1

Suppose that $BAA' = CAA'$ holds, where $A$ is a $n\times k$ matrix and $C,B$ are (arbitrary) matrices of size $n\times n$. Then it follows that $$BA = CA.$$ I don't think it's true but I could not come up with a counter example. If I set $n=k=1$, the above obviously holds. For $n=k=2$ and possible non-singular choices of $A$, $B$ and $C$, I couldn't come up with a working example because I usually end up with $0 = 0$ implies $0 = 0$... $n=k\leq 2$ couldn't help either. Any hints?

1 Answers1

3

The statement is equivalent to: $XAA^T = 0$ implies $XA=0$ (just take $X=B-C$).

If $XAA^T=0$, then $XAA^TX^T=0$, and therefore $Trace((XA)(XA)^T) = Trace(XAA^TX^T)=0$. But for any matrix $M$, $Trace(MM^T)$ is the sum of the squares of the entries of $M$. Therefore $XA=0$.

Catalin Zara
  • 6,187
  • nice! It seems as if I was too convinced that the statement is wrong to actually consider a prove for it...

    btw if one does not want to use the trace argument: https://math.stackexchange.com/questions/1044527/prove-that-aat-0-implies-a-0

    – Syd Amerikaner Jan 06 '18 at 00:31