0% found this document useful (0 votes)
32 views

Assignment 4

This document contains the solution to Question 1 of Assignment 4. It involves transforming between an inertial reference frame K and a non-inertial rotating reference frame K'. It derives the transport theorem, which relates derivatives in K and K', and uses it to derive the expression for the fictitious Coriolis force seen in the non-inertial frame. Key steps include expressing vectors in both frames, applying the product rule to time derivatives, and substituting known relations between velocities and accelerations in the two frames.

Uploaded by

Abhinav Pradeep
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
32 views

Assignment 4

This document contains the solution to Question 1 of Assignment 4. It involves transforming between an inertial reference frame K and a non-inertial rotating reference frame K'. It derives the transport theorem, which relates derivatives in K and K', and uses it to derive the expression for the fictitious Coriolis force seen in the non-inertial frame. Key steps include expressing vectors in both frames, applying the product rule to time derivatives, and substituting known relations between velocities and accelerations in the two frames.

Uploaded by

Abhinav Pradeep
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 20

Assignment 4

Abhinav Pradeep
Tutorial group 12
October 20, 2023

1 Question 1.
1.1 a
π  π 
n̂ = cos + φ ĝ + sin + φ k̂
2 2

ê = ĥ

û = cos (φ)ĝ + sin (φ)k̂


In matrix form,

cos π2 + φ 0 sin π2 + φ
      
n̂ ĝ
 ê  =  0 1 0   ĥ
û cos (φ) 0 sin (φ) k̂
To get the inverse transformation,
−1  
cos π2 + φ 0 sin π2 + φ
   
ĝ n̂
ĥ =  0 1 0   ê 
k̂ cos (φ) 0 sin (φ) û
This can geometrically be seen to be:
    
ĝ − sin (φ) 0 cos (φ) n̂
ĥ =  0 1 0   ê 
k̂ cos (φ) 0 sin (φ) û

1.2 b
Problem involves 2 refrence frames: the inertial frame K with basis vectors V = (ĝ, ĥ, k̂), and the non-
inertial frame K ′ with basis vectors V ′ = (n̂, ê, û). K and K ′ have the same origin. For an observer
in K basis vectors V are stationary where basis vectors V ′ change with time. For an observer on K ′
basis vectors V ′ are fixed whereas basis V moves as a function of time. That is:

dVi′
=0
dt K ′

1
As the non-initial frame is rotating about its shared origin with the inertial frame with an angular
velocity of ω, velocity of Vi′ would be tangential velocity.

dVi′
= ω × Vi′
dt K
Consider some vector A. This can be written as:
3
X
A= Ai Vi
i=1

3
X
A= A′i Vi′
i=1

Let A move along with K in addition to its own velocity. That is,
3
X dA′ (t)
dA
= i
Vi′
dt K ′
i=1
dt
However, in the inertial frame K basis vectors V ′ move as functions of time. Hence,
3
X d (A′ (t)V ′ (t))
dA i i
=
dt K i=1
dt
By the product rule,
3
dA′i ′ dV ′
 
dA X
= Vi + A′i i
dt K i=1
dt dt
3
dA′i ′
 
dA X
′ ′
= V + Ai (ω × Vi )
dt K i=1
dt i
3
dA′i ′
 
dA X
′ ′
= V + (ω × Ai Vi )
dt K i=1
dt i

X dA′ 3 3
dA i ′
X
= Vi + (ω × A′i Vi′ )
dt K i=1
dt i=1

dA dA
= +ω×A
dt K dt K ′
This the ’transport theorem’ and relates derivatives in K and K ′ .

