LAG13

Fundamentals of Linear Algebra and Geometry
Lecture 13: Eigenvalues
In Lecture 12 we have introduced the notion of similarity for square matrices: two n×n matrices
A and B are similar if there exists an invertible n × n matrix H such that B = H −1 AH.
We have also seen that two matrices are similar if and only if there exists a basis B or Rn such
that B = M (fA , B) and then H = M (idV ; B, En ). During our discussion it will be convenient to
use both the point of view of matrices and that of endomorphisms.
Given two matrices A, B ∈ M (n × n, R) we would like to tell whether they are similar (and find
a matrix H such that B = H −1 AH. We remark first that similarity is an equivalence relation
on matrices. So we will rephrase our question as: given a matrix A finding “the simplest
possible”matrix A0 similar to A. Then A and B are similar if and only if they have the same
“simplest form”. This formulation is vague, we have to say what we mean by “simple”. We will
mostly be interested in a diagonal form (i.e. all entries outside the diagonal are zero)¿
Definition A matrix A ∈ M (n × n, R) is diagonalisable if there exists a diagonalmatrix
D ∈ M (n × n, R) similar to A. Equivalent: an endomorphism f : V → V is diagonalisable if
there exists a basis B = {v1 , . . . vn } of V such that M (f ; B) is a diagonal matrix.
By the definition of M (f ; B), that means that there exist scalars λ1 , . . . , λn ∈ R such that
f (vi ) = λi vi i.e. f acts as an homothety on the basis vectors. Vectors with this property
are called eigenvectors and the coefficients eigenvalues. So we may say that f : V → V is
diagonalisable if and only if there exists a basis of eigenvectors.
For any λ ∈ R, it is readily seen that the subset V (λ) = {v ∈ V | f (v) = λv} = ker (f − λidV )
is a subspace of V , called eigenspace. So λ is an eigenvalue if and only if f − λidV is not
injective.
If A = M (f, C) is the matrix of f in any basis C, then M (f − λidV , C) = A − λIn . So λ is an
eigenvalue for f if and only if rk (A − λIn ) dim V . In the next lecture, we will introduce a
tool to detect this property.
These topics are detailed in § 1 and 2 of Chapter 4. Take the time to read these carefully. Then
proceed with the exercises below. Post your solutions, questions etc. in the thread Lecture 13
in the Exercises Forum.
Solutions will be published at the end of the day.
 
1 2 3 4
4 3 2 1
Exercise 1. Let A = 1 1 1
. Prove that 2 is an eigenvalue and compute an eigenvector.
1
1 2 1 5
Can you find another eigenvalue for A?
Exercise 2. Let f (x, y, z) = (3x, 5x+2y−z, 3z) and recall the subspace U =< (1, 2, 3), (1, 3, 2) >
and W =< (0, 1, 0) > of exercise 3 in Lecture 8.
a) Write the matrix M of f in the canonical bases.
b) Compute f (U ) and f (W ).
c) Is f diagonalisable?
d) If it is, find the matrix H such that H −1 M H is diagonal.

a b a c
Exercise 3. Show that the matrix is similar to its trasnpose .
c a b a
[Hint: elementary operations.]
Solutions
Solution to Exercise 1. The matrix A will have the eigenvalue 2 if and only if rk (A−2I4 ) ≤ 3.
So we compute the echelon form of
       
1 2 3 4 2 0 0 0 −1 2 3 4 −1 2 3 4
4 3 2 1 0 2 0 0  4 1 2 1  0 9 14 17
1 1 1 1 − 0 0 2 0 =  1 1 −1 1
A − 2I4 =       
 0 3 2 5

1 2 1 5 0 0 0 0 1 2 1 3 0 4 4 7
       
−1 2 3 4 −1 2 3 4 −1 2 3 4 −1 2 3 4
 0 9 14 17  0 1 1 2 0 1 1 2 0 1 1 2
       
 0 3 2 5  0 9 14 17  0 0 −4 −1  0 0 −4 −1
0 1 1 2 0 3 2 5 0 0 −4 −1 0 0 0 0
(with apologies for my idiosyncrasy for preferring to avoid denominators).
So indeed rk (A − 2I4 ) = 3, and 2 is an eigenvalue of A. Furthermore the eigenspace is
V (2) = ker(A − 2I4 ) =< (1, −6, −1, 4) > .
It is not difficult to see that rk A ≤ 3, as the sum of the first two rows is 5 times the third. So
0 is also an eigenvalue.
 
3 0 0
Solution to Exercise 2. We compute M = M (f, E3 ) = 5  2 −1 . Furthermore:
0 0 3
f (U ) =< f (1, 2, 3), f (1, 3, 2) >=< (3, 6, 9), (3, 9, 6) >= U ;
f (W ) =< f (0, 1, 0) >=< (0, 2, 0) >= W.

In the process, we notice that f multiplies by 3 the vectors in U and by 2 the vectors in W .
So U = V (3) and W = V (2). The endomorphism is diagonalisable because we know that
U ⊕ W = R3 , so {(1, 2, 3), (1, 3, 2), (0, 1, 0)} is a basis of eigenvectors. This also gives
 
1 1 0
H = 2 3 1 
3 2 0

a b
Solution to Exercise 3. Taking the hint, we notice that if we permute the rows of
c a
and then permute the colums,
weend up with the transpose. Now permuting the rows is
0 1
multiplying on the left by and permuting the colums is multiplying of the right by
1 0
the same matrix. Since
−1
0 1 0 1 1 0 0 1 0 1
= ⇐⇒ =
1 0 1 0 0 1 1 0 1 0
we are done:
a c 0 1 a b 0 1
= .
b a 1 0 c a 1 0

LAG13

Uploaded by

Copyright:

Available Formats

LAG13

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

LAG13

Uploaded by

Copyright:

Available Formats

Fundamentals of Linear Algebra and Geometry

Lecture 13: Eigenvalues

a) Write the matrix M of f in the canonical bases.

f (W ) =< f (0, 1, 0) >=< (0, 2, 0) >= W.

You might also like