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

Let $A,B$ be a $n \times n$ matrices such that $\det(B) = 1$. Will the spectrum (set of eigenvalues) $AB$ be same as that of $A$. Or, at least is $\mbox{Trace}(A) = \mbox{Trace}(AB)$ ? If not, what can we say about the change in spectrum and trace in $A$ and $AB$.

share|cite|improve this question
The spectrum changes, even in dimension $2$. Consider the case $B$ diagonal, of diagonal elements $u$ and $u^{-1}$ where $u\neq 0$, and $A$ a diagonal matrix of diagonal elements $a_1$ and $a_2$. The spectrum of $AB$ is $\{ua_1,u^{—1}a_2\}$ and the trace of $AB$ is $ua_1+u^{-1}a_2$, which is not $a_1+a_2$, except special cases of $u$. – Davide Giraudo Jul 1 '12 at 13:02
up vote 1 down vote accepted

If two matrices $A,B$ have the same nonzero determinant, then $\det(A^{-1}B)=1$ so $A$ and $B$ are in the same class under multiplication by matrices of determinant $1$. Hence among invertible matrices the only invariant quantity under such multiplication is the determinant itself (of course any function of the determinant is also invariant; this is not very interesting). So there is no chance that the characteristic polynomial, spectrum, trace or whatever is invariant.

share|cite|improve this answer

No. Consider $n=2$, $A=I$, $B$ a diagonal matrix with entries $2, 1/2$ on diagonal. $\det B=1$, but both trace and spectrum of $AB$ are different from those of $A$.

You can't really say anything about the spectrum except that it will not change the product (that is, determinant), which is easy to see by generalizing the example. Choose any $A$, wlog $A$ is in Jordan normal form. Then by a suitable diagonal $B$ we can change the spectrum onto any whatsoever, as long as the product of it stays fixed (and there are as many zeroes for singular $A$).

From that we can see that the trace can be any $\sum_j x_j$ with $\prod x_j=\det (A)$. Unless $A$ is singular, in which case there must be as many zero $x_j$'s as there are zero eigenvalues of $A$ (without any restriction on the others), so for nonzero singular $A$ the trace can be any number.

share|cite|improve this answer
are there class of matrices which will not change the trace when multiplied? – DurgaDatta Jul 1 '12 at 13:08
DurgaDatta, commuting matrices maybe. – Peter Sheldrick Jul 1 '12 at 13:09
@PeterSheldrick: not really. Not even mutiples of identity matrix preserve trace (consider $-I$ for even dimension, for example), so there's little hope for any interesting class of matrices that would preserve trace. – tomasz Jul 1 '12 at 13:18

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.