To derive the fictitious Coriolis force, consider that for some position vector R,
dR dR
= +ω×R
dt K dt K ′
dR dR
Let vector VK = dt K
and VK ′ = dt K ′
Therefore the above equation can be written as:
VK = VK ′ + ω × R
dVK
To find dt K
once again apply the transport theorem:
dVK dVK
= + ω × VK
dt K dt K ′
RHS of the above expression must be re-written only in terms of derivatives in K ′ . This is in an effort
express newtons first law in the inertial frame in terms of velocities observed in the non-inertial frame,
ie. find newtons frist law for the non-inertial frame.
dVK
First find expression for dt K ′
:
dVK d
= (VK ′ + ω × R)
dt K ′ dt K ′
dVK dVK ′ d
= + (ω × R)
dt K ′ dt K ′ dt K ′
By the product rule:
   
dVK dVK ′ dω dR
= + ×R + ω×
dt K ′ dt K ′ dt K ′ dt K ′
dR dVK ′
Substitute in VK ′ = dt K ′
and let AK ′ = in
dt K ′
 
dVK dω
= AK ′ + × R + (ω × VK ′ )
dt K ′ dt K ′
Next find expression for ω × VK :

ω × VK = ω × (VK ′ + ω × R)

ω × VK = (ω × VK ′ ) + (ω × (ω × R))
Therefore:
 
dVK dω
+ ω × VK = AK ′ + × R + (ω × VK ′ ) + (ω × VK ′ ) + (ω × (ω × R))
dt K ′ dt K ′
 
dVK dω
= AK ′ + × R + 2 (ω × VK ′ ) + (ω × (ω × R))
dt K dt K ′
dVK
Let AK = dt K
 

AK = AK ′ + × R + 2 (ω × VK ′ ) + (ω × (ω × R))
dt K ′
Multiply in m for the mass of the object.
 

mAK = mAK ′ + m × R + 2m (ω × VK ′ ) + m (ω × (ω × R))
dt K ′
Re-arranging in terms of mAK ′
 

mAK ′ = mAK − m × R − 2m (ω × VK ′ ) − m (ω × (ω × R))
dt K ′
This is Newtons first law in the non-inertial frame. Of these terms the Coriolis force is:

Fc = −2m (ω × VK ′ )
In this case it is given that

VK ′ = ẋê + ẏn̂
It is Assumed that ẋ and ẏ are in S.I units of m/s.

To find ω, convert from rad/day to rad/second. That is:



2πrad/day = rad/second
86400
Hence,

ω=k̂
86400
By applying transformation rules from 1a., where φ is the angle P makes with ĝ
    
ĝ − sin (φ) 0 cos (φ) n̂
ĥ =  0 1 0   ê 
k̂ cos (φ) 0 sin (φ) û
That is:

k̂ = cos (φ)n̂ + sin (φ)û

2π 2π 2π
k̂ = cos (φ)n̂ + sin (φ)û
86400 86400 86400
2π 2π
ω= cos (φ)n̂ + sin (φ)û
86400 86400
Therefore,
 
n̂ û ê
2π 2π
ω × VK ′ = det  86400 cos (φ) 86400 sin (φ) 0 
ẏ 0 ẋ
     
2π 2π 2π
ω × VK ′ = sin (φ)ẋ n̂ + − cos (φ)ẋ û + − sin (φ)ẏ ê
86400 86400 86400

Fc = −2m (ω × VK ′ )
      
2π 2π 2π
Fc = −2m sin (φ)ẋ n̂ + − sin (φ)ẏ ê + − cos (φ)ẋ û
86400 86400 86400
     
4mπ 4mπ 4mπ
Fc = − sin (φ)ẋ n̂ + sin (φ)ẏ ê + cos (φ)ẋ û
86400 86400 86400
1.3 c
Define non inertial frame K ′ in similar manner to the above with point P set to be the equilibrium
position of the pendulum.

