability that no eigenvalues lie in I is equal to �(a). Also �(a) is a