CS6015: Linear Algebra and Random Processes
Lecture 16: The Eigenstory begins, computing eigenvalues and eigenvectors.
Learning Objectives
What is the eigenstory? (the outline of it)
(for today's lecture)
What are eigenvalues and eigenvectors?
How do you compute them?
What are eigenvalues/eigenvectors of some special matrices?
The Eigenstory
real
imaginary
distinct
repeating
(basis)
powers of A
PCA
optimisation
diagonalisation
How to compute eigenvalues/vectors?
What are the possible values?
What are the eigenvalues of some special matrices ?
What is the relation between the eigenvalues of related matrices?
What do eigen values reveal about a matrix?
What are some applications in which eigenvalues play an important role?
Identity
Projection
Reflection
Markov
Rotation
Singular
Orthogonal
Rank one
Symmetric
Permutation
det(A−λI)=0
det(A - \lambda I) = 0
trace
determinant
invertibility
rank
nullspace
columnspace
(positive semidefinite matrices)
positive pivots
(independent eigenvectors)
(orthogonal eigenvectors)
... ...
(symmetric)
(A−λI)x=0
(A - \lambda I)\mathbf{x} = \mathbf{0}
A⊤
A−1
AB
A⊤A
A+B
U
R
A2
A+kI
distinct values
independent eigenvectors
⟹
steady state
(Markov matrices)
Intuition
x
\mathbf{x}
The span of a vector is a line cu ∀c∈R
What happens when a matrix hits a vector?
Ax
A\mathbf{x}
[1221][12]
\begin{bmatrix}
1&2\\
2&1\\
\end{bmatrix}\begin{bmatrix}
1\\
2\\
\end{bmatrix}
=[54]
=\begin{bmatrix}
5\\
4
\end{bmatrix}
Most vectors get knocked off their span
(i.e., most vectors change their direction)
But some special vectors stay on the span
[1221][11]
\begin{bmatrix}
1&2\\
2&1\\
\end{bmatrix}\begin{bmatrix}
1\\
1\\
\end{bmatrix}
=[33]
=\begin{bmatrix}
3\\
3
\end{bmatrix}
These are called eigenvectors and the scaling factor is called the eigenvalue
=3[11]
=3\begin{bmatrix}
1\\
1
\end{bmatrix}
(i.e., they only get scaled or squished)
Definition
x
\mathbf{x}
The span of a vector is a line cu ∀c∈R
[1221][22]
\begin{bmatrix}
1&2\\
2&1\\
\end{bmatrix}\begin{bmatrix}
2\\
2\\
\end{bmatrix}
=[66]
=\begin{bmatrix}
6\\
6
\end{bmatrix}
Observation 1:
=3[22]
=3\begin{bmatrix}
2\\
2
\end{bmatrix}
If x is an eigenvector, then so is cx
(i.e., any vector in the span of x)
The corresponding eigenvalue also remains the same
The eigenvectors corresponding to an eigenvalue form a subspace
The vector x is an eigenvector of A if
Ax=λx
A\mathbf{x} = \lambda\mathbf{x}
λ is the corresponding eigenvalue
Definition
x
\mathbf{x}
The vector x is an eigenvector of A if
Ax=λx
A\mathbf{x} = \lambda\mathbf{x}
Observation 2:
λ is the corresponding eigenvalue
If x is an eigenvector, then it lies in the column space of A
(this is useful and we will return back to it later)
Definition
x
\mathbf{x}
The vector x is an eigenvector of A if
Ax=λx
A\mathbf{x} = \lambda\mathbf{x}
Observation 3:
λ is the corresponding eigenvalue
Eigenvectors only make sense for square matrices
(these two vectors can never be the same as they are in two different spaces)
Ax=λx′
A\mathbf{x} = \lambda\mathbf{x'}
m×n
m\times n
n×1
n\times1
m×1
m\times1
How do we compute the eigenvalues?
Ax=λx
A\mathbf{x} = \lambda\mathbf{x}
⟹Ax=λIx
\implies A\mathbf{x} = \lambda I\mathbf{x}
⟹Ax−λIx=0
\implies A\mathbf{x} - \lambda I\mathbf{x} = 0
(we now have a matrix on both sides)
⟹(A−λI)x=0
\implies (A - \lambda I)\mathbf{x} = 0
(trivial solution: x = 0 -- not very interesting)
(we are looking for a non-zero eigenvector)
⟹ null space of A−λI is non-zero
⟹ columns of A−λI are not independent
What does x being non-zero imply?
