Let us consider $A,B$ two $n \times n$ matrices with rational elements. Prove that if $(A+B)^2=AB$, then $\det(AB-BA)=0$.

My try: By the conclusion, I tried to prove that $\mathrm{rank}(AB-BA) \lt n$ and I thought about Sylvester's or Frobenius' inequalities. I can't go any further.

  • $\begingroup$ As it is stated the question is false. For a counterexample see here. $\endgroup$
    – user26857
    Jan 12, 2021 at 22:25
  • $\begingroup$ I suppose the OP wanted to write $(A-B)^2=AB$. This question is solved here. $\endgroup$
    – user26857
    Jan 12, 2021 at 22:40

1 Answer 1


Let $a = \frac{3+\sqrt{5}}{2}$ and $b = \frac{3-\sqrt{5}}{2}$, $ab = 1$.
Then, $\det(A-aB) = c+d\sqrt{5} $ and $\det(A-bB) = c-d\sqrt{5}$.
So $\det(A-aB)\det(A-bB)$ is a rational number.
Thus, by calculating $(A-aB)(A-bB)$ and then applying the determinant, we obtain that $\det(AB-BA) = 0$.

This problem is similar to the following problem:

Let $A$, $B$ be two $n\times n$ matrices with real elements. Prove that if $A^2+B^2=AB$, then $\det(AB−BA)=0$.
Here you take $\epsilon$ a non-real number, $\epsilon^3=1$, the root for the equation $x^2+x+1=0$.
Then you calculate $(A+\epsilon B)(A+\epsilon^2 B)$, apply the determinant, and then the conclusion is proved.

  • $\begingroup$ For the second claim you probably want $n$ not divisible by $3$. $\endgroup$
    – user26857
    Jan 12, 2021 at 21:32
  • $\begingroup$ For $n=3$ you can find a counterexample here: math.stackexchange.com/questions/1768707 $\endgroup$
    – user26857
    Jan 12, 2021 at 22:14
  • $\begingroup$ To conclude: the question is false and this answer as well. $\endgroup$
    – user26857
    Jan 13, 2021 at 14:31

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.