Consider Newtons first law in the non-inertial frame K ′ :


 

mAK ′ = mAK − m × R − 2m (ω × VK ′ ) − m (ω × (ω × R))
dt K ′
Below is a visualisation of the forces acting on a pendulum:

Newtons first law in the inertial frame K for the pendulum can be written as

mAK = mg + T
Hence,
 

mAK ′ = mg + T − m × R − 2m (ω × VK ′ ) − m (ω × (ω × R))
dt K ′

= 0 (zero vector) (as ω is not time dependant). Hence,
dt K ′
mAK ′ = mg + T − 2m (ω × VK ′ ) − m (ω × (ω × R))
To neglect the centrifugal force set m (ω × (ω × R)) = 0 (zero vector). Hence,

mAK ′ = mg + T − 2m (ω × VK ′ )
To find AK ′ , let

AK ′ = ÿn̂ + ẍê + z̈ û
It can be assumed that z ≈ 0. Hence z̈ = 0 and:

AK ′ = ÿn̂ + ẍê
From 1b. the Coriolis force acting on the pendulum is:
     
4mπ 4mπ 4mπ
Fc = − sin (φ)ẋ n̂ + sin (φ)ẏ ê + cos (φ)ẋ û
86400 86400 86400
Where φ is the altitude / angle point P makes with the horizontal plane of the inertial frame K.

The gravity vector can be written as:

g = 0n̂ + 0ê − |g|û


Let,

T = Ty n̂ + Tx ê + Tz û
Consider the below diagram. WOLG the system has been simplified by setting the initial conditions
such that the equilibrium point of the pendulum is the origin of frame K ′ .

From the diagram it is evident that △ABC ∼ △AOD (approximately), △ABE ∼ △AOF (approxi-
mately), and △AOB ∼ △ABG. Using this the below can be said:

|T | l |T ||x|
≈ ⇒ Tx ≈
Tx |x| l
|T | l |T ||y|
≈ ⇒ Ty ≈
Ty |y| l
|T | l |T ||OB|
= ⇒ Tz =
Tz |OB| l
By considering the directions involved Tx and Ty can be written as,

|T |x
Tx ≈ −
l
|T |y
Ty ≈ −
l
|T ||OB|
Tz =
l
To find Tz Let θ be the angle that the rope makes with the û axis. Hence,
l
|OB| =
cos (θ)
By the small angle approximation, cos (θ) ≈ 1. Hence,
l
|OB| ≈
1
Hence, by this approximation

|T |l
Tz ≈ ⇒ Tz ≈ |T |
l
Hence,

|T |y |T |x
T ≈− n̂ − ê + |T |û
l l
Therefore the net equation can be written in column vector form as:
     −|T |y   4mπ 
mÿ 0 l
− 86400
sin (φ)ẋ
mẍ =  0  +  −|T |x  +  + 4mπ sin (φ)ẏ 
l 86400
4mπ
0 −m|g| |T | + 86400 cos (φ)ẋ
That is,

−|T |y 4mπ
mÿ = − sin (φ)ẋ
l 86400
−|T |x 4mπ
mẍ = + sin (φ)ẏ
l 86400
4mπ
0 = −m|g| + |T | + cos (φ)ẋ
86400
To calculate |T |, the third equation can be manipulated such that:
4mπ
|T | = m|g| + cos (φ)ẋ
86400
Hence, the EOM can be written as:
 
−y 4mπ 4mπ
mÿ = m|g| + cos (φ)ẋ − sin (φ)ẋ
l 86400 86400
 
−x 4mπ 4mπ
mẍ = m|g| + cos (φ)ẋ + sin (φ)ẏ
l 86400 86400
Expanding out the brackets,

−m|g|y 4mπ cos (φ) 4mπ


mÿ = − ẋy − sin (φ)ẋ
l 86400 86400
−m|g|x 4mπ cos (φ) 4mπ
mẍ = − ẋx + sin (φ)ẏ
l 86400 86400
π cos (φ) π cos (φ)
To simplify the above system, consider that the product of small numbers 86400
× ẋ × y and 86400
×
ẋ × x will be negligible compared to the other terms. Hence,

