# can eigenvector be found without computing the eigenvalue [closed]

Is there any ways to compute the eigen vector without computing explicitly the associated eigenvalue?

Actually, I'd like to compute the largest eigenvalue of a positive matrix from its eigen vector, so I have to know its eigenvector first.

-

## closed as off topic by Fernando Muro, Suvrit, Lee Mosher, Andres Caicedo, Igor RivinSep 23 '12 at 18:47

Questions on MathOverflow are expected to relate to research level mathematics within the scope defined by the community. Consider editing the question or leaving comments for improvement if you believe the question can be reworded to fit within the scope. Read more about reopening questions here. If this question can be reworded to fit the rules in the help center, please edit the question.

–  Goldstern Sep 23 '12 at 8:43
Obviously no. If you manage to compute an eigenvector $\mathbf{v}$ of a matrix $A$ then to automatically get the associated eigenvalue $\lambda$ by the formula $A\mathbf{v}=\lambda\mathbf{v}$. –  Fernando Muro Sep 23 '12 at 9:53
@Fernando: the comment is a bit tautological, since this is exactly what the OP is proposing to do. –  Igor Rivin Sep 23 '12 at 18:45
@Igor: you're right, it seems I didn't pay attention to the question's second paragraph. Anyway, everything looks tautological in this post. –  Fernando Muro Sep 28 '12 at 8:22

If you pick a random vector $v$ and look at $v_n=A^n v/\| A^n v\|,$ that will converge to the dominant eigenvector.

-
Small correction: pick a positive vector. If you just pick at random you might get very unlucky and pick one that is orthogonal to the eigenvector. –  Felix Goldberg Sep 24 '12 at 10:20
Good point (in the OP's stochastic context...) –  Igor Rivin Sep 25 '12 at 4:09

What do you know about the matrix?

If we know that the rows all have the same sum (but not what that sum is) then we would essentially find it by multiplying by the corresponding eigenvector, $\mathbb{j}$, the all $1$'s vector. This will be the largest eigenvalue provided that the entries are non-negative.

One way this could happen (but not the only one) is if the rows are identical or merely each is a permutation of the first.

If certain rows are equal then we know that $0$ is a eigenvalue although we never "computed" it. Then we do know an eigenvector.

In a circulant matrix we know all the eigenvectors (not just $\mathbb{j}$) and we essentially use them to compute the corresponding eigenvalues.

-
Aaron, Actually I'd like to compare the spectral radius of 2 matrices, say A1 and A2. The information about the matrices are: - They are doubly sub-stochastic matrix, with one row sum = a<1 ; their determinant are the same; their traces are the same. we know that they both have spectral radius between a and 1, but I'd like to compare them without computing them explicitly. E.g: A1 = [1/4 1/4 1/4 0; 1/4 3/4 0 0; 1/4 0 1/2 1/4; 0 0 1/4 3/4] A2 = [2/4 1/4 0 0; 1/4 1/4 1/4 1/4; 0 1/4 3/4 0; 0 1/4 0 3/4] We get, rho(A1) < rho(A2), but can we infer w/o computing it? – hayu 0 secs ago –  hayu Sep 23 '12 at 13:57