3
$\begingroup$

I'm fairly certain I have this worked out but I want to make sure of it.

Let $A$ and $B$ be $5\times 5$ matrices and $P(A)$ be the vector space of all solutions to the homogeneous system $Ax=0$ . Given that $rank(B)=3$, then:

  1. $rank(AB)=3 \Leftrightarrow$ A is invertible.

  2. If $AB=0$ then $dim P(A)\leq 2$

I believe the first statement is false because the rank of the product of matricial multiplication is at most the rank of the matrix on the left side of the operation. If $A$ is invertible its rank is 5, but the question specified 3, so I know it's possible that there is some $A,\ 3\leq rank(A)<5\Rightarrow rank(AB)=3$ which disproves the statement.

As for the second statement I believe it is false as well. My reasoning is as follows: if $AB=0$ that means that every column of $B$ is itself a solution to $Ax=0$, and since $rank(B)=3$ that means that the dimension of $B$'s column space is also 3. That means that $dimP(A)= 3$ or 4 if $B$ doesn't contain a column of $0$s, but either way the statement is false.

Am I correct in my assertions?

$\endgroup$
  • $\begingroup$ Is this better? $\endgroup$ – Or Bairey-Sehayek Jan 6 '16 at 15:29
  • 1
    $\begingroup$ The reasoning for the first question is incorrect $\endgroup$ – Omnomnomnom Jan 6 '16 at 16:02
1
$\begingroup$

Your assertions are correct (both statements are false), however, your reasoning is a bit off.

It is indeed true that if $A$ is invertible, then $AB$ will have rank $3$, which is the same rank as $B$. I'm not sure where you're getting your inequalities from, but this situation is entirely possible.

It is not true, however, that if $AB$ has rank $3$, then $A$ must be invertible. For an example where this implication fails, consider $$ A = B = \begin{bmatrix} 1 & 0 & 0 & 0 & 0\\ 0 & 1 & 0 & 0 & 0\\ 0 & 0 & 1 & 0 & 0\\ 0 & 0 & 0 & 0 & 0\\ 0 & 0 & 0 & 0 & 0\\ \end{bmatrix} $$

For the second assertion, it suffices (and is technically necessary) to provide a counterexample. For instance, take $A = 0$.

$\endgroup$
  • $\begingroup$ $A = 0$ would not be counterexample, right, because then $\text{dim}(P(A)) = 0 \le 2$? Good catch on the first one, though, I did not look at the reasoning closely enough. $\endgroup$ – Eli Rose Jan 6 '16 at 16:08
  • $\begingroup$ @EliRose $\dim(P(A)) = 5 > 2$ $\endgroup$ – Omnomnomnom Jan 6 '16 at 16:09
  • $\begingroup$ Oh, any $x$ would be a solution -- gotcha. $\endgroup$ – Eli Rose Jan 6 '16 at 16:11
0
$\begingroup$
  1. You're correct. For an explicit example consider

$$ A = B = \begin{bmatrix} 1 & 0 & 0 & 0 & 0\\ 0 & 1 & 0 & 0 & 0\\ 0 & 0 & 1 & 0 & 0\\ 0 & 0 & 0 & 0 & 0\\ 0 & 0 & 0 & 0 & 0\\ \end{bmatrix} $$

  1. Yep. For example,

$$ AB = \begin{bmatrix} 1 & 0 & 1 & 0 & 0\\ 0 & 1 & 0 & 1 & 0\\ 0 & 0 & 1 & 0 & 1\\ 0 & 0 & 0 & 0 & 0\\ 0 & 0 & 0 & 0 & 0\\ \end{bmatrix} \begin{bmatrix} 1 & 0 & 0 & 0 & 0\\ 0 & 1 & 0 & 0 & 0\\ -1 & 0 & 1 & 0 & 0\\ 0 & -1 & 0 & 0 & 0\\ 0 & 0 & -1 & 0 & 0\\ \end{bmatrix} = \mathbf{0} $$

These matrices are in exactly the situation you described.

$\endgroup$

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.