2
$\begingroup$

I need the following answer for research purposes.

Let $A$ be a $m \times n$ random matrix with iid ${\rm Bernoulli}(p)$ entries. Is there any result on the spectral gap of $AA^{T}$ (similar to well known random matrix laws like Tracy-Widom)?

$\endgroup$

1 Answer 1

1
$\begingroup$

The spectrum of random matrices has some universality properties, in the sense that they do not depend on details of the distribution of the matrix elements. For instance, the spectral gap, i.e. the smalles eigenvalue of your $AA^T$, should be the same as for real Gaussian $A$ matrices (when $A$ is real Gaussian, the ensemble of $AA^T$ is called the real Wishart ensemble). At least for $p=1/2$, this is actually proved in the paper Random Matrices: The Distribution of the Smallest Singular Values, by Terence Tao and Van Vu (Geom. Funct. Anal. 20, 260–297, 2010). For large dimensions, the eigenvalue density is called the Marchenko–Pastur distribution.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.