−m|g|y 4mπ
mÿ = − sin (φ)ẋ
l 86400
−m|g|x 4mπ
mẍ = + sin (φ)ẏ
l 86400
Dividing out mass,

−|g|y 4π
ÿ = − sin (φ)ẋ
l 86400
−|g|x 4π
ẍ = + sin (φ)ẏ
l 86400
49
Pendulum is at latitude φ = 180π
. Hence,


−|g|y 4π 49
ÿ = − sin ẋ
l 86400 180π
 
−|g|x 4π 49
ẍ = + sin ẏ
l 86400 180π
2π 49
 p
Set a = 86400 sin 180π and Ω = |g|/l. This square root will not be imaginary because |g| will always
be positive and negative length does not make sense. With these substitutions and some rearranging:

ẍ = 2aẏ − Ω2 x

ÿ = −2aẋ − Ω2 y
1.4 d
ẍ = 2aẏ − Ω2 x

ÿ = −2aẋ − Ω2 y
Multiply the second equation by the imaginary unit i

ẍ = 2aẏ − Ω2 x

ÿi = −2aẋi − Ω2 yi
Note that 2aẏ = −2aẏi2

ẍ = −2aẏi2 − Ω2 x

ÿi = −2aẋi − Ω2 yi
Add the equations together

ẍ + ÿi = −2aẏi2 − 2aẋi − Ω2 x − Ω2 yi

ẍ + ÿi = −2ai (ẋ + ẏi) − Ω2 (x + yi)

ζ̈ = −2aiζ̇ − Ω2 ζ

ζ̈ + 2aiζ̇ + Ω2 ζ = 0
By anstaz solutions will look like ζ = eλt . To find values of λ, consider the characteristic polynomial
for the above:

λ2 + 2aiλ + Ω2 = 0
By the quadratic formula
p
−2ai ± (2ai)2 − 4Ω2
λ=
2

−2ai ± −4a2 − 4Ω2
λ=
2
p
−2ai ± −4(a2 + Ω2 )
λ=
2
p
−2ai ± 2 (a2 + Ω2 )i
λ=
2
p
λ = −ai ± (a2 + Ω2 )i
 p 
2 2
λ = −a ± (a + Ω ) i
r !
a2
λ= −a ± Ω +1 i
Ω2
Talk about approx later

λ ≈ (−a ± Ω) i

λ1 = (−a + Ω) i , λ1 = (−a − Ω) i
Therefore, the general solution takes the form:

ζ = C 1 e λ1 t + C 2 e λ2 t
Where C1 , C2 ∈ C

ζ = C1 e(−a+Ω)it + C2 e(−a−Ω)it

ζ = C1 (cos (Ωt − at) + sin (Ωt − at)i) + C2 (cos (−at − Ωt) + sin (−at − Ωt)i)
As cos is an even function and sin is an odd function,

ζ = C1 (cos (Ωt − at) + sin (Ωt − at)i) + C2 (cos (Ωt + at) − sin (Ωt + at)i)
In terms of real and complex components,

ζ = (ℜ(C1 ) + ℑ(C1 )i) (cos (Ωt − at) + sin (Ωt − at)i)+(ℜ(C2 ) + ℑ(C2 )i) (cos (Ωt + at) − sin (Ωt + at)i)

ζ = ℜ(C1 ) cos (Ωt − at) + ℜ(C1 ) sin (Ωt − at)i + ℑ(C1 ) cos (Ωt − at)i − ℑ(C1 ) sin (Ωt − at)+
ℜ(C2 ) cos (Ωt + at) − ℜ(C2 ) sin (Ωt + at)i + ℑ(C2 ) cos (Ωt + at)i + ℑ(C2 ) sin (Ωt + at) (1)

