Linear Algebra: Lecturer: Tracy Craig E-Mail: T.s.craig@utwente - NL

Linear Algebra
Lecturer : Tracy Craig

E-mail: t.s.craig@utwente.nl
1/78 Electrical Engineering, Mathematics and Computer Science EEMCS

Recall: determinant
The determinant of a square matrix is a number.
! This number immediately indicates whether a matrix is invertible
! For 2 " 2 matrices, the determinant is associated with an area

and, for 3 " 3 matrices, the determinant is associated with a
volume which yields a connection to a scale factor needed in
linear transformations (Vector Calculus).

Recall
The determinant of a 2 ! 2 matrix

! "
a b
AD
c d
is given by:
det A D ad " bc

Recall: Definition
For n > 2 the determinant of an n ! n-matrix A D Œaij ! equals:
1Cn
det A D a11 det A11 " a12 det A12 C a13 det A13 C # # # C ."1/ a1n det A1n
n
X
1Cj
D ."1/ a1j det A1j
j D1
This is referred to as an expansion across the first row

Recall: Determinant and elementary row operations
Theorem Let A be an n ! n matrix. Then:
" If a multiple of one row is added to another row,

the determinant does not change
" If two rows are interchanged,

the determinant is multiplied by #1
" If a row is multiplied by a number k,

the determinant is multiplied by k

Recall: Determinant and elementary column operations
! If a multiple of one column is added to another column then the

determinant does not change.
! If two columns are interchanged then the determinant is

multiplied by "1.
! If a column is multiplied by a number k then the determinant is

multiplied by k.

Recall: Fast way to compute determinant
! Create, using elementary row- or column operations, a row or

column with as many zeros as possible
Take into account whether the determinant changes !!
! Use expansion for the determinant along this row or column.

Recall
Theorem If A is a triangular matrix then det A is the product of the

entries on the main diagonal of A.
Theorem If A is a square matrix, then

T
det A D det A:

Recall
Theorem Let A be an n ! n matrix. Then,
A is invertible if and only if det A ¤ 0

Recall
Theorem Let A and B be n ! n matrices. Then:
det.AB/ D det A " det B

Recall
Theorem Let A be a 2 ! 2 matrix. The area of the parallelogram

spanned by the columns of A is equal to:
j det Aj

Recall
Theorem The absolute value of the determinant of a 3 ! 3 matrix A

is the volume of the parallelepiped spanned by the columns of A.

Three masses connected by identical springs
k k
m M m
x1 x2 x3
mxU 1 D !k.x1 ! x2 /
M xU 2 D !k.x2 ! x1 / ! k.x2 ! x3 /
mxU 3 D !k.x3 ! x2 /

mxU 1 D !k.x1 ! x2 /
M xU 2 D !k.x2 ! x1 / ! k.x2 ! x3 /
mxU 3 D !k.x3 ! x2 /
Define:
˙ ! ˙ k k
!
x1 !m m 0
xD ; AD k k k :
x2 M !2 M M
k k
x3 0 m !m
This can be rewritten as:

xU D Ax:

Electrical circuit
R C2
RC1 xT 1 D !x1
C2 xT 2 D x3
C1 L LxT 3 D !x2

RC1 xT 1 D !x1
C2 xT 2 D x3
LxT 3 D !x2
Choose L D 1, C1 D C2 D 1 and R D 1. We get:
xT D Ax
where ˙ ! ˙ !
x1 !1 0 0
xD x2 ; AD 0 0 1
x3 0 !1 0

Definition
An eigenvector of an n ! n matrix A is a vector x ¤ 0 such that
Ax D !x (?)
A scalar ! is an eigenvalue of A if there exists a non-trivial solution

x of (?).

Example
˙! ˙ !
1 4 !1 6
xD 1 ; AD 2 1 6
1 2 !1 8
The vector x is an eigenvector of A with eigenvalue 9 since:
˙ !˙ ! ˙ !
4 !1 6 1 9
Ax D 2 1 6 1 D 9 D 9x
2 !1 8 1 9

Example
The matrix: ˙ !
4 !1 6
AD 2 1 6
2 !1 8
has eigenvalue 2.

