4
$\begingroup$

How could we prove that

If $\lambda_1,\lambda_2,\lambda_3,\cdots \lambda_n $ are the eigenvalues of a non-singular square matrix $A$ then eigenvalues of adj $\space A$ are $\frac {\det A}{\lambda_1},\frac {\det A}{\lambda_2},\frac {\det A}{\lambda_3},\cdots \frac {\det A}{\lambda_n}$.

I stumbled upon this property while solving a MCQ type question, in the solution there is no proof, I was just wondering if anybody could show me how to prove this one.

Thanks,

$\endgroup$
  • 3
    $\begingroup$ Most uninformative title. $\endgroup$ – Did Jan 18 '12 at 16:40
  • $\begingroup$ @Didier:I tried to make it more informative but $150$ is the limit :( $\endgroup$ – Quixotic Jan 18 '12 at 17:05
  • 1
    $\begingroup$ ...The limit worries you because you put useless stuff in the title. Try something like Eigenvalues of adjoint of non-singular matrix. $\endgroup$ – Did Jan 18 '12 at 18:01
  • $\begingroup$ @Didier:That's cogent, i am putting it right now:) $\endgroup$ – Quixotic Jan 18 '12 at 18:28
8
$\begingroup$

As Davide answer shows, using the identity $adj(A)=\det(A)A^{-1}$ this problem can be reduced to showing that the eigenvalues of $A^{-1}$ are exactly the inverses of the ones of $A$.

This is intuitively obvious, since $Ax=\lambda x \Rightarrow \frac{1}{\lambda}x = A^{-1}x$, but there could be issues with the multiplicities.

To formally prove it, note that

$$\det(\lambda I -A^{-1}) = \frac{\det(A) \det(\lambda I -A^{-1})}{\det(A)}= \frac{\lambda^n \det(A- \frac{1}{\lambda}I)}{\det(A)} \,.$$

This way you can relate the characteristic polynomials of $A$ and $A^{-1}$.

$\endgroup$
  • $\begingroup$ This is more clear and neat, I got it just by reading once. Thanks :) $\endgroup$ – Quixotic Jan 18 '12 at 17:56
4
$\begingroup$

The key is the identity $\operatorname{adj} A\cdot A=\det A \cdot I_n$, and since $A$ is not singular we have $\operatorname{adj} A=\det(A)\cdot A^{-1}$. The eigenvalues of $A^{-1}$ are the respective inverses of the eigenvalues of $A$ with the same algebraic multiplicity as @N. S. showed.

$\endgroup$
  • $\begingroup$ $^t\operatorname{adj} A$ does this means transpose of $\operatorname{adj} A$? But then I am only aware of the identity $\operatorname{adj} A\cdot A= A \cdot \operatorname{adj} A=\det A \cdot I_n $ $\endgroup$ – Quixotic Jan 18 '12 at 15:52
  • $\begingroup$ What is your definition of $\operatorname{adj}(A)$? $\endgroup$ – Davide Giraudo Jan 18 '12 at 16:11
  • $\begingroup$ Same as here $\endgroup$ – Quixotic Jan 18 '12 at 16:22
  • $\begingroup$ Ok, what I took for adjoint is in fact the matrix $C$. So I will edit my answer in order to be conform with your link. $\endgroup$ – Davide Giraudo Jan 18 '12 at 21:36

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.