ζ = (ℜ(C1 ) cos (Ωt − at) − ℑ(C1 ) sin (Ωt − at) + ℜ(C2 ) cos (Ωt + at) + ℑ(C2 ) sin (Ωt + at)) +
(ℜ(C1 ) sin (Ωt − at) + ℑ(C1 ) cos (Ωt − at) − ℜ(C2 ) sin (Ωt + at) + ℑ(C2 ) cos (Ωt + at)) i (2)

As,

ζ = x + yi

x = ℜ(C1 ) cos (Ωt − at) − ℑ(C1 ) sin (Ωt − at) + ℜ(C2 ) cos (Ωt + at) + ℑ(C2 ) sin (Ωt + at)

y = ℜ(C1 ) sin (Ωt − at) + ℑ(C1 ) cos (Ωt − at) − ℜ(C2 ) sin (Ωt + at) + ℑ(C2 ) cos (Ωt + at)
1.5 e
x = ℜ(C1 ) cos (Ωt − at) − ℑ(C1 ) sin (Ωt − at) + ℜ(C2 ) cos (Ωt + at) + ℑ(C2 ) sin (Ωt + at)

y = ℜ(C1 ) sin (Ωt − at) + ℑ(C1 ) cos (Ωt − at) − ℜ(C2 ) sin (Ωt + at) + ℑ(C2 ) cos (Ωt + at)

x(0) = 0

0 = ℜ(C1 ) + ℜ(C2 )

ℜ(C1 ) = −ℜ(C2 )

y(0) = 0

0 = ℑ(C1 ) + ℑ(C2 )

−ℑ(C2 ) = ℑ(C1 )

ẋ = −ℜ(C1 )(Ω − a) sin (Ωt − at) − ℑ(C1 )(Ω − a) cos (Ωt − at) −
ℜ(C2 )(Ω + a) sin (Ωt + at) + ℑ(C2 )(Ω + a) cos (Ωt + at) (3)

ẏ = ℜ(C1 )(Ω − a) cos (Ωt − at) − ℑ(C1 )(Ω − a) sin (Ωt − at) −
ℜ(C2 )(Ω + a) cos (Ωt + at) − ℑ(C2 )(Ω + a) sin (Ωt + at) (4)

ẋ(0) = 0

0 = −ℑ(C1 )(Ω − a) + ℑ(C2 )(Ω + a)


As ℑ(C2 ) = −ℑ(C1 )

0 = ℑ(C2 )(Ω − a) + ℑ(C2 )(Ω + a)

0 = ((Ω − a) + (Ω + a)) ℑ(C2 )

0 = (2Ω)ℑ(C2 )
As (2Ω) ̸= 0,

ℑ(C2 ) = 0
Hence,

ℑ(C1 ) = 0

ẏ(0) = v

v = ℜ(C1 )(Ω − a) − ℜ(C2 )(Ω + a)


As ℜ(C1 ) = −ℜ(C2 ),

v = ℜ(C1 )(Ω − a) + ℜ(C1 )(Ω + a)

v = ((Ω − a) + (Ω + a)) ℜ(C1 )

v = 2Ωℜ(C1 )

v
ℜ(C1 ) =
2Ω
−v
ℜ(C2 ) =
2Ω
Hence, the initial conditions are:
v
C1 =
2Ω
−v
C2 =
2Ω
Hence the EOM are:
v v
x= cos (Ωt − at) − cos (Ωt + at)
2Ω 2Ω
v v
y= sin (Ωt − at) + sin (Ωt + at)
2Ω 2Ω
By applying the sum to product trigonometric identities,
v
x=− (sin (Ωt) sin (−at))

v
y= (sin (Ωt) cos (−at))

As sin is odd and cos is even,
v
x= (sin (Ωt) sin (at))

v
y= (sin (Ωt) cos (at))

In vector form, this solution can be seen as:
v  v 
s(t) = (sin (Ωt) cos (at)) n̂ + (sin (Ωt) sin (at)) ê
Ω Ω
v
s(t) = sin (Ωt) (cos (at)n̂ + sin(at)ê)

This solution is that of a pendulum that is nudged from its equillibrium position with some velocity
v in the north direction. The Ωv sin (Ωt) term describes the motion of a pendulum with amplitude Ωv
and frequency Ω. The second is a unit vector that rotates in the n̂-ê plane with frequency a. Hence,
the pendulum not only oscillates but also rotates.