˙ !
4 !1 6
AD 2 1 6
2 !1 8
˙ !
2 !1 6
Ax D 2x ” Ax!2I x D 0 ” .A!2I /x D 0 ” 2 !1 6 x D 0
2 !1 6
This system clearly has nontrivial solutions.

Note that A ! 2I is not invertible and so det.A ! 2I / D 0.

A scalar ! is an eigenvalue of an n ! n matrix A if and only if !
satisfies the characteristic equation:
det.A " !I / D 0
For an n ! n matrix A we call det.A " !I / the

characteristic polynomial of A.

Example
Determine the eigenvalues of the matrix

! "
2 3
AD
3 !6

Characteristic equation
! "
2 3
AD
3 !6
! "
2!! 3
A ! !I D
3 !6 ! !
2
0 D det.A ! !I / D .2 ! !/.!6 ! !/ ! 9 D ! C 4! ! 21
So the eigenvalues are ! D !7 and ! D 3.

Consider the matrix
! "
1 1
AD
1 1
Determine the eigenvalues of A

! "
1 1
AD
1 1
It can be derived from the characteristic polynomial
.1 ! !/.1 ! !/ ! 1 D 0
which yields:
2
! ! 2! D 0:

Example
Determine the eigenvalues of the matrix:

˙ !
4 !1 6
AD 2 1 6
2 !1 8

Row operations
˙ ! ˙ !
4!! !1 6 2!! 0 !!2
A ! !I D 2 1!! 6 " 0 2!! !!2
2 !1 8!! 2 !1 8!!

Column operations
˙ ! ˙ !
2!! 0 !!2 2!! 0 0
0 2!! !!2 " 0 2!! !!2
2 !1 8!! 2 !1 10 ! !
˙ !
2!! 0 0
" 0 2!! 0
2 !1 9!!

Characteristic equation
˙ ! ˙ !
4!! !1 6 2!! 0 0
A ! !I D 2 1!! 6 " 0 2!! 0
2 !1 8!! 2 !1 9!!
2
0 D det.A ! !I / D .2 ! !/ .9 ! !/
So the eigenvalues are ! D 2 and ! D 9.

Eigenvectors
Finding x such that:

Ax D !x
is equivalent to finding x such that:
.A ! !I /x D 0

Eigenvectors for ! D 9
˙ ! ˙ !
4 !1 6 !5 !1 6
AD 2 1 6 H) A! 9I D 2 !8 6
2 !1 8 2 !1 !1
So we must solve:
˙ !
!5 !1 6
2 !8 6 xD0
2 !1 !1

˙ !
Reduced echelon form
0 1
!5 !1 6 1 0 !1
B C
2 !8 6 "B 0 j 1 !1 C
B C
@ A
2 !1 !1 0 0 0
Simplified system:
8
< x1 ! x3 D 0
: x2 ! x3 D 0
So: ˙!
1
xD˛ 1
1
for any ˛ ¤ 0.
Eigenvectors for ! D 2
˙ ! ˙ !
4 !1 6 2 !1 6
AD 2 1 6 H) A! 2I D 2 !1 6
2 !1 8 2 !1 6
Simplified system:
n
2x1 ! x2 C 6x3 D 0
We can choose ˙! ˙ !
1 !3
xD˛ 2 Cˇ 0
0 1
for any ˛; ˇ not both equal to 0.

Recall
A scalar ! is an eigenvalue of an n ! n matrix A if and only if !

satisfies the characteristic equation:
det.A " !I / D 0
For an n ! n matrix A we call det.A " !I / the

characteristic polynomial of A.

Let A be an n ! n matrix. The following three conditions are
equivalent:
(i) The matrix A is invertible
(ii) The number 0 is not an eigenvalue of A.
(iii) The determinant of A is not zero.

Definition Let A be an n ! n matrix and let ! be an eigenvalue of
A. We define the eigenspace E! associated to the eigenvalue ! by:
n
E! D fv 2 R j Av D !vg
E! contains all possible eigenvectors associated to ! as well as 0.

We have:
E! D Null.A " !I /

Theorem Let A be a n ! n matrix and let fv1 ; v2 ; : : : ; vn g be a
n
basis of R consisting of eigenvectors with associated eigenvalues
!1 ; !2 ; : : : ; !n . Let
! "
P D v1 v2 " " " vn
!
be the matrix with the vi ’s as columns. In that case, we have:
!1 0 """ 0
"
:: ::
!1
0 !2 : :
P AP D :: :: ::
: : : 0
0 """ 0 !n

Definition We call a square matrix A diagonalizable if there exists
an invertible matrix P such that !1
P AP is a diagonal matrix.

