Let the adjugate of a matrix be defined as the transpose of the cofactor matrix, denoted $A^{*}$. (Also termed the Classical Adjoint)

It can be proven that for any $n\times n$ matrix $A$,

1) if rank($A$)= $n$ then rank(adj($A$)) = $n$

2) if rank($A$) = $n-1$ then rank(adj($A$)) = $1$

2) if rank($A$) < $n-1$ then rank(adj($A$)) = $0$

Suppose we were given an arbitrary $n\times n$ matrix $A$. If we know that rank((adj($A$)) = $1$, does it necessarily imply that the rank of $A$ is $n-1$? For example, suppose $A$ has an unknown entry but it is given that rank(adj($A$)) =$1$, is it a valid approach to conclude that the rank of $A$ is $n-1$ and proceed to determine the unknown value by solving $\det(A)=0$?

Similarly for the other two statements. Are they single direction statements or are they really if and only if statements?

  • $\begingroup$ I'm not familiar with the statement you posted. What is your definition of adjoint matrix? Either $\operatorname{rank} A = n$, $\operatorname{rank} A = n-1$, or $\operatorname{rank} A < n-1$. Assuming your statement is correct, if $\operatorname{rank}\operatorname{adj} A = 1$ and $\operatorname{rank} A \neq n-1$ you have a contradiction so $\operatorname{rank}(A)$ must be $n-1$. $\endgroup$ – tch Dec 30 '18 at 15:23
  • $\begingroup$ @TylerChen : thank you for the reply, an adjoint matrix is the conjugate transpose of a matrix. I updated the post with some queries. $\endgroup$ – NetUser5y62 Dec 30 '18 at 15:35
  • $\begingroup$ How do you use that definition of $A$ is not invertible? In any case, I don't think that is the definition of conjugate transpose. $\endgroup$ – tch Dec 30 '18 at 15:53
  • $\begingroup$ @TylerChen thank you for correcting that, I didn’t realize that was just a property thats applies when $A$ is invertible. I have removed it from the definition. $\endgroup$ – NetUser5y62 Dec 30 '18 at 16:02

1)Yes. If $\mathrm{rank}(\mathrm{adj}(A))=n$ then $\mathrm{adj}(A)$ is invertible, so $A=\mathrm{det}(A)\mathrm{adj}(A)^{-1}$ is invertible, because it can not be zero.

2)Yes. If $\mathrm{rank}(\mathrm{adj}(A))=1$ then $A$ is not invertible so $\mathrm{rank}(A)\leq n-1$ but by 3) we can not have $\mathrm{rank}(A)< n-1$, since it would imply $\mathrm{adj}(A)=0$. Thus $\mathrm{rank}(A)= n-1$.

3)Yes. If $\mathrm{adj}(A)=0$ then all $(n−1)×(n−1)$ minors of $A$ are zero, hence $\mathrm{rank}(A)\leq 2$


If by adjoint of $A$ you mean adjoint operator, i.e. $\operatorname{adj}(A) = A^*$ then your claimed properties are not true. For instance, any Hermetian matrix satisfies $A=A^*$ so the rank of $A$ is the rank of $A^*$.

More generally, for any matrix $A$, the conjugate transpose $A^*$ always has the same rank.

As an explicit example, consider the $3\times 3$ matrix: $$ A = \begin{bmatrix} 1 & 0 & 0 \\ 0 & 1 & 0 \\ 0 & 0 & 0 \end{bmatrix} $$

This is clearly rank $2$, but the conjugate transpose is also rank 2.

That said, if there exists some definition of "adjoint" satisfying the 3 properties you stated, then it is true that $\operatorname{rank}\operatorname{adj}(A)=1$ implies $\operatorname{rank}(A) = n-1$. To see this, note that if $\operatorname{rank}(A)$ did not equal $n-1$ then $\operatorname{rank}\operatorname{adj}(A)\neq1$

  • $\begingroup$ $\mathrm{adj}(A)$ is the transpose of the cofactor matrix of $A$ which make it different from $A^*$ the transpose of the conjugate. $\endgroup$ – jijijojo Dec 31 '18 at 15:15

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.