1
$\begingroup$

I got the eigenvalues which equal $4$, $-2$ and $-2$. The matrix is $$\begin{pmatrix}1 &-3& 3 \\3& -5 &3\\ 6 &-6 &4\end{pmatrix}$$

Now, usually, when I solve these, the second row always ends up automatically equaling zero. Is the same method applied when solving for a $3\times3$ matrix? Only ending up with the top row? Because I'm messing around with the matrix and I think that's possible.

Edit: I'm messing around with it and I don't think that is possible.

$\endgroup$
1
$\begingroup$

The eigenvalues you got are correct. I'm not sure what you mean by one row always equaling zero. It is true that when you compute $A-\lambda I$ and row-reduce it, that you will get at least one zero row precisely when $\lambda$ is an eigenvalue. You may get more than one zero row (in the case you have given, you will) when the dimension of the eigenspace is greater than one.

$\endgroup$
  • $\begingroup$ Ahh! I see. Thanks $\endgroup$ – user149635 May 14 '14 at 14:05
2
$\begingroup$

Eigenvalues solve the equation $\det A-\lambda I = 0$, which means that the eigenvalues $\lambda$ are precisely the values where the matrix $A-\lambda I$ is singular, as any matrix with determinant zero is singular.

Whenever a matrix is singular, it has a null space of dimension greater than zero, which is another way of saying "row reduction induces a fully-zero row."

Similarly, eigenvectors solve $Ax = \lambda x$, or $(A-\lambda I)x = 0$. This is another way of saying that eigenvectors are precisely the vectors in the null space of $A-\lambda I$. If the null space of $A-\lambda I$ has dimension 1, then when solving $(A-\lambda I)x=0$, you will get a single zero row. If the null space has dimension 2, you'll get two such rows, and so on.

In fact, eigenvectors form a basis for the null space of $A-\lambda I$. So if you know what the eigenvalues are, then finding the eigenvectors is nothing more than computing the basis of the null space of $A-\lambda I$ for each eigenvalue $\lambda$.

$\endgroup$

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.