Theorem A square n ! n matrix A is diagonalizable if and only if
n
there exists a basis of R consisting of eigenvectors of A.
n
In case A is an n ! n matrix with n distinct eigenvalues, then R has
a basis consisting of eigenvectors of A.

Let A be a n ! n matrix and let fv1 ; v2 ; : : : ; vn g be a basis of Rn
consisting of eigenvectors with associated eigenvalues !1 ; : : : ; !n ,
then
!1
P AP D D;
where
! "
" P D v1 # # # vn
" D is the diagonal matrix with the eigenvalues !1 ; : : : ; !n on its

main diagonal (in this order!).

Example
Consider the matrix ˙ !

4 !1 6
AD 2 1 6
2 !1 8
The eigenvalues are ! D 2 and ! D 9, with eigenspaces
8̂ ˙!˙ ! 9 8̂ 9 ˙!
ˆ
ˆ 1 !3 >>
> ˆ
ˆ 1 >
>
>
< = < =
E2 D Span 2 ; 0 E9 D Span 1
ˆ
ˆ >
> ˆ
ˆ >
>
:̂ 0 1
>
; :̂ 1 ;
>

The eigenvalues are ! D 2 and ! D 9, with eigenspaces
8̂ ˙!˙ ! 9 8̂ 9 ˙!
ˆ
ˆ 1 !3 >>
> ˆ
ˆ 1 >
>
>
< = < =
E2 D Span 2 ; 0 E9 D Span 1
ˆ
ˆ >
> ˆ
ˆ >
>
:̂ 0 1
>
; :̂ 1
>
;
A basis of eigenvectors is fp1 ; p2 ; p3 g, where

˙! ˙ ! ˙!
1 !3 1
p1 D 2 ; p2 D 0 ; p3 D 1 ;
0 1 1
corresponding to !1 D 2; !2 D 2 and !3 D 9.

˙! ˙ ! ˙!
1 !3 1
p1 D 2 ; p2 D 0 ; p3 D 1 ;
0 1 1
corresponding to !1 D 2; !2 D 2 and !3 D 9. Now take
˙ ! ˙ !
1 !3 1 2 0 0
P D 2 0 1 DD 0 2 0
0 1 1 0 0 9
Note that the order of the eigenvectors in P corresponds to the order
of the eigenvalues in D. We get:
!1
P AP D D

Three masses connected by identical springs
k k
m M m
x1 x2 x3

mxU 1 D !k.x1 ! x2 /
M xU 2 D !k.x2 ! x1 / ! k.x2 ! x3 /
mxU 3 D !k.x3 ! x2 /
We have:
xU D Ax
where ˙ k k
!
!m m 0
AD k k k :
M !2 M M
k k
0 m !m

˙ k k
!
!m m 0
AD k k k :
M !2 M M
k k
0 m !m
We can easily verify that:

! "! "
k k k
det.A ! !I / D !! ! C !C2 C
m M m
We get
k k k
!1 D 0; !2 D ! ; !3 D !2 !
m M m

Note that these eigenvalues are distinct and hence the matrix A is
diagonalizable. If we write
!1
P AP D D
then we can define:

!1
pDP x
and we get:
pU D Dp:

If ˙ !
!1 0 0
DD 0 !2 0
0 0 !3
then we find:
pU1 D !1 p1
pU2 D !2 p2
pU3 D !3 p3
Since these differential equations are no longer coupled, these
equations can be easily solved using the techniques from Calculus
(homogeneous second order linear differential equations with
constant coefficients).

The first differential equation will have a solution describing a linear
translation over time:
p1 .t / D a1 C a2 t

The second equation is an oscillation where the middle mass stays
put and the outer two masses move symmetrically with respect to
each other.
r ! r !
k k
p2 .t / D b1 cos t C b2 sin t
m m

The final equation also yields an oscillation:
r ! r !
k k k k
p3 .t / D c1 cos t 2 C C c2 sin t 2 C
M m M m
However, this oscillation involves all three masses and is hence a bit
more involved.