Continued on next page


2 Question 2.
2.1 a
% Define parameters
Omega = sqrt(9.83/67); % Angular frequency
a = 2 * pi * sin(49/180*pi) / 86400; % Latitude-related parameter
v = 1; % Initial velocity
% Define the system of ODEs
System = @(t, Y) [Y(2); 2 * a * Y(4) - Omega^2 * Y(1); Y(4);
-2 * a * Y(2) - Omega^2 * Y(3)];
% Initial conditions
initialConditions = [0; 0; 0; v];
% Time intervals
timeSpan = [0, 1200];
[t1, Y1] = ode45(System , timeSpan, initialConditions);
% Analytical solution
TimeArray = linspace(0, 1200, 1000);
xAnalytic = (v/(2*Omega)) * cos(Omega * TimeArray - a * TimeArray) -
(v/(2*Omega)) * cos(Omega * TimeArray + a * TimeArray);
yAnalytic = (v/(2*Omega)) * sin(Omega * TimeArray - a * TimeArray) +
1(v/(2*Omega)) * sin(Omega * TimeArray + a * TimeArray);
% Plot the trajectory
figure;
plot(Y1(:,1), Y1(:,3), ’b’, xAnalytic, yAnalytic, ’r’);
xlabel(’x’);
ylabel(’y’);
title(’Pendulum Motion (0-1200 seconds)’);
legend(’ODE45’, ’Analytical Solution’);

As can be seen from the figure, the analytic and computational solution agree to a very high degree.
This can be attributed to the relatively short timespan.
2.2 b
% Define parameters
Omega = sqrt(9.83/67); % Angular frequency
a = 2 * pi * sin(49/180*pi) / 86400; % Latitude-related parameter
v = 1; % Initial velocity
% Define the system of ODEs
System = @(t, Y) [Y(2); 2 * a * Y(4) - Omega^2 * Y(1); Y(4);
-2 * a * Y(2) - Omega^2 * Y(3)];
% Initial conditions
initialConditions = [0; 0; 0; v];
% Time intervals
timeSpan = [0, 86400];
[t1, Y1] = ode45(System , timeSpan, initialConditions);
% Analytical solution
TimeArray = linspace(0, 86400, 1000);
xAnalytic = (v/(2*Omega)) * cos(Omega * TimeArray - a * TimeArray) -
(v/(2*Omega)) * cos(Omega * TimeArray + a * TimeArray);
yAnalytic = (v/(2*Omega)) * sin(Omega * TimeArray - a * TimeArray) +
(v/(2*Omega)) * sin(Omega * TimeArray + a * TimeArray);
% Plot the trajectory
figure;
plot(Y1(:,1), Y1(:,3), ’b’, xAnalytic, yAnalytic, ’r’);
xlabel(’x’);
ylabel(’y’);
title(’Pendulum Motion (0-86400 seconds)’);
legend(’ODE45’, ’Analytical Solution’);

As can be seen from the figure, the analytic and computational solution do not agree to a very high
degree. This can be attributed to the relatively long timespan. This would compound the inherent
numerical errors in the ode45 solver.
3 Question 3.
p(t)y ′′ + q(t)y ′ + r(t)y = 0

3.1 a
y(t) = c1 y1 (t) + c2 y2 (t)

3.2 b
y(t0 ) = y0

y(t0 ) = c1 y1 (t0 ) + c2 y2 (t0 )

c1 y1 (t0 ) + c2 y2 (t0 ) = y0

y ′ (t0 ) = y0′

y ′ (t0 ) = c1 y1′ (t0 ) + c2 y2′ (t0 )

c1 y1′ (t0 ) + c2 y2′ (t0 ) = y0′


To solve for c1 and c2 , solve:

c1 y1 (t0 ) + c2 y2 (t0 ) = y0

c1 y1′ (t0 ) + c2 y2′ (t0 ) = y0′


In matrix form,
    
y1 (t0 ) y2 (t0 ) c1 y
′ ′ = 0′
y1 (t0 ) y2 (t0 ) c2 y0
By Cramer’s rule,
 
y0 y2 (t0 )
det ′
y y ′ (t0 )
c1 =  0 2 
y1 (t0 ) y2 (t0 )
det ′
y1 (t0 ) y2′ (t0 )
y0 y2′ (t0 ) − y2 (t0 )y0′
c1 =
y1 (t0 )y2′ (t0 ) − y2 (t0 )y1′ (t0 )
 
y1 (t0 ) y0
det ′
y1 (t0 ) y0′
c2 =  
y1 (t0 ) y2 (t0 )
det ′
y1 (t0 ) y2′ (t0 )
y1 (t0 )y0′ − y0 y1′ (t)
c2 =
y1 (t0 )y2′ (t0 ) − y2 (t0 )y1′ (t0 )
3.3 c
The Wronskian W (y1 , y2 )(t) is defined as
 
y1 (t) y2 (t)
W (y1 , y2 )(t) = det ′
y1 (t) y2′ (t)

W (y1 , y2 )(t) = y1 (t)y2′ (t) − y2 (t)y1′ (t)


For the initial value problem

p(t)y ′′ + q(t)y ′ + r(t)y = 0


With conditions

y(t0 ) = y0

y ′ (t0 ) = y0′
Given two solutions y1 and y2 its solvability can be defined by the existence of c1 and c2 such that
some superposition of solutions y1 and y2 is a solution of the IVP. If such c1 and c2 do not exist, the
system must be unsolvable for the provided initial conditions.

From the above derivations,

y0 y2′ (t0 ) − y2 (t0 )y0′


c1 =
y1 (t0 )y2′ (t0 ) − y2 (t0 )y1′ (t0 )
y1 (t0 )y0′ − y0 y1′ (t)
c2 =
y1 (t0 )y2′ (t0 ) − y2 (t0 )y1′ (t0 )
It is evident that the denominators of c1 and c2 are the Wronskian of solutions y1 and y2 . That is:

y0 y2′ (t0 ) − y2 (t0 )y0′


c1 =
W (y1 , y2 )(t0 )
y1 (t0 )y0′ − y0 y1′ (t)
c2 =
W (y1 , y2 )(t0 )
Hence a necessary condition for c1 and c2 to exist is that W (y1 , y2 )(t0 ) ̸= 0. Hence, given the initial
conditions (t0 , y0 , y0′ ) and solutions y1 and y2 , the IVP is only solvable iff the Wronskian of solutions y1
and y2 are non-zero at t0 . The general set of solutions would require y1 and y2 such that the Wronskian
is non-zero for any set of initial conditions (ti , yi , yi′ ). That is:

W (y1 , y2 )(ti ) ̸= 0 ∀ti ∈ t


3.4 d
2t2 y ′′ + ty ′ − 3y = 0

y1 (t) = t−1
Let y2 (t) = y1 (t)v(t) be a solution of the differential equation.

y2′ = y1 (t)′ v(t) + y1 (t)v(t)′

y2′′ = y1 (t)′′ v(t) + 2y1 (t)′ v(t)′ + y1 (t)v(t)′′


As y2 is a solution,

2t2 y2′′ + ty2′ − 3y2 = 0


Sub y2 (t) = y1 (t)v(t)

2t2 (y1 (t)′′ v(t) + 2y1 (t)′ v(t)′ + y1 (t)v(t)′′ ) + t(y1 (t)′ v(t) + y1 (t)v(t)′ ) − 3y1 (t)v(t) = 0

