2
$\begingroup$

I am calculating the characteristic polynomial for this matrix:

$$A = \begin{pmatrix} 1 & 2 & \cdots & n \\ 1 & 2 & \cdots & n \\ \vdots & \vdots & \cdots & \vdots \\ 1 & 2 &\cdots & n \end{pmatrix}$$ First I was asked to figure out that $0$ is an eigenvalue, and since it is not invertible then $0$ is an eigenvalue, and its' geometric multiplicity is $n-1$. Now I need to calculate the characteristic polynomial but I am finding this determinant hard!

$$\mbox{det}\begin{pmatrix} \lambda - 1 & -2 & \cdots & -n \\ -1 & \lambda - 2 & \cdots & -n \\ \vdots & \vdots & \cdots & \vdots \\ -1 & -2 & \cdots & \lambda -n \end{pmatrix} = ? $$

$\endgroup$
  • 2
    $\begingroup$ Hint: You know $0$ is an eigenvalue o multiplicity $n-1.$ Calculating the trace should tell you the other eigenvalue. $\endgroup$ – Geoff Robinson Jun 1 '13 at 14:55
4
$\begingroup$

The last eigenvalue $\lambda$ is the trace of the matrix $A$ so $$\lambda=1+2+\cdots+n=\frac{n(n+1)}{2}$$ hence the chararcteristic polynomial is $$\chi_A=x^{n-1}\left(x-\frac{n(n+1)}{2}\right)$$

$\endgroup$
  • $\begingroup$ Why is the last eigenvalue the trace of the matrix? $\endgroup$ – TheNotMe Jun 1 '13 at 15:02
  • $\begingroup$ @TheNotMe: you are aware that the trace is equal to the sum of all the eigenvalues, and you also know that $n-1$ of the eigenvalues are zero. Thus... $\endgroup$ – J. M. is a poor mathematician Jun 1 '13 at 15:17
  • $\begingroup$ Is that always valid? I mean, $A$ is not a diagonal matrix, is your theorem still valid? $\endgroup$ – TheNotMe Jun 1 '13 at 15:41
  • $\begingroup$ @TheNotMe To determine the Characteristic polynomial it suffices to know the eigenvalues of the matrix whatever it is diagonalizable or not diagonalizable $\endgroup$ – user63181 Jun 1 '13 at 15:48
  • $\begingroup$ Well, if you don't mind me asking an extra question. Given that exact above matrix $A$. How do we figure out what the minimum ploynomial is? I mean surely I can't try $n-1$ options to figure out which option gives $m_A(A) = 0$. Any idea? $\endgroup$ – TheNotMe Jun 1 '13 at 15:52
0
$\begingroup$

Suggestion: add to the first column all the others. We get $\lambda-\frac{n(n+1)}2$ for each entry. Then use linearity with respect to this column, and finally do $L_j\leftarrow L_j+jL_1$.

$\endgroup$

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.