The solution of the original differential equation is then given by:
˙ !
p1 .t /
x.t / D P p2 .t / D p1 .t /v1 C p2 .t /v2 C p3 .t /v3
p3 .t /
where v1 ; v2 ; v3 are the eigenvectors associated to the eigenvalues

!1 ; !2 ; !3 . In this special case we get:
˙! ˙ ! ˙ !
1 1 1
v1 D v2 D v3 D m
1 ; 0 ; !2 M :
1 !1 1

Are all matrices diagonalisable?
No! For example, take

! "
0 1 2
AD : Then det.A ! !I / D !
0 0
The only eigenvalue is ! D 0 but

8 9
< 1 =
!"
E0 D Span
: 0 ;
which yields no basis of eigenvectors.

Electrical circuit
R C2
RC1 xT 1 D !x1
C2 xT 2 D x3
C1 L LxT 3 D !x2

Choose: C1 D C2 D 1, L D 1 and R D 1
We get:
xT D Ax
where ˙ ! ˙ !
x1 !1 0 0
xD x2 ; AD 0 0 1
x3 0 !1 0

˙ !
!1 0 0
AD 0 0 1
0 !1 0
Let us first compute the eigenvalues of A. We have:

2
det.A ! !I / D !.! C 1/.! C 1/
We find only one real eigenvalue ! D !1.

The vector space Cn Example:
8̂ !" ˇ 9
# ˘ !1
x1
ˇ >
1Ci
ˆ
ˆ ˇ >
>
ˆ
ˆ ˇ >
>
x2
ˆ ˇ >
5
< =
n
C D
ˇ
ˇ xi 2 C 1!i 2C
:: ˇ
ˆ
ˆ
ˆ
ˆ : ˇ
ˇ
>
>
>
> 0
ˆ >
xn
:̂ ˇ >
;
ˇ 5

Addition Example
! " ! "! " ! "

u1 C v1 1 i 1Ci
u2 C v2 1Ci !1 C i 2i
uCvD :: C D
: 1!i i 1
un C vn i 1 1Ci

Complex conjugate Example
!" vx1
! "! " 1 1
vx2
x
vD 1Ci 1!i
:: D
: 1!i 1Ci
vxn i !i
Note that if v D a C bi with a and b real numbers then vx D a ! bi .

Scalar multiplication Example
! "
cv1
! "! "
1 1Ci
cv2 1Ci 2i
c!vD :: .c 2 C/ .1Ci /! D
: 1"i 2
cvn i "1 C i

Definition Consider a square n ! n matrix A. In that case ! 2 C is
called a (complex) eigenvalue of the matrix A if there exists a
nonzero vector x 2 Cn such that
Ax D !x
In that case x is called an eigenvector of A associated with the

eigenvalue !.

The matrix A and the vector v are given by:
! " ! "
!1 !5 3!i
AD vD
2 5 !2
Choose the best answer.
(a) v is an eigenvector of A, associated with eigenvalue 2 ! i .
(b) v is an eigenvector of A, associated with eigenvalue 2 C i .
(c) v is an eigenvector of A, associated with eigenvalue !2 ! i .
(d) v is not an eigenvector of A.

! " ! "
!1 !5 3!i
AD vD
2 5 !2
Solution
! "! " ! "

!1 !5 3!i !.3 ! i / C 10
Av D D
2 5 !2 2.3 ! i / ! 10
! " ! "
7Ci 3!i
D D .2 C i / D .2 C i /v:
!4 ! 2i !2

Lemma Let A be a real, square matrix with a complex eigenvalue
x is also an eigenvalue of the
!. In that case, the complex conjugate !
matrix A. Moreover, if x is an eigenvector associated with eigenvalue
x
! then x is an eigenvector associated with eigenvalue !.

Example
Determine eigenvalues and eigenvectors of the matrix

˙ !
0 !1 2
AD 1 !3 3
2 !3 2

˙ !
0 !1 2
AD 1 !3 3
2 !3 2
˙ ! ˙ 2
!
!! !1 2 0 !1 ! 3! ! ! 2 C 3!
A!!I D 1 !3 ! ! 3 " 1 !3 ! ! 3
2 !3 2!! 0 3 C 2! !4 ! !
2
det.A ! !I / D !.!! ! 3! ! 1/.!4 ! !/ C .3 C 2!/.2 C 3!/
3 2
D !! ! ! C 2
2
D !.! ! 1/.! C 2! C 2/

2
det.A ! !I / D !.! ! 1/.! C 2! C 2/
So the eigenvalues are 1, !1 ! i and !1 C i .

