5
$\begingroup$

Let $A=\left[\begin{matrix} 0 & 1\\1 & 0 \end{matrix}\right]$ represent reflection about the line $y=x$.

I can calculate eigenvalues and eigenvectors mathematically, but I have hard time getting the results geometrically. There are some similar posts available but they are not of any help. For example in this post, I don't know how they found 1 and -1 as eigenvalues? I must be missing some points. So the question is,

how to find eigenvalues and eigenvectors geometrically? Why in this case eigenvalues are 1 and -1? and how to obtain corresponding eigenvectors?

$\endgroup$
2
  • 3
    $\begingroup$ Draw a few vectors and their images. Try to understand when it's the case that a vector and its image have the same direction (we count the exact opposite direction as the same direction too; it corresponds to a negative eigenvalue). Those vectors are eigenvectors. $\endgroup$ Commented Jul 15, 2022 at 15:41
  • 2
    $\begingroup$ A reflection can't change lengths, so its eigenvalues can only be $1$ and $-1$. The eigenvector corresponding to $1$ is a vector left unchanged by the reflection, while the one corresponding to $-1$ gets turned completely around by the reflection. $\endgroup$ Commented Jul 15, 2022 at 16:00

1 Answer 1

6
$\begingroup$

$A=\left[\begin{matrix} 0 & 1\\1 & 0 \end{matrix}\right]$

Consider the linear map associated with $A$ , $T_A:\Bbb{R}^2\to\Bbb{R}^2$ defined by $T_A(x, y) =(y, x) $

$\lambda$ is an eigenvalue of $T$ implies $\exists v\in\Bbb{R^2}\setminus\{0\}$ such that $\text{span}\{v\}$ is $1$- dimensional invariant subspace of $\Bbb{R^2}$ i.e for any vector $u\in \text{span}\{v\} , Tu\in\text{span}\{v\}$

A one dimensional subspace of $\Bbb{R}^2$ is a line through origin.

Hence you have to find a line through origin such that for any point $(x,y)\in\Bbb{R}^2$ on that line, after applying the transformation $T$ i.e after switching the co-ordinate the point still lie on that line.

Choose the line $y=x$ i.e the subspace $\{(x,x):x\in\Bbb{R}\}$ . Now it's easy to see after switching the co-ordinate it will be still one the line $ y=x$ . In that case eigenvalue is $1$.

There is another one $, y=-x$ i.e the subspace $\{(x,-x):x\in\Bbb{R}\}$ . Then after switching the co-ordinate any points will be in the same line but change it's direction ( correspond negative eigenvalue $-1$)

$\endgroup$
2
  • $\begingroup$ why did you not pick the subspace {(-x, x : x in R)}? In this case direction vector would be different, i.e. v= (x, -x) $\endgroup$
    – BasicTex
    Commented Jul 17, 2022 at 20:29
  • 1
    $\begingroup$ $U_1=\{(x,-x):x\in \Bbb{R}\}$ and $U_2=\{(-x,x):x\in \Bbb{R}\}$ represent the same subspace. They are the same line. Draw it. A subspace is closed under scalar multiplication. $(x, -x) \in U_1$ implies $-1\cdot (x, -x) =(-x, x) \in U_1$ $\endgroup$ Commented Jul 19, 2022 at 8:23

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .