A second order matrix differential equation

Algebra Level 3

A 2 × 2 2 \times 2 matrix B ( t ) B(t) satisfies the differential equation,

d 2 B d t 2 + A 2 B = 0 \dfrac{d^2 B }{d t^2} + A^2 B = 0

where A = [ 6 1 5 0 ] A = \begin{bmatrix} 6 && 1 \\ -5 && 0 \end{bmatrix}

subject to the initial conditions,

B ( 0 ) = [ 1 1 2 3 ] B(0) = \begin{bmatrix} 1 && 1 \\ 2 && 3 \end{bmatrix}

d B d t ( 0 ) = [ 2 0 0 5 ] \dfrac{dB}{dt}(0) = \begin{bmatrix} 2 && 0 \\ 0 && 5 \end{bmatrix}

Determine B ( t ) B(t) , then find B 11 ( 2 ) B_{11}(2) .


The answer is -1.882924.

This section requires Javascript.
You are seeing this because something didn't load right. We suggest you, (a) try refreshing the page, (b) enabling javascript if it is disabled on your browser and, finally, (c) loading the non-javascript version of this page . We're sorry about the hassle.

3 solutions

Hosam Hajjir
May 16, 2019

Here's an analytic solution of the problem. The matrix B ( t ) B(t) will be determined as a function of t t .

Similar to the scalar case, it can be shown that the general solution of the given differential equation is

B ( t ) = cos ( A t ) C 1 + sin ( A t ) C 2 ( 1 ) B(t) = \cos(A t) C_1 + \sin(A t) C_2 \hspace{12pt} (1)

where the square matrices C 1 C_1 and C 2 C_2 depend on the initial conditions.

To determine the matrix sinusoid sin ( A t ) \sin(A t) , we use the following procedure which stems from consequences of the Cayley-Hamilton theorem. According to this procedure, we can express a given function of a matrix as the following expansion:

sin ( A t ) = α 0 I + α 1 ( A t ) + α 2 ( A t ) 2 + + α n 1 ( A t ) n 1 ( 2 ) \sin(A t) = \alpha_0 I + \alpha_1 (A t) + \alpha_2 (A t)^2 + \cdots + \alpha_{n-1} (A t)^{n-1} \hspace{12pt} (2)

where n n is the size of the square matrix A A . In our case, n = 2 n = 2 , so

sin ( A t ) = α 0 I + α 1 ( A t ) ( 3 ) \sin(A t) = \alpha_0 I + \alpha_1 (A t) \hspace{12pt} (3)

To determine the constants α 0 \alpha_0 and α 1 \alpha_1 , we use the fact that the eigenvalues of the matrix ( A t ) (A t) satisfy the same equation. So if λ \lambda is an eigenvalue of A A , then

sin ( λ t ) = α 0 + α 1 ( λ t ) ( 4 ) \sin( \lambda t ) = \alpha_0 + \alpha_1 (\lambda t) \hspace{12pt} (4)

Now it is a straight forward excercise to determine the eigenvalues of the given matrix A A . They are λ 1 = 1 \lambda_1 = 1 and λ 2 = 5 \lambda_2 = 5 . Plugging these two values in equation (4), we obtain a linear equation in the two unknowns α 0 \alpha_0 and α 1 \alpha_1 , which upon solving yields,

α 0 = 5 4 sin ( t ) 1 4 sin ( 5 t ) ( 5 a ) \alpha_0 = \frac{5}{4} \sin(t) - \frac{1}{4} \sin(5 t) \hspace{12pt} (5a)

and,

α 1 = sin ( 5 t ) sin ( t ) 4 t ( 5 b ) \alpha_1 = \dfrac{ \sin(5t) - \sin(t) } { 4 t} \hspace{12 pt} (5b)

Plugging these expressions in equation (3), gives us the desired matrix sin ( A t ) \sin(A t) , namely

sin ( A t ) = [ 1 4 sin t + 5 4 sin 5 t 1 4 sin t + 1 4 sin 5 t 5 4 sin t 5 4 sin 5 t 5 4 sin t 1 4 sin 5 t ] ( 6 ) \sin(A t) = \begin{bmatrix} -\frac{1}{4} \sin t + \frac{5}{4} \sin 5 t && - \frac{1}{4} \sin t + \frac{1}{4} \sin 5 t \\ \frac{5}{4} \sin t - \frac{5}{4} \sin 5 t && \frac{5}{4} \sin t - \frac{1}{4} \sin 5 t \end{bmatrix} \hspace{12pt} (6)

And, in exactly the same manner, we determine the matrix cos ( A t ) \cos(A t) to be,

cos ( A t ) = [ 1 4 cos t + 5 4 cos 5 t 1 4 cos t + 1 4 cos 5 t 5 4 cos t 5 4 cos 5 t 5 4 cos t 1 4 cos 5 t ] ( 7 ) \cos(A t) = \begin{bmatrix} -\frac{1}{4} \cos t + \frac{5}{4} \cos 5 t && - \frac{1}{4} \cos t + \frac{1}{4} \cos 5 t \\ \frac{5}{4} \cos t - \frac{5}{4} \cos 5 t && \frac{5}{4} \cos t - \frac{1}{4} \cos 5 t \end{bmatrix} \hspace{12pt} (7)

Now, we need to determine the matrices C 1 C_1 and C 2 C_2 . Plugging t = 0 t = 0 in equation (1), yields,

B ( 0 ) = C 1 ( 8 a ) B(0) = C_1 \hspace{12pt} (8a)

And differentiating equation (1), and plugging in t = 0 t = 0 yields,

d B d t ( 0 ) = A C 2 ( 8 b ) \dfrac{d B}{dt} (0) = A C_2 \hspace{12pt} (8b)

Since A A is invertible, we can determine C 2 C_2 . What remains is to plug in the above-obtained expressions in equation (1), and the result is,

B ( t ) = [ 3 4 cos t + 7 4 cos 5 t 1 2 sin t + 1 2 sin 5 t cos t + 2 cos 5 t 5 4 sin t + 1 4 sin 5 t 15 4 cos t 7 4 cos 5 t + 5 2 sin t 1 2 sin 5 t 5 cos t 2 cos 5 t + 25 4 sin t 1 4 sin 5 t ] ( 9 ) B(t) = \begin{bmatrix} -\frac{3}{4} \cos t + \frac{7}{4} \cos 5 t - \frac{1}{2} \sin t + \frac{1}{2} \sin 5 t && - \cos t + 2 \cos 5 t - \frac{5}{4} \sin t + \frac{1}{4} \sin 5 t \\ \frac{15}{4} \cos t - \frac{7}{4} \cos 5 t + \frac{5}{2} \sin t - \frac{1}{2} \sin 5 t && 5 \cos t - 2 \cos 5 t + \frac{25}{4} \sin t - \frac{1}{4} \sin 5 t \end{bmatrix} \hspace{12pt} (9)

Finally, we can evaluate B 11 ( 2 ) B_{11}(2) ,

B 11 ( 2 ) = 3 4 cos 2 + 7 4 cos 10 1 2 sin 2 + 1 2 sin 10 = 1.882924 B_{11}(2) = -\frac{3}{4} \cos 2 + \frac{7}{4} \cos 10 - \frac{1}{2} \sin 2 + \frac{1}{2} \sin 10 =\boxed{ -1.882924}

Steven Chase
May 16, 2019

Here's a very elementary numerical integration approach, just to illustrate that the process works the same with matrices as with scalars.

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
import math
import numpy as np

##############################################

# Constants

A = np.array([[6.0,1.0],[-5.0,0.0]])
Asq = np.dot(A,A)

# Initialize B and derivatives

B = np.array([[1.0,1.0],[2.0,3.0]])
Bd = np.array([[2.0,0.0],[0.0,5.0]])
Bdd = -np.dot(Asq,B)

# Initialize time

t = 0.0
dt = 10.0**(-6.0)

##############################################

while t <= 2.0:

    B = B + Bd * dt      # Euler integration for B and first derivative
    Bd = Bd + Bdd * dt

    Bdd = -np.dot(Asq,B)

    t = t + dt

##############################################

print B

# Result

#[[-1.88296442 -2.53466354]
 #[ 2.45311577  5.41656049]]

Karan Chatrath
May 16, 2019

Consider the given differential equation:

Let us introduce a change of variables as follows:

X 1 = B X_1 = B X 2 = d B d t X_2 = \frac{dB}{dt}

By doing so, the second order differential equation can be transformed into a system of first order equations. Note that X 1 X_1 and X 2 X_2 each are square matrices of size 2.

d X 1 d t = X 2 \frac{dX_1}{dt} = X_2 d X 2 d t = A 2 X 1 \frac{dX_2}{dt} = -A^2X_1

This can be combined in a matrix form as such:

[ X 1 ˙ X 2 ˙ ] = [ 0 2 × 2 I 2 A 2 0 2 × 2 ] [ X 1 X 2 ] \begin{bmatrix}\dot{X_1}\\\dot{X_2}\\\end{bmatrix} = \begin{bmatrix}0_{2\times2} & I_2\\-A^2 & 0_{2\times2}\end{bmatrix}\begin{bmatrix}{X_1}\\{X_2}\\\end{bmatrix}

We Denote: K = [ 0 2 × 2 I 2 A 2 0 2 × 2 ] K = \begin{bmatrix}0_{2\times2} & I_2\\-A^2 & 0_{2\times2}\end{bmatrix}

and:

X = [ X 1 X 2 ] X = \begin{bmatrix}{X_1}\\{X_2}\\\end{bmatrix}

Here 0 2 × 2 0_{2\times2} is a matrix of 2 rows and 2 columns with all entries as zero. I 2 I_2 is the identity matrix of size 2. The subscripts are dropped in further computations for convenience. The matrix X X has 4 rows and 4 columns.

The system of differential equations becomes:

X ˙ = K X \dot{X} = KX

Where the initial condition is:

X o = [ B ( 0 ) d B d t ( 0 ) ] X_o = \begin{bmatrix}B(0)\\\frac{dB}{dt}(0)\\\end{bmatrix}

The solution to this equation is of the form: X = e K t X o X = e^{Kt}X_o

Here e K t e^{Kt} is the matrix exponential since K is a square matrix of size 4.

Now, it is just a matter of substituting values and obtaining the required answer. I used a computer for the last step since computing a matrix exponential is not so trivial except in some special circumstances.

The required answer is: -1.883

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...