Next, we need to determine the eigenvectors!

˙ !
0 !1 2
AD 1 !3 3
2 !3 2
˙ ! 0 1
!1 !1 2 1 0 !1
B C
A! 1I D 1 !4 3 "B 0 j 1 !1 C
B C
@ A
2 !3 1 0 0 0
An eigenvector associated with eigenvalue ! D 1 is given by:

˙!
1
x1 D 1
1

˙ !
0 !1 2
AD 1 !3 3
2 !3 2
˙ ! ˙ !
1Ci !1 2 1 !2 C i 3
A ! .!1 ! i / I D 1 !2 C i 3 " 1Ci !1 2
2 !3 3Ci 2 !3 3Ci
˙ !
1 !2 C i 3
" 0 2Ci !1 ! 3i
0 1 ! 2i !3 C i

˙ ! ˙ !
1 !2 C i 3 1 !2 C i 3
A ! .!1 ! i / I " 0 2Ci !1 ! 3i " 0 1 !1 ! i
0 1 ! 2i !3 C i 0 1 ! 2i !3 C i
˙ !
1 0 !i
" 0 1 !1 ! i
0 0 0
An eigenvector associated with eigenvalue ! D !1 ! i is given by:

˙ !
i
x2 D 1Ci
1

An eigenvector associated with eigenvalue ! D !1 ! i is given by:
˙ !
i
x2 D 1Ci
1
Therefore, an eigenvector associated with eigenvalue ! D !1 C i is

given by:
˙ !
!i
x3 D 1!i
1

Electrical circuit
R C2
RC1 xT 1 D !x1
C2 xT 2 D x3
C1 L LxT 3 D !x2

Choose: C1 D C2 D 1, L D 1 and R D 1
We get:
xT D Ax
where ˙ ! ˙ !
x1 !1 0 0
xD x2 ; AD 0 0 1
x3 0 !1 0

Let us first compute the eigenvalues of A. We have:
2
det.A ! !I / D !.! C 1/.! C 1/
We find one real eigenvalue !1 D !1 and two complex eigenvalues

!2 D i and !3 D !i .

For the associated eigenvectors, we find
˙! ˙! ˙ !
1 0 0
v1 D 0 ; v2 D 1 ; v3 D 1 :
0 i !i

We define:
!1
pDP x
and we get:
pT D Dp:
which can be written as:
pT1 D !1 p1
pT2 D !2 p2
pT3 D !3 p3

We get using the techniques from Calculus:
!t it !it
p1 .t / D e p1 .0/; p2 .t / D e p2 .0/; p3 .t / D e p3 .0/
We then find:
˙ !
p1 .t /
x.t / D P p2 .t / D p1 .t /v1 C p2 .t /v2 C p3 .t /v3
p3 .t /

We get:
!t it !it
x.t / D e p1 .0/v1 C e p2 .0/v2 C e p3 .0/v3
˙! ˙!
1 0
! "
!t it 1 i
D e x1 .0/ 0 Ce 2 x2 .0/ ! 2 x3 .0/ 1
0 i
˙ !
0
! "
!it 1 i
Ce 2 x2 .0/ C 2 x3 .0/ 1
!i
˙ ! !t
˙ ! ˙ !
e 0 0
D x1 .0/ 0 C x2 .0/ cos t C x3 .0/ sin t
0 ! sin t cos t

Linear Algebra: Lecturer: Tracy Craig E-Mail: T.s.craig@utwente - NL

Uploaded by

Linear Algebra: Lecturer: Tracy Craig E-Mail: T.s.craig@utwente - NL

Uploaded by

Linear Algebra

Lecturer : Tracy Craig

1/78 Electrical Engineering, Mathematics and Computer Science EEMCS

The determinant of a square matrix is a number.

