ODE Homework 3

Stefan van Dijk

October 18, 2017

Exercise 1
Consider the following 5 5 matrix and its characteristic polynomial:

5 1 3 1 3
3 1 4 1 3
det(A I) = ( 2)4 ( + 1)

0
A= 4 4 4 0,
3 1 3 1 3
6 3 8 3 4

a to d
We can clearly see that this matrix has 2 eigenvalues namely 1 = 2 with
algebraic multiplicity m1 = 4 and 2 = 1 with algebraic multiplicity m2 = 1.
For the second eigenvalue with algebraic multiplicity 1 there exists a single
corresponding linearly independent eigenvector.
First for 2 = 1 we find the corresponding eigenspace namely we compute:
E2 = ker(A + 1I) (1)
5
= {x R : (A + 1I)x = 0} (2)
Using matlab we can show that the row-echelon form of the matrix (A + 1I)
is given by:
1 0 0 0 21
0 1 0 0 12

0 0 1 0 0

0 0 0 1 12

0 0 0 0 0
Implying that a solution of the equation (A + 1I)x = 0 is given by the vector

1
1

x= 0

1
2

1
So we have that the eigenspace E2 is given by :


1

1



E2 = span 0
1




2

So we have our first block for our J matrix namely the 11 block containing the
element 1 and we have our first element of the Q matrix, namely the vector x
from above.
Now lets proceed to the eigenvalue 1 = 2 with m1 = 4. We need to compute
Ek1 for k = 1, 2, ... until the dimension of that generalized eigenspace becomes
equal to 4. We will do this using Matlab:
1.
E1 = ker(A 2I)
By matlab we can show that

3 1 3 1 3 1 0 0 0 1
3 1 4 1 3 0 1 0 1 0



rref 0 4 2 4 0 0 0 1 0 0



3 1 3 1 3 0 0 0 0 0
6 3 8 3 6 0 0 0 0 0

Thus
1
1


1 1

E1 = span 0 , 0

1
1




1 1

So we have dim E1 = 2
2.
E21 = ker(A 2I)2
By matlab we can show:

9 2 8 2 9 1 0 1 0 1
9 2 8 2 9 10 12 1 0


rref 0 0 0 0 0 0 0 0 0 0


9 2 8 2 9 0 0 0 0 0

18 2 17 2 18 0 0 0 0 0

2
 This implies that:


1 0 1
1 2 2



E21 0 2 2
= span , ,


1 1 1


1 2 1

So we have dim E21 = 3.

3.
E31 = ker(A 2I)3

By matlab we can show:

27 0 27 0 27 1 0 1 0 1
27 0 27 0 27
0 0 0 0 0


rref 0 0 0 0 0 0 0 0 0 0



27 0 27 0 27 0
0 0 0 0
54 0 54 0 54 0 0 0 0 0

As we can see we have 4 free variables so dim E31 = 4 = m1

We get that then:


0 1 0 0

1 0 1 0




E31 = span 0 , 0 , 0 , 1



2 0 1 0

0 1 0 1

We now then have:

r1 = dim E1 = 2 (3)
r2 = dim E21 dim E1 = 1 (4)
r3 = dim E31 dim E21 =1 (5)

*
So the dot diagram becomes:

(6)
(7)
(8)

So we get 2 cycles of lengths 3 and 1

We have from here that our J matrix is as follows:

3

2 1 0 0 0
0 2 1 0 0

0
J = 0 2 0 0
0 0 0 2 0
0 0 0 0 1
We want a 3-cycle {(A 2I)2 v, (A 2I)1 v, v} with v E31 \ E1 . Choosing

v = (0, 0, 1, 0, 1)T

yields the following 3-cycle:

1 0 0

1 1 0




C = 0 , 2 , 1

1


0 0


1 2 1

We also want a 1-cycle {v} with v E1 The choice here is:

1
1

0

1
1

Finally we want a 1-cycle {v} with v E2

The choice here is
1
1

0

1
2

1 0 0 1 1
1 1 0 1 1

Q= 0 2 1 0 0

1 0 0 1 1
1 2 1 1 2
A quick computation with Matlab shows that indeed:

QJQ1 = A

4
e
Finally we want to compute eJt . Notice that:

2 1 0 0 0 2 0 0 0 0 0 1 0 0 0
0 2 1 0 0 0 2 0 0 0 0 0 1 0 0

J = 0 0 2 0 0 = 0 0 2 0 0 + 0 0 0 0 0=D+N

0 0 0 2 0 0 0 0 2 0 0 0 0 0 0
0 0 0 0 1 0 0 0 0 1 0 0 0 0 0
So we get:
eJt = eDt eN t
2t
e 0 0 0 0
0 e2t 0 0 0
Dt

e 0
= 0 e2t 0 0

0 0 0 e2t 0
0 0 0 0 et
1 1 1
eN t = I + N t + (N t)2 + (N t)3 + (N t)4
2 3! 4!
So we get:
1 t 21 t2 0 0
0 1 t 0 0

Nt
e = 0 0 1 0 0

0 0 0 1 0

0 0 0 0 1
So we get that:

1 2 2t
e2t te2t 2t e 0 0
0 e2t 2t
te 0 0


Jt
e =0 0 e2t 0 0


0 0 0 e2t 0


0 0 0 0 et

Exercise 2
Consider the following non-linear IVP:
d2
u + 4u 4u3 = 0, u(0) = , u0 (0) =
dt2

part a)
Substitute x(t) = u(t) and y(t) = u0 (t) = y 0 (t) = u(t) in the equation gives:
y 0 (t) + 4x(t) 4(x(t))3 = 0 = y 0 (t) = 4x3 (t) 4x(t)
Which is a first-order differential equation.

5
part b)
Consider the chain rule:
dy dy dx dy
= = y
dt dx dt dx
However we also have that:
dy
= 4x3 4x
dt
Which in turn implies together with the chain rule that:

dy 4x3 4x
=
dx y
Which we then can rewrite as:

y dy = (4x3 4x) dx

So we get that
y dy (4x3 4x) dx = 0
So we have our g and h.

part c)
We need to solve:
y dy = (4x3 4x) dx
Integrating both sides yields:
1 2
y = x4 2x2 + C
2
Which is equivalent to:
1 2
F (x, y) = y x4 + 2x2 = C
2
Where C is the integrating constant. Note that the partial derivatives of F
correspond with g and h from part b.

6
part d,e)
Using wolfram alpha we get the contour plot:

Plotting closer to the origin we get:

Note that we have said that x(t) = u(t) and y(t) = u0 (t). So for t = 0 we get
that x = , y = But then we get x2 + y 2 = constant which means that x and
y lie on a circle Note that if and are relatively small then (x, y) lies on one
of the contour lines(as seen in the above plot) since they are kind of a circle,
which then implies that if it lies on one of those lines as t changes we stay on
those lines which have a periodic nature.