17
$\begingroup$

Let $A$ be a $m \times n$ matrix with entries from some field $F$. Define the determinant rank of $A$ to be the largest possible size of a nonzero minor, i.e. the size of the largest invertible square submatrix of $A$. It is true that the determinant rank is equal to the rank of a matrix, which we define to be the dimension of the row/column space.

It's not difficult to see that $\text{rank} \geq \text{determinant rank}$. If some submatrix of $A$ is invertible, then its columns/rows are linearly indepedent, which implies that the corresponding rows/columns of $A$ are also linearly indepedent.

Is there a nice proof for the converse?

$\endgroup$
30
$\begingroup$

If the matrix $A$ has rank $k$, then it has $k$ linearly independent lines. Those form an $k\times n$ submatrix, which of course also has rank $k$. But if it has rank $k$, then it has $k$ linearly independent columns. Those form a $k\times k$ submatrix of $A$, which of course also has rank $k$. But a $k\times k$ submatrix with rank $k$ is a full-rank square matrix, therefore invertible, thus is has a non-zero determinant. And therefore the determinant rank has to be at least $k$.

$\endgroup$
  • $\begingroup$ Nice! Now that I see this proof, I can't believe I missed it.. $\endgroup$ – spin Aug 27 '12 at 15:33
  • $\begingroup$ Why do you say "And therefore the determinant rank has to be at least k"? Doesn't your answer show that it will be exactly equal? Why "at least"? $\endgroup$ – agdhruv May 6 '18 at 13:05
  • $\begingroup$ @agdhruv: Because that is what is asked for in the question. $\endgroup$ – spin Jun 4 '18 at 5:57

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.