! This number immediately indicates whether a matrix is invertible

! For 2 " 2 matrices, the determinant is associated with an area

2/78 Electrical Engineering, Mathematics and Computer Science EEMCS

The determinant of a 2 ! 2 matrix

3/78 Electrical Engineering, Mathematics and Computer Science EEMCS

For n > 2 the determinant of an n ! n-matrix A D Œaij ! equals:

This is referred to as an expansion across the first row

4/78 Electrical Engineering, Mathematics and Computer Science EEMCS

Theorem Let A be an n ! n matrix. Then:

" If a multiple of one row is added to another row,

" If two rows are interchanged,

" If a row is multiplied by a number k,

5/78 Electrical Engineering, Mathematics and Computer Science EEMCS

! If a multiple of one column is added to another column then the

! If two columns are interchanged then the determinant is

! If a column is multiplied by a number k then the determinant is

6/78 Electrical Engineering, Mathematics and Computer Science EEMCS

! Create, using elementary row- or column operations, a row or

! Use expansion for the determinant along this row or column.

7/78 Electrical Engineering, Mathematics and Computer Science EEMCS

Theorem If A is a triangular matrix then det A is the product of the

Theorem If A is a square matrix, then

8/78 Electrical Engineering, Mathematics and Computer Science EEMCS

Theorem Let A be an n ! n matrix. Then,

A is invertible if and only if det A ¤ 0

9/78 Electrical Engineering, Mathematics and Computer Science EEMCS

Theorem Let A and B be n ! n matrices. Then:

det.AB/ D det A " det B

10/78 Electrical Engineering, Mathematics and Computer Science EEMCS

Theorem Let A be a 2 ! 2 matrix. The area of the parallelogram

11/78 Electrical Engineering, Mathematics and Computer Science EEMCS

Theorem The absolute value of the determinant of a 3 ! 3 matrix A

12/78 Electrical Engineering, Mathematics and Computer Science EEMCS

13/78 Electrical Engineering, Mathematics and Computer Science EEMCS

This can be rewritten as:

14/78 Electrical Engineering, Mathematics and Computer Science EEMCS

15/78 Electrical Engineering, Mathematics and Computer Science EEMCS

Choose L D 1, C1 D C2 D 1 and R D 1. We get:

16/78 Electrical Engineering, Mathematics and Computer Science EEMCS

An eigenvector of an n ! n matrix A is a vector x ¤ 0 such that

A scalar ! is an eigenvalue of A if there exists a non-trivial solution

17/78 Electrical Engineering, Mathematics and Computer Science EEMCS

18/78 Electrical Engineering, Mathematics and Computer Science EEMCS

19/78 Electrical Engineering, Mathematics and Computer Science EEMCS

This system clearly has nontrivial solutions.

20/78 Electrical Engineering, Mathematics and Computer Science EEMCS

For an n ! n matrix A we call det.A " !I / the

21/78 Electrical Engineering, Mathematics and Computer Science EEMCS

Determine the eigenvalues of the matrix

22/78 Electrical Engineering, Mathematics and Computer Science EEMCS

So the eigenvalues are ! D !7 and ! D 3.

23/78 Electrical Engineering, Mathematics and Computer Science EEMCS

Determine the eigenvalues of A

24/78 Electrical Engineering, Mathematics and Computer Science EEMCS

It can be derived from the characteristic polynomial

25/78 Electrical Engineering, Mathematics and Computer Science EEMCS

Determine the eigenvalues of the matrix:

26/78 Electrical Engineering, Mathematics and Computer Science EEMCS

27/78 Electrical Engineering, Mathematics and Computer Science EEMCS

28/78 Electrical Engineering, Mathematics and Computer Science EEMCS

So the eigenvalues are ! D 2 and ! D 9.

29/78 Electrical Engineering, Mathematics and Computer Science EEMCS

Finding x such that:

30/78 Electrical Engineering, Mathematics and Computer Science EEMCS

31/78 Electrical Engineering, Mathematics and Computer Science EEMCS

33/78 Electrical Engineering, Mathematics and Computer Science EEMCS

A scalar ! is an eigenvalue of an n ! n matrix A if and only if !

For an n ! n matrix A we call det.A " !I / the

34/78 Electrical Engineering, Mathematics and Computer Science EEMCS

(i) The matrix A is invertible

(ii) The number 0 is not an eigenvalue of A.

(iii) The determinant of A is not zero.