# Eigenvector question

#### makris

Hi all,

I have the following question.

A = nxn non singular matrix
I = nxn identity matrix
li = eigevalues of A i=1,2...n
ui = eigenvectors corresponding to the previous eigenvalues.

It true that

( A - l1 * I ) * x =0

is satisfied by any vector of the form x = a1 * u1 where a1= arbitrary real number

Lanczos in his book Applied Analysis p. 61 claims that the following quadratic equation in A

( A - l1 * I ) * ( A - l2 * I ) * x = 0

is satisfied by an arbitrary linear combination of the first two eigenvectors

x = a1 * u1 + a2 * u2

It is not very obvious to me why this happens.
(Extending this to include n eigenvectors and eigenvalues will eventually lead to the so called Cayley-Hamilton theorem.)

I was wondering if you could give me a hint starting from first principals.

Thanks

Related Linear and Abstract Algebra News on Phys.org

#### lurflurf

Homework Helper
( A - l1 * I ) * ( A - l2 * I ) = ( A - l2 * I ) *( A - l1 * I )
so if
T=( A - l1 * I ) * ( A - l2 * I ) = ( A - l2 * I ) *( A - l1 * I )
T(a1 * u1)=a1*T(u1)=0
T(a2 * u2)=a1*T(u2)=0
hence by linearity
T(a1 * u1 + a2 * u2)=0

### Physics Forums Values

We Value Quality
• Topics based on mainstream science
• Proper English grammar and spelling
We Value Civility
• Positive and compassionate attitudes
• Patience while debating
We Value Productivity
• Disciplined to remain on-topic
• Recognition of own weaknesses
• Solo and co-op problem solving