Section

Tutorial Sheet 5
Eigenvalues and Eigenvectors

/ 1. Consider the linear system Ax = b, where

 
6 −2 1
A =  −4 12 −8 
4 −16 24

and b ∈ R3 is any given vector. Does the Jacobi iterative sequence of the given
system converge for every initial guess x(0) ∈ R3 ? Justify your answer.
W
2. Let A be a matrix whose eigenvalues are λ1 = −3, λ2 = 2, and λ3 = 9 with
corresponding eigenvectors v 1 = (1, 0, 1)T , v 2 = (0, 1, −1)T , and v 3 = (0, 0, 1)T .
To which eigenvalue and the corresponding eigenvector does the power method
converge (up to a subsequence) if we start with the initial guess x(0) = (2, −3, 5)?
Justify your answer.
⑥
3. Consider the matrix 
12.25 0.125 0.42


A =  −1.05 −14 0.5 

0.006 0.045 2.25
whose eigenvalues are real.

v
i) Without calculating the eigenvalues explicitly, show that the eigenvalues of
A can be labelled as λ1 , λ2 , and λ3 satisfying

|λ1 | > |λ2 | > |λ3 |.

⑥
ii) Construct the iterative sequences {µ } and {x
k } (based on power method)
(k)

converging to λ3 (as in (3i)) and a corresponding eigenvector, respectively.

⑤
iii) Starting with the initial guess x = (1, 2, 3), perform one iteration of the
(0)

sequences constructed in (3ii) using Doolittle factorization. (Inverse power

⑤
4. Find the optimal bounds for the eigenvalues of the matrix
methodl)
 
−7 0.5 −0.75
A=  0.65 5 0.4 
0 0.1 1
given by the Gerschgorin theorem.

1
S. Sivaji Ganesh and S. Baskar Spring 2022-23

/
1.

We that not dominant, 12

seeA is

Hence we cannot
diagonally
that
say convergence=1-41+1-8.
always
will occur.

12 commage, else
w ill
3(3) is it
-

B0
If
=
·

it won't.

2.

18887) (
2 =
(e) 2,2
=
=
-

3,
5
e z ez
- + =

0
ez
= =

As dominant
coeff for ingenvalue =p, i.e ez 5 =

is dominant
non-zero, and X isunique ugenvalue.So
power method is
we can
say
that
+ likely to comerge,
to
z (0,0,.)
=
and 1 0.
=
3.

i) (z -

12.25/10.645,12+14/1155,12-225/10.05.

All distinct and

eigen values are non
zero

(
-
i(8): (2.25,0) (122580)

ii) As we want
itto
converge
to the smallest
ign
value, we use the power method
inverse

Ay() x(0), M =py((((,x" y

=
(1)
=

=> Ay(+1) xe(H), MAT =

y(HM,CHINA
=

MA
iii) A 20 =

x(0)
(Vy() z
yy()
=
=

(z x(0)
=
(y() zw
=
= =
4. =)
(z 7(11.25 =)/(21 7(71.255.755/2118.25
+
-

(2 -

5)=105=)((21 5/21053.455/2/16.05
-

(z -
1) 50. =)1(21-11101 =) 0.PS/2/
=> 0.81/2/18.25

(2+7(10.65 =)6.355(2117.65
12 -

5/10-6 (.42/2/15.6
=

12 1/11.15
-

=) 0X(z1 = 1.5
02/2/17.65
=>

=) 0.81/2/17.65