2t2 y1 (t)′′ v(t) + 4t2 y1 (t)′ v(t)′ + 2t2 y1 (t)v(t)′′ + ty1 (t)′ v(t) + ty1 (t)v(t)′ − 3y1 (t)v(t) = 0

2t2 y1 (t)′′ v(t) + ty1 (t)′ v(t) − 3y1 (t)v(t) + 4t2 y1 (t)′ v(t)′ + 2t2 y1 (t)v(t)′′ + ty1 (t)v(t)′ = 0

(2t2 y1 (t)′′ + ty1 (t)′ − 3y1 (t))v(t) + 4t2 y1 (t)′ v(t)′ + 2t2 y1 (t)v(t)′′ + ty1 (t)v(t)′ = 0

As y1 is a solution, 2t2 y1 (t)′′ + ty1 (t)′ − 3y1 (t) = 0. Hence,

4t2 y1 (t)′ v(t)′ + 2t2 y1 (t)v(t)′′ + ty1 (t)v(t)′ = 0

2t2 y1 (t)v(t)′′ + 4t2 y1 (t)′ + ty1 (t) v(t)′ = 0




Let u(t) = v ′ (t)

2t2 y1 (t)u(t)′ + 4t2 y1 (t)′ + ty1 (t) u(t) = 0




2t2 t−1 u(t)′ + 4t2 (−t−2 ) + tt−1 u(t) = 0




2t2 t−1 u(t)′ + (−4 + 1) u(t) = 0

2tu(t)′ − 3u(t) = 0

2tu(t)′ = 3u(t)

du
2t = 3u
dt
1 1
du = dt
3u 2t
Z Z
1 1
du = dt
3u 2t
ln |u| ln |t|
= +C
3 2
 
ln |t|
ln |u| = 3 +C
2
ln |t|
u = e3( 2
+C )
Let C1 = 3C
3
u = e( 2 ln |t|+C1 )
 3 
ln |t 2 |+C1
u=e
3
u = eC1 eln |t 2 |
Let C2 = eC1
3
u = C2 eln |t 2 |

3
u(t) = C2 t 2
As u(t) = v ′ (t)
dv
= u(t)
dt
dv 3
= C2 t 2
dt
Z
3
v = C2 t 2 dt
 
2 5
v = C2 t 2 + C3
5
Let K1 = 52 C2 and K2 = C2 C3
5
v = K1 t 2 + K2
Hence, as y2 (t) = y1 (t)v(t)
5
y2 (t) = K1 y1 (t)t 2 + K2 y1 (t)

5
y2 (t) = K1 t−1 t 2 + K2 t−1

3
y2 (t) = K1 t 2 + K2 t−1
3.5 e
Consider the Wronskian of y1 and y2 ,

W (y1 , y2 )(t) = y1 (t)y2′ (t) − y2 (t)y1′ (t)

y1′ (t) = −t−2

3K1 1
y2′ (t) = t 2 − K2 t−2
2
  
−1 3K1 1 −2 3

W (y1 , y2 )(t) = t t 2 − K2 t − K1 t 2 + K2 t−1 (−t−2 )
2
  
−1 3K1 1 −2 3

W (y1 , y2 )(t) = t t 2 − K2 t + K1 t 2 + K2 t−1 (t−2 )
2
3K1 1 −1 3
W (y1 , y2 )(t) = t 2 − K2 t−2−1 + K1 t 2 −2 + K2 t−1−2
2
3K1 − 1 1
W (y1 , y2 )(t) = t 2 − K2 t−3 + K1 t− 2 + K2 t−3
2
5K1 − 1
W (y1 , y2 )(t) = t 2
2
∀K1 ̸= 0, W (y1 , y2 )(t) ̸= 0 ∀t. Hence y1 and y2 form a fundamental set of solutions. K1 = 0 would
yield the trivial solution.

You might also like