Constraint for a full rank matrix

Nonconvex
1

Hi All!

I’m trying to solve some problem with the help of cvx. I should say that I’m also new at cvx and optimiztion problems.
Anyways, at the end, I’m looking for a matrix which I need it to be full rank. But as far as I’m concerned, this is not a convex constraint.

So what should the constraint be in order to get as a solution a full rank matrix?

Thanks a lot,

Alan

2

As you rightly point out, your constraint is not convex. CVX probably is not your solution. The closest thing CVX could do would be to bound a positive definite matrix away from zero; e.g., X-I==semidefinite(n).