2
$\begingroup$

I understand that if $v_1,..,v_r$ are the eigenvectors that correspond to distinct eigenvalues then they are linearly independent (*)

However what if I have say two linearly independent eigenvectors corresponding to one eigenvalue and an eigenvector corresponding to another, with $A$ a $3$x$3$ matrix and $Av=\lambda v$. Are these three eigenvectors linearly independent? Does this follow from (*)?

$\endgroup$
3
$\begingroup$

The answer is yes. Let's assumme that $v_1, v_2$ are the eigenvectors that correspond to the same eigenvalue $\alpha_1$. Observe that $\lambda_1 v_1 + \lambda_2 v_2$ is an eigenvector for the eigenvalues $\alpha_1$ and is therefore linearly independent of our third vector $v_3$. This means that if $\lambda_1 v_1 + \lambda_2 v_2 + \lambda_3 v_3 = 0$ we necessarily have $\lambda_3 = 0$. Now this implies $\lambda_1 v_1 + \lambda_2 v_2 = 0$, which by assumption yields $\lambda_1 = \lambda_2 = 0$.

Note that this is nothing else than observing that the sum of eigenspaces to different eigenvalues is a direct sum.

$\endgroup$
6
  • $\begingroup$ Why is $\lambda_1 v_1 + \lambda_2 v_2$ linearly independent of $v_3$? $\endgroup$ – usainlightning Aug 23 '15 at 22:07
  • 1
    $\begingroup$ $v_3$ is meant to be an eigenvector to an eigenvalue $\alpha_2 \ne \alpha_1$, so this follows from the lemma (*) in your post. $\endgroup$ – Dominik Aug 23 '15 at 22:09
  • $\begingroup$ Where you wrote $\lambda_2 v_2 + \lambda_2 v_2 + \lambda_3 v_3 = 0$ I assume you meant something else? $\endgroup$ – usainlightning Aug 23 '15 at 22:27
  • $\begingroup$ Indeed, there were some mistakes in the indices. I've corrrected it now. $\endgroup$ – Dominik Aug 23 '15 at 22:29
  • $\begingroup$ I don't understand why setting $\lambda_1 v_1 + \lambda_2 v_2 + \lambda_3 v_3 = 0$ implies that $\lambda_3=0$ $\endgroup$ – usainlightning Aug 23 '15 at 23:00

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.