Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

All matrices I consider are complex hermitian $N\times N$ matrices. None of them are Diagonal matrices. Consider a positive semi definite matrix $A$ and a matrix $B$. I am interested in the sum $C=A-\gamma B$ which $\gamma$ is a non-negative scalar.


What is the smallest gamma such that $C$ is negative semi-definite when $B$ is 1) positive definite 2) positive semi-definite?

If $\gamma=0$, $C$ is positive semidefinite, as $\gamma$ increases, the "positive semi-definiteness" of $C$ decreases. At some point, it becomes, indefinite, and after a while, it will become a negative semidefinite matrix. I want to know for what gamma this happens.

Solution of Case 1

It is quite straightforward to prove that the solution of case 1 is $\gamma=\lambda_{max}(B^{-1}A)$ which follows from expressing $\gamma$ as the solution to a generalized eigenvalue problem of $A$ and $B$ and using the fact $B$ is invertible.

(please see this MSE question for a distantly related problem, but from which solution can't be deduced for this one).

share|cite|improve this question
The prove for case 1 is wrong, the author seems to ignore the fact that $B^{-1} A$ is not positive matrix or semi-positive matrix. – Wangyan Li Jun 30 '15 at 6:17
up vote 1 down vote accepted

For diagonal matrices, the above question can't be true!!. (thanks to the author of this answer, this edit was added by the questioneer).

Let $A=\pmatrix{1 & 0 \\ 0 & 0 }$, $B=\pmatrix{0 & 0 \\ 0 & 1 }$. Then $C=\pmatrix{1 & 0 \\ 0 & -\gamma }$, which can never be negative semi-definite.

Addendum (after question was modified):

Let $A=\pmatrix{1 & 1 \\ 1 & 1 }$, $B=\pmatrix{1 & -1 \\ -1 & 1 }$. Then $C=\pmatrix{1-\gamma & 1+\gamma \\ 1+\gamma & 1-\gamma }$. Then we note that $C \binom{ 1 }{ 1 } = 2 \binom{ 1 }{ 1 }$, and hence $C$ always has an eigenvalue at 2 and can never be negative semi-definite.

share|cite|improve this answer
oops! I missed that, edited the question, for non-diagonal matrices. Thanks a lot!! – dineshdileep Nov 4 '12 at 4:20
A slight modification yields non-diagonal matrices. – copper.hat Nov 4 '12 at 8:00

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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