⟹ A−λI is not invertible (singular)
⟹ det(A−λI)=0
[1221]x=λx
\begin{bmatrix}
1&2\\
2&1\\
\end{bmatrix}\mathbf{x} = \lambda\mathbf{x}
[1221]x=λ[1001]x
\begin{bmatrix}
1&2\\
2&1\\
\end{bmatrix}\mathbf{x} = \lambda\begin{bmatrix}
1&0\\
0&1\\
\end{bmatrix}\mathbf{x}
[1221]x−λ[1001]x=0
\begin{bmatrix}
1&2\\
2&1\\
\end{bmatrix}\mathbf{x} - \lambda\begin{bmatrix}
1&0\\
0&1\\
\end{bmatrix}\mathbf{x} = 0
[1−λ221−λ]x=0
\begin{bmatrix}
1-\lambda&2\\
2&1-\lambda\\
\end{bmatrix}\mathbf{x}= 0
How do we compute the eigenvalues?
det(A−λI)=0
A=[1221]
A=\begin{bmatrix}
1&2\\
2&1\\
\end{bmatrix}
A−λI=[1−λ221−λ]
A-\lambda I=\begin{bmatrix}
1-\lambda&2\\
2&1-\lambda\\
\end{bmatrix}
det(A−λI)=(1−λ)⋅(1−λ)−2⋅2=0
det(A-\lambda I)= (1-\lambda)\cdot(1-\lambda) - 2\cdot2 = 0
λ2−2λ−3=0
\lambda^2 - 2\lambda - 3 = 0
(λ−3)(λ+1)=0
(\lambda - 3)(\lambda + 1) = 0
λ=3,λ=−1
\lambda = 3, \lambda = -1
(this is called the characteristic equation)
How do we compute the eigenvectors?
det(A−λI)=0
A=[1221]
A=\begin{bmatrix}
1&2\\
2&1\\
\end{bmatrix}
(A−λI)x=0
(A-\lambda I)\mathbf{x} = 0
λ=3,λ=−1
\lambda = 3, \lambda = -1
[1−3221−3]x=0
\begin{bmatrix}
1-3&2\\
2&1-3\\
\end{bmatrix}\mathbf{x} = \mathbf{0}
(Ax = 0 - we know how to solve this)
λ=3
\lambda = 3
[−222−2]x=0
\begin{bmatrix}
-2&2\\
2&-2\\
\end{bmatrix}\mathbf{x} = \mathbf{0}
x=[11]
\mathbf{x}=\begin{bmatrix}
1\\
1\\
\end{bmatrix}
[1−(−1)221−(−1)]x=0
\begin{bmatrix}
1-(-1)&2\\
2&1-(-1)\\
\end{bmatrix}\mathbf{x} = \mathbf{0}
λ=−1
\lambda = -1
[2222]x=0
\begin{bmatrix}
2&2\\
2&2\\
\end{bmatrix}\mathbf{x} = \mathbf{0}
x=[1−1]
\mathbf{x}=\begin{bmatrix}
1\\
-1\\
\end{bmatrix}
The n×n case
det(A−λI)=0
a11−λa21a31⋯⋯an1a12a22−λa32⋯⋯an2a13a23a33−λ⋯⋯an3⋯⋯⋯⋯⋯⋯a1na2na3n⋯⋯ann−λ
\begin{bmatrix}
a_{11}-\lambda&a_{12}&a_{13}&\cdots&a_{1n}\\
a_{21}&a_{22-\lambda}&a_{23}&\cdots&a_{2n}\\
a_{31}&a_{32}&a_{33}-\lambda&\cdots&a_{3n}\\
\cdots&\cdots&\cdots&\cdots&\cdots\\
\cdots&\cdots&\cdots&\cdots&\cdots\\
a_{n1}&a_{n2}&a_{n3}&\cdots&a_{nn}-\lambda\\
\end{bmatrix}
Characteristic Equation:
λn+⋯=0
\lambda^n + \cdots = 0
Observation:
For a n×n matrix
We expect n roots: n eigenvalues, n eigenvectors
(but sometimes things could go wrong)
What could go wrong?
