Assume I have $n$ random variables with covariance matrix $\Sigma$. Now, I want to find $m$ groups of variables such that they very correlated inside each group, but their correlation between groups is small.

This is kind of saying that I want to shuffle the random variables such that the covariance matrix looks as close as possible to a block diagonal matrix.

Do I you know if there's an algorithm for this? I guess I would want something similar to the Jordan normal form of the matrix, but that implies taking linear combinations of the random variables, if I am not mistaken. Is there another way to do it?


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.