0% found this document useful (0 votes)
16 views11 pages

Eigenvalues and Eigenspaces in Data Science

The document outlines an assignment on the mathematical foundations of data science, focusing on eigenvalues, eigenspaces, and the PageRank algorithm. It includes matrix manipulations, solving for eigenvalues, and determining the importance of nodes in a network represented by a matrix. The final results rank nodes A, B, C, and D in order of importance based on their calculated eigenvector values.
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
16 views11 pages

Eigenvalues and Eigenspaces in Data Science

The document outlines an assignment on the mathematical foundations of data science, focusing on eigenvalues, eigenspaces, and the PageRank algorithm. It includes matrix manipulations, solving for eigenvalues, and determining the importance of nodes in a network represented by a matrix. The final results rank nodes A, B, C, and D in order of importance based on their calculated eigenvector values.
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd

Mathematical Foundations of Data science

ASSIGNMENT -

1 .
Given A = 0 -2

03 O

201_

To solve /XI-Al
eigenvalues 0
=

I 007 [00 I
xi +

+
=

00 I

( % ) -0 I
xF -

A =
-

I I
= X -

1 02

O x -

3 O

2 O x -
1

1 XI-Al =
0 ,
then

I I
X -

1 O 2
=
0
O X -

3 O

2 0 X-1

-
o(x 1)(0) -
-

(0)(2) + (x -

3)((x -

1)(X D
-
-

4) -

0 =
0
I
(x -
3)((x -

x -

x+ 1 -

a)) =
0

(x 3)(x2- -
2x -

3) =
0
-

X2-2X-3
Solving for 0 we
get
=
,

X2 -

2X -

3 =
0
2
X -
3X + X -

3 = 0

x(x 3) + 1(x 3)
- - =
0

(x 3)(x + 1)= = 0

from equation O , we have

(x -

3)(x 3)(x + 1) - =
0

X = 3 3 .
-1
,

Therefore ,
the
eiganvalves are X =
3 ,
3 - 1

Eigenspace for X =
3

(31-A) X =
0

I 30- I
(31 A)
-
=
·

I 30 %
)
=

202
(31 A)
-

X =
0

i .
.
2

I 2020)
RE, !
O

1200
R
,
-
R1/2

I%
O

we
equation
get the x+ z =
0

x = -

therefore the
ciganspace for X = 3 is

11
A = I

O
-

Eigenspace where X = I

( -

I =
A) X =
0

(E +A) X =
0
I I
(I + A) =
2 0 -

O
O 4
-

2 O 2

The system of equations is

2x -

22 = 0

4y = 0

2n + 22 =
0

i .
e x =
2 0
y
=
.
,

Therefore the
eigenspace for X =
-1 is

(0)
* -

I I
2 .
a) Given B =
2 a -

O 2 b

-- I -

13

()(i)
Given
Figurectors v =

we know that Br , =
xv
-

@
(f)
Br =

I + ) (2
- 2 +
=

1 + 1 + 0

substituting the result in


eqn ① we
get

I ) (: ):
O
=

2 -

a = X -


-

2 =
- X

get
X 2
substituting X C in
equ we
get
=
we =
,

2 -

a =
2

a = 0

similarly Brc = Xvz-

solving get
I [i]
Bvz we = 2 O

O
-

I I
=> 2(1) + o + 2( -

1)
O + 2) 1) -

+ 2b
1 + 1 + 6
(e)
=

subtituting the above result in


egn we
get

I I (i)
=
8

the system of equations are

x X
4 = 4
-

=
-
=

2b -

2= -
X

Since ,
X =
4 ,
2b -

2 = -

4
2b = -2

b =
I
-

therefore the values


of a and b are o and I

respectively
.
2b .
Trace (B) =
b11 + b22 +
bas

=
2 + 2 + 3 = 7

know that
we
eigenvalues of B are

x1 = 2

x z =
4
By properties of rigevalues
Xi + x2 + X3 = trace (B)