The good case
The not-so-good case
A=[1221]
A=\begin{bmatrix}
1&2\\
2&1\\
\end{bmatrix}
λ=3,λ=−1
\lambda = 3, \lambda = -1
n real, distinct eigenvalues
n independent eigenvectors
(I see a basis there - coming soon)
A=[01−10]
A=\begin{bmatrix}
0&-1\\
1&0\\
\end{bmatrix}
1−λ221−λ=0
\begin{vmatrix}
1-\lambda&2\\
2&1-\lambda\\
\end{vmatrix} = 0
λ2−2λ−3=0
\lambda^2 - 2\lambda - 3 = 0
0−λ1−10−λ=0
\begin{vmatrix}
0-\lambda&-1\\
1&0-\lambda\\
\end{vmatrix} = 0
λ2+1=0
\lambda^2 + 1 = 0
λ=i,λ=−i
\lambda = i, \lambda = -i
imaginary eigenvalues
imaginary eigenvectors
(In many real world applications, imaginary values are not good)
A=[3013]
A=\begin{bmatrix}
3&1\\
0&3\\
\end{bmatrix}
3−λ013−λ=0
\begin{vmatrix}
3-\lambda&1\\
0&3-\lambda\\
\end{vmatrix} = 0
(3−λ)(3−λ)=0
(3 - \lambda)(3 - \lambda) = 0
λ=3,λ=3
\lambda = 3, \lambda = 3
repeating eigenvalues
<n independent eigenvectors
(I see an incomplete basis there - coming soon)
The Eigenstory
real
imaginary
distinct
repeating
A⊤
A−1
AB
A⊤A
(basis)
powers of A
PCA
optimisation
diagonalisation
A+B
U
R
A2
A+kI
How to compute eigenvalues?
What are the possible values?
What are the eigenvalues of some special matrices ?
What is the relation between the eigenvalues of related matrices?
What do eigen values reveal about a matrix?
What are some applications in which eigenvalues play an important role?
Identity
Projection
Reflection
Markov
Rotation
Singular
Orthogonal
Rank one
Symmetric
Permutation
det(A−λI)=0
det(A - \lambda I) = 0
trace
determinant
invertibility
rank
nullspace
columnspace
(positive semidefinite matrices)
positive pivots
(independent eigenvectors)
(orthogonal eigenvectors)
... ...
(symmetric)
(where are we?)
(characteristic equation)
(desirable)
(A−λI)x=0
(A - \lambda I)\mathbf{x} = \mathbf{0}
distinct values
independent eigenvectors
⟹
steady state
(Markov matrices)
EVs of some special matrices: Identity
I=100010001
I=\begin{bmatrix}
1&0&0\\
0&1&0\\
0&0&1\\
\end{bmatrix}
I=1−λ0001−λ0001−λ
I=\begin{bmatrix}
1-\lambda&0&0\\
0&1-\lambda&0\\
0&0&1-\lambda\\
\end{bmatrix}
(1−λ)3=0
(1-\lambda)^3= 0
λ1=1,λ2=1,λ3=1
\lambda_1=1,\lambda_2=1,\lambda_3=1
repeating eigenvalues
but no shortage of eigenvectors
(every n dimensional vector is an eigenvector of I)
EVs of some special matrices:Projection
x in column space of A
column space of A
p
\mathbf{p}
a1∈Rn
\mathbf{a}_1 \in \mathbb{R}^n
a2
\mathbf{a}_2
P=A(A⊤A)−1A⊤
P = A(A^\top A)^{-1}A^\top
If x∈Rn there are three possibilities
(Px = 1.x)
x is orthogonal to column space of A
(Px = 0.x)
x is at some angle to the column space of A
(Px != c.x) - direction will change
Possible eigenvalues: 0, 1 (nothing else)
Corresponding eigenvectors: nullspace of A⊤, column space of A
(angle = 0)
(angle = 90)
(angle != 0,90)
EVs of some special matrices:Rotation
A=[cosθsinθ−sinθcosθ]
A=\begin{bmatrix}
cos\theta&-sin\theta\\
sin\theta&cos\theta\\
\end{bmatrix}
A=[01−10]
A=\begin{bmatrix}
0&-1\\
1&0\\
\end{bmatrix}
θ=90°
\theta=90\degree
The purpose of a rotation matrix is to rotate (move) vectors to a different direction!
