Consider the following block matrix

$$A = \left(\begin{matrix} B & T\\ T & 0 \end{matrix} \right)$$

where all submatrices are square and

  • matrix $B = \mbox{diag}\left(b_1 ,0,0,\dots,0,b_n \right)$ with $b_1, b_n > 0$.

  • matrix $T$ is self-adjoint and positive semidefinite.

What can one say about the lowest eigenvalue of this matrix $A$? In particular, how does it depend on the spectrum of $T$ and the entries of $B$? Are there any known results?

  • $\begingroup$ $A$ may have negative eigenvalues, you want the smallest in absolute value or the most negative one? $\endgroup$ – Carlo Beenakker Feb 19 '19 at 22:57
  • 1
    $\begingroup$ the lowest one (which may be negative)... $\endgroup$ – Sascha Feb 19 '19 at 23:47

If $\lambda_\max$ is the greatest eigenvalue of $T$, the least eigenvalue of $A$ is between $-\lambda_\max$ and $\max(b_1, b_n) - \lambda_\max$.

| cite | improve this answer | |
  • 1
    $\begingroup$ thank you, however do you mind briefly explaining how you get this? $\endgroup$ – Sascha Feb 20 '19 at 0:28
  • 2
    $\begingroup$ The eigenvalues of $\pmatrix{0 & T\cr T & 0\cr}$ are the eigenvalues of $T$ and $-T$. You're adding a positive semidefinite matrix. Use the min-max theorem. $\endgroup$ – Robert Israel Feb 20 '19 at 0:34

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.