Distribution approximation of covariance matrix eigenvalues

Abstract

In multivariate analysis, the eigenvalues of the covariance matrix are crucial. Thus, there is a demand among users to find a good, easy-to-use chi-squared approximation. However, there are few good approximations for eigenvalues. Therefore, in this paper, we focus on the chi-squared approximation, proposing a new approximation and investigating its accuracy.

Keywords:

• Eigenvalues
• Chi-squared approximation
• Wishart matrix
• Principal component analysis