What is the question that we are asking?
(Which vectors will not move?)
Will we get a good answer?
(Obviously not!)
λ1=i
\lambda_1=i
u1=[i1]
u_1=\begin{bmatrix}
i\\1
\end{bmatrix}
λ2=−i
\lambda_2=-i
u2=[−i1]
u_2=\begin{bmatrix}
-i\\1
\end{bmatrix}
EVs of some special matrices:Singular
If A is singular (non-invertible), we know that
(non-zero)
(special solution(s) of Ax = 0)
Hence, 0 is always an eigenvalue of any singular (non-invertible) matrix
Ax=0
A\mathbf{x}=\mathbf{0}
Ax=0⋅x
A\mathbf{x}=\mathbf{0}\cdot\mathbf{x}
What are the corresponding eigenvectors?
I=111222333
I=\begin{bmatrix}
1&2&3\\
1&2&3\\
1&2&3\\
\end{bmatrix}
λ1=0,λ2=0,λ3=6
\lambda_1 = 0, \lambda_2 = 0, \lambda_3 = 6
x1=−301
\mathbf{x_1} = \begin{bmatrix}
-3\\0\\1
\end{bmatrix}
x2=−210
\mathbf{x_2} = \begin{bmatrix}
-2\\1\\0
\end{bmatrix}
x3=111
\mathbf{x_3} = \begin{bmatrix}
1\\1\\1
\end{bmatrix}
A is not invertible if and only if 0 is an eigenvalue of A
has a non-zero solution
Do EVs tell us anything about the rank?
A is not invertible if and only if 0 is an eigenvalue of A
⟹ if there is no 0 eigenvalue, A is invertible
⟹ if there is no 0 eigenvalue, A has rank n
False argument:
each 0 eigenvalue corresponds to one unique special solution of Ax=0
⟹ number of 0 eigenvalues = dimension of nullspace
⟹ number of 0 eigenvalues = n−r
⟹r=n − number of 0 eigenvalues
(Case1: no 0 eigenvalue)
(Case2: one or more 0 eigenvalues)
(Counter Example)
I=000300030
I=\begin{bmatrix}
0&3&0\\
0&0&3\\
0&0&0\\
\end{bmatrix}
r=2λ1=λ2=λ3=0
r=2\\
\lambda_1=\lambda_2=\lambda_3 = 0
Explore more in HW5
The Eigenstory
real
imaginary
distinct
repeating
A⊤
A−1
AB
A⊤A
(basis)
powers of A
PCA
optimisation
diagonalisation
A+B
U
R
A2
A+kI
How to compute eigenvalues?
What are the possible values?
What are the eigenvalues of some special matrices ?
What is the relation between the eigenvalues of related matrices?
What do eigen values reveal about a matrix?
What are some applications in which eigenvalues play an important role?
Identity
Projection
Reflection
Markov
Rotation
Singular
Orthogonal
Rank one
Symmetric
Permutation
det(A−λI)=0
det(A - \lambda I) = 0
trace
determinant
invertibility
rank
nullspace
columnspace
(positive semidefinite matrices)
positive pivots
(independent eigenvectors)
(orthogonal eigenvectors)
... ...
(symmetric)
(where are we?)
(desirable)
HW5
distinct values
independent eigenvectors
⟹
(characteristic equation)
(A−λI)x=0
(A - \lambda I)\mathbf{x} = \mathbf{0}
steady state
(Markov matrices)
Learning Objectives
What is the eigenstory? (the outline of it)
(achieved)
What are eigenvalues and eigenvectors?
How do you compute them?
What are eigenvalues/eigenvectors of some special matrices?
CS6015: Linear Algebra and Random Processes Lecture 16: The Eigenstory begins, computing eigenvalues and eigenvectors.
CS6015: Lecture 16
By Mitesh Khapra
CS6015: Lecture 16
Lecture 16: The Eigenstory begins, computing eigenvalues and eigenvectors
- 2,576
