A Matrix Gaussian Distribution

arXiv1804.11010MaRDI QIDQ6300962FDOQ6300962

Publication date: 29 April 2018

Abstract: In this note, we define a Gaussian probability distribution over matrices. We prove some useful properties of this distribution, namely, the fact that marginalization, conditioning, and affine transformations preserve the matrix Gaussian distribution. We also derive useful results regarding the expected value of certain quadratic forms based solely on covariances between matrices. Previous definitions of matrix normal distributions are severely under-parameterized, assuming unrealistic structure on the covariance (see Section 2). We believe that our generalization is better equipped for use in practice.

This page was built for publication: A Matrix Gaussian Distribution

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6300962)