2 + 4 + xz = 7

* 3 = 7 -
6

* 1
3
=

therefore ,
the
remaining eigenvalue of B is 43 = .
1

I I
3a .
Given D = I -0

2 23

20 A

we know that PD" = I

Performing row operations to turn P into I


by using
(PII)
augmented matrix

I I I
(P(l) = I -

1 O I 00

2 2 301 O

2 0 + 001

R2 +
Rc =
2 R1 , R3-R3-2R

I I I
I -
I ⑧ I O O

O 4 3 -

2 I O

02 I -

2 O I
R2 -
R2 X /4

( I I
1 1 O

yz9 %
-

3/4
-

O I

02 1 -
201

R1 - R + R2 , R3-R3-2R2

I I
I O 3/4 12 O
O I 314- 240
-
-2 1
-

O O 21
-

R3 +
2Rz
-

I I I
I O 34 "10
3/4 120
-

O I

00 12 / -

RI - R ,
-

3/4R3 , R2-Rc-3/4R3

I :
I ]
10

O I

O O I 21 -

(i)
=
therefore
2 I -

2
3b . Matrin C can be written as PDP

: .
Ch = ( PDP + )n
+
= PDnp

I I
D =
10 O

O O O

O O C

1608 I
Dn =

00 ch

PD"D+
Solving

/100 I
= 312

2 1 -

(2 % ) I
=

21 -

I I
: 1
(-1) + 0 + 01 t(z) + 0 + 0 1(3(2) + o + 0
2( -

1) + o + 3c(2)2() + o + 3 2(32) + + (2)(3c)


o

2) 1)-
+o + 2 2( + 2) +o + ch2(32) + o + 1 2)c
-

=on I
3/2
6ch-2

ch -

1 2ch -

3
7/30/24, 12:03 PM MFDS_A5_Q4 - Jupyter Notebook

MATHS7027 Mathematical Foundations of


Data Science
Assignment 5 - Question 4

0 
Consider the network comprising nodes 𝐴 , 𝐵 , 𝐶 , and 𝐷, and represented by the matrix

 
1
0 0
1 
2

𝑀= .
1 1
0
0 0
2 2

 
1
0
0 0
2
1 1
2 2

4(a)
Enter 𝑀 as an array .

In [10]: #importing libraries


import numpy as np

#defining M matrix
M = [Link]([
[0, 0, 0, 0.5],
[1, 0, 0.5, 0.5],
[0, 0.5, 0, 0],
[0, 0.5, 0.5, 0]
])
print (M)

[[0. 0. 0. 0.5]
[1. 0. 0.5 0.5]
[0. 0.5 0. 0. ]
[0. 0.5 0.5 0. ]]

4(b)
 𝑥𝐴 
 

Let 𝒙 = 
𝑥𝐵 
 . Use the PageRank algorithm to solve 𝑀𝒙 = 𝒙.
 𝐶
 𝑥𝐷 
𝑥

Hint: You may find Computer Exercise 5


([Link] helpful.

localhost:8888/notebooks/MFDS_A5_Q4.ipynb 1/2
7/30/24, 12:04 PM MFDS_A5_Q4 - Jupyter Notebook

In [15]: results = [Link](M)



eigenvector = results[1][:,0].real
# Try comparing this eigenvector to the answer you got for x above.
eigenvector

Out[15]: array([0.26832816, 0.71554175, 0.35777088, 0.53665631])

4(c)
Hence, order the nodes 𝐴 , 𝐵 , 𝐶 , and 𝐷 from most important to least important.

In [14]: # Define the corresponding nodes


nodes = ['A', 'B', 'C', 'D']

# Create a list of tuples where each tuple is (node, value)
paired_nodes = list(zip(nodes, eigenvector))

# Sorting the list of tuples based on the value in descending order
sorted_nodes = sorted(paired_nodes, key=lambda x: x[1], reverse=True)

# Extracting the sorted nodes
sorted_nodes = [node for node, value in sorted_nodes]

print("The order of the nodes A, B, C and D from most important to least im

The order of the nodes A, B, C and D from most important to least importan
t is ['B', 'D', 'C', 'A']

localhost:8888/notebooks/MFDS_A5_Q4.ipynb 2/2

You might also like