You are on page 1of 3

B.Sc.

(H) Physics Semester 5 Computational Physics Lab


September 03-04, 2020 Time Limit : 4 Hrs

Name Roll No.

The following rules apply:

i
Question: 1 2 3 Total

lh
• Each step carries weight.

De
Points: 2 5 3 10
• Step should be commented.
Score:

of
• Answers must be supported.

ty
rsi
Experiment 3
Linear Algebra : Eigenvalues and Eigenvectors

ive
First Principle

Un
The objective is to solve the equation AX = λX with no prior knowledge of either
X or λ or both.
This equation can be written as (A − λI)X = 0 and for X to exist we need
g
r

|A − λI| = 0
s.o

The left hand side of the above equation represents a polynomial in λ,


sic

p(λ) = |A − λI|
hy

known as the Characteristic Polynomial, the roots of which gives the eigenvalues of
ssp

matrix A. Any such polynomial can expanded as,

p(λ) = (−1)n λn + a1 λn−1 + ... + an


with appropriate values of the coefficient. In this experiment, to find the coef-
e

ficients ai we employ the Faddeev le Verrier method ak = k1 tr(Bk ) & Bk =


eg

A(Bk−1 − ak−1 I) for all k > 1 starting with B1 = A. Once the characteristics
oll

polynomial has been determined, plot it to find out approximate values of roots of
the Characteristic equation. Use an appropriate method to find the exact roots.
aC

Using these eigenvalues solve the eigen-equation for the respective eigen-vectors.
als
Kh
TB
SG
B.Sc. (H) Physics Semester 5 Page 2 of 3 September 03-04, 2020

Methodology
Since λ and X are unknown we need to find λ and X knowing or not knowing both.

• STEP-I To find λ, the eigenvalue use the Fadeev Le Verrier method : to find the coefficients
of the characteristic polynomial.
Given A, let
B1 = A; a1 = trB1
B2 = A(B1 − a1 I); a2 = 21 trB2

lh i
B3 = A(B2 − a2 I); a3 = 13 trB3
.

De
.
.
Bk = A(Bk−1 − ak−1 I); ak = k1 trBk

of
• Use the characteristic polynomial p(λ) to find root values. But note that

ty
rsi
p(λ) = |A − λI|

can expanded as,

ive
p(λ) = (−1)n λn + a1 λn−1 + ... + an

Un
with appropriate values of the coefficient.
Also Note : the (−1)n comes from the following understanding
For a quadratic polynomial ax2 + bx + c = 0 , the sum of roots = −b/a and product of roots
= c/a so if x2 + a1 x + a2 = 0 , the sum of roots is = −a1 and product of roots = a2
Similarly, recall that for a cubic polynomial ax3 + bx2 + cx + d = 0 , the sum of roots = −b/a
g

and product of roots = −d/a so if x3 + a1 x2 + a2 x + a3 = 0 , the sum of roots = −a1 and


r

product of roots = (−1)3 a3


s.o

• STEP-II Numerically use Newton-Raphson method ( simple and good convergence)


sic

(i) Plot p(λ) and estimate value of approximate root λk (ii) Use
hy

p(λk )
λk+1 = λk −
p0 (λk )
ssp

(iii) Use the convergence limit say |p(λ∗k )| < 10−8

• STEP-III For each eigenvalue since we know λ, AX = λX becomes a set of linear equations
So
(A − λI)X = 0
e

Perform Gauss elimination or Gauss Jordan and get


eg

  
a11 a12 a13 x
oll

 0 a22 a23   y = 0
0 0 a33 z
aC
als

• STEP-IV Perform Back substitution


V (3) = a33 a33 ∗ z = 0
Kh

V (2) = − aa23
22
∗ a33 a22 ∗ y + a23 ∗ z = 0
h i
V (1) = 1
a11
[−a12 V (2) − a13 V (3)] = 1
a11
−a12 (− aa23
22
∗ a 33 ) − a 3 a 33 ; a11 ∗x+a12 ∗y +a13 ∗z = 0
TB
SG
B.Sc. (H) Physics Semester 5 Page 3 of 3 September 03-04, 2020

1. Method
(a) (1/2 point) (Eigenvalue Problem)
Briefly explain the meaning of eigenvalue and eigenvector.
(b) (11/2 points) (Condition for Existence)
Briefly explain the method of finding eigenvalue and eigenvector.

2. Coding
(a) (2 points) (Characteristic Equation)

lh i
(Faddeev le Verrier Method) Find it’s characteristic polynomial using Faddeev le Ver-
rier method.

De
B1 = A , Bk = A(Bk−1 − ak−1 I)
1

of
ak = tr(Bk )
k
Plot the characteristic polynomial.

ty
(b) (3 points) (Eigenvalues & Eigenvectors)

rsi
(Newton Raphson Method) Find all the roots (eigenvalues) of the characteristic equa-
tion using the Newton-Raphson method.

ive
(Gauss Elimination Method) Using Gauss elimination method and backward substi-
tution, find the eigenvector corresponding to each eigenvalue.

Un
3. Applications
(a) (3 points) (Test Matrix)
 
8 −6 2
A = −6 7 −4
g

2 −4 3
r
s.o
sic
hy
ssp
e
eg
oll
aC
als
Kh
TB
SG

You might also like