I have two problems related to eigenvalues of negative definite matrices:

  1. I have a matrix $M\prec0$ (symmetric and all eigenvalues are negative) and $S=M_{11}-M_{12}M_{22}^{-1}M_{21}$ by taking $M=[M_{ij}]$. Now I want to derive the relation between the eigenvalues of $M$ and $S$. I am particularly looking for a relation between the minimum and maximum eigenvalues.

  2. Now if $M=A^T+A\prec 0$ then what can be said about the relationship between the eigenvalues of $M$ and $A$. I have tried that if the condition of symmetry is not imposed then $A, A^T \prec 0$ with the real part of all eigenvalues negative. But if this is so, can we have any eigenvalue inequality on real parts only.


As for your first question, $S^{-1}$ is a diagonal block of $M^{-1}$. You can only say that its eigenvalues are interlaced with those of $M^{-1}$. Since all the eigenvalues of $M$ and $S$ are negative, it amounts to saying that the eigenvalues of $S$ are interlaced with those of $M$~: $$\lambda_1\le\mu_1\le\lambda_2\le\cdots\le\mu_{n-1}\le\lambda_n.$$

As for your second question, see my previous answer https://mathoverflow.net/q/52588.

  • $\begingroup$ Thanks Prof. Denis. I got the first one. For the second one, how can I prove the relationship given by you in the other reply and what can I comment on eigenvalues of $M$ (it was $N$ in your earlier reply). $\endgroup$ – Parikshit Pareek Sep 6 '18 at 6:26

Your Answer

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

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