# Meaning of eigenvectors and values of a 2x2 matrix (2nd order tensor)

1. Apr 11, 2012

### hiroman

Hi! I am a new user who is not an expert with Linear Algebra at all.

I have some questions about eigen values/vectors and their meaning with relation to a 2x2 matrix, or tensor, which was obtained by the tensor product of 2 vectors.

First, I have two 2-dimensional 2x1 vectors "v1" and "v2" on one point from which I wish to construct a 2x2 matrix "T" using tensor product, ie T=v1 (circle x) v2.

Then, I compute the eigen values and eigen vectors of the matrix (tensor) T.

Questions:

Is using tensor product the correct way to represent the vectors v1 and v2 on a 2x2 matrix T?

What's the meaning of the eigen values and eigen vectors of T? What is their relation with the original vectos v1 and v2? Also, most importantly, what is the meaning of having eigen values that are repeated?

I have read that if the eigen values of T are repeated, then that means that any eigen vector is associated with T, but still cannot figure out its underlying meaning with respect to the original vectors that constructed T.

Thanks!

2. Apr 12, 2012

### HallsofIvy

Geometrically, we can think of a two by two matrix. A, as "warping" points in a plane. If u and v are eigenvectors of A then they point in the "principal directions" of A. Think of four people holding a rubber sheet and pulling on it. Points on the line connecting two diagonally opposite people are just moved along that line. Those lines are in the direction of the eigenvectors and and the eigenvalues tell how far they are moved. Other points are moved part toward one person and partly toward another .

3. Apr 12, 2012

### hiroman

Thanks for the illustration on eigenvalues and eigenvectors. Then, is it correct to consider that eigenvectors of a matrix are the same if the orientation is different? Per the example, the eigenvectors would be the same if the people are stretching or contracting the rubber sheet?

4. Apr 12, 2012

### HallsofIvy

Yes, though in one case (stretching) the eigenvalues would be positive and in the other (contracting or compressing) they would be negative.