NATIONAL UNIVERSITY OF SINGAPORE
Department of Mathematics
2025-2026 (Semester 1) MA1513 Lecture Problem Set 6 (Solutions)
0 1
1. (a) The underlying matrix A = has a characteristic polynomial λ2 + 1,
−1 0
which gives a pair of complex conjugate eigenvalues ±i.
i −1 x 0
Solving the homogeneous system (iI − A)x = 0 ⇒ = ,
1 i y 0
1
we get the eigenvectors t . (We just need to solve any one of the two
i
equations in the system, say ix − y = 0.)
The two real
solutions
of the system willbegiven by
1 0 1 0
XRe = cos t − sin t and XIm = sin t + cos t
0 1 0 1
cos t sin t
Fundamental set of solutions: and .
− sin t cos t
General solution is y1 = c1 cos t + c2 sin t, y2 = −c1 sin t + c2 cos t.
−4 0
(b) The underlying matrix A = has repeated eigenvalue −4. Since
0 −4
2
this is a scalar matrix, every non-zero vector in R is an eigenvector. So
1 0
we may just take the standard basis vectors , as a pair of linearly
0 1
independent eigenvectors. −4t
e 0
Fundamental set of solutions: , .
0 e−4t
General solution is y1 = c1 e−4t , y2 = c2 e−4t .
2 −2
(c) The underlying matrix A = has a characteristic polynomial (λ −
2 2
2)2 + 4 = λ2 − 4λ + 8, which gives a pair of complex conjugate eigenvalues
2 ± 2i.
2i 2 x
Solving the homogeneous system ((2 + 2i)I − A)x = 0 ⇒ =
−2 2i y
0 1
, we get the eigenvectors t . (We just need to solve any one of the
0 −i
two equations in the system, say 2ix + 2y = 0.)
The two realsolutions
of the system
will be given by
2t 1 0 2t 1 0
XRe = e cos 2t − sin 2t and XIm = e sin 2t + cos 2t
0 −1 0
−1
cos 2t 2t sin 2t
Fundamental set of solutions: e and e2t .
sin 2t − cos 2t
General solution is y1 = c1 e2t cos 2t + c2 e2t sin 2t, y2 = c1 e2t sin 2t − c2 e2t cos 2t.
−1 1
(d) The underlying matrix A = has a characteristic polynomial (λ +
−4 3
1)(λ − 3) + 4 = (λ − 1)2 , which gives a repeated eigenvalue 1.
1
2 −1 x 0
Solving the homogeneous system (I − A)x = 0 ⇒ = , we
4 −2 y 0
1/2 1
get the eigenvectors t . Let’s take v = (with t = 2).
1 2
Since there is only one independent eigenvector, we need to find a generalized
eigenvector u:
−2 1 x 1
Solving the non-homogeneous system (A−I)u = v ⇒ =
−4 2 y 2
(k − 1)/2 1
to get the generalized eigenvectors for any k. Let’s take u =
k 3
(with k = 3).
t t 1 t t t+1
Fundamental set of solutions: e v = e and e (tv + u) = e .
2 2t + 3
General solution: y1 = c1 et + c2 et (t + 1), y2 = 2c1 et + c2 et (2t + 3).
3 2
(e) The underlying matrix A = has a characteristic polynomial (λ −
−8 −5
3)(λ + 5) + 16 = (λ + 1)2 , which gives a repeated eigenvalue −1.
−4 −2 x 0
Solving the homogeneous system (−I − A)x = 0 ⇒ =
1 8 4 y 0
−2 −1
to get the eigenvectors t . Let’s take v = (with t = 2).
1 2
Since there is only one independent eigenvector, we need to find a generalized
eigenvector u:
4 2 x
Solving the non-homogeneous system (A + I)u = v ⇒ =
−8 −4 y
−1 −(2k + 1)/4
to get the generalized eigenvectors for any k. Let’s
2 1 k
−4
take u = (with k = 0).
0
1
−t −t −1 −t −t −t − 4
Fundamental set of solutions: e v = e and e (tv+u) = e .
2 2t
General solution: y1 = −c1 e−t − c2 e−t (t + 41 ), y2 = 2c1 e−t + c2 e−t (2t).
−1 1
2. (a) (i) Eigenvalues -2, 2 with eigenvectors and . Since eigenvalues
3 1
are real and opposite sign, so the equilibrium point is a saddle point, and is
unstable.
(ii)
−2t −1 2t 1
(iii) y = c1 e + c2 e
3 1
−3 3
(b) (i) Eigenvalues 3, 9 with eigenvectors and . Since eigenvalues are
1 1
real, distinct and both positive, so the equilibrium point is a node, and is
unstable.
(ii)
2
3t −3 9t 3
(iii) y = c1 e + c2 e
1 1
√
√ 2
(c) (i) Eigenvalues ±3 2i with eigenvectors . Since eigenvalues are purely
±3i
imaginary, so the equilibrium point is a center, and is neutrally stable.
(ii)
√ √ √ √
2 cos 3√ 2t 2 sin 3√ 2t
(iii) y = c1 + c2
−3 sin 3 2t 3 cos 3 2t
3
1 0
(d) (i) Eigenvalues -4, -3 with eigenvectors and . Since eigenvalues are
0 1
real, distinct and both negative, so the equilibrium point is a node, and is
asymptotically stable.
(ii)
−4t 1 −3t 0
(iii) y = c1 e + c2 e
0 1
2
(e) (i) eigenvalues −1 ± 3i with eigenvectors . Since eigenvalues are
−1 ± 3i
complex with negative real parts, so the equilibrium point is a spiral point,
and is asymptotically stable.
(ii)
−t 2 cos 3t −t 2 sin 3t
(iii) y = c1 e + c2 e
− cos 3t − 3 sin 3t − sin 3t + 3 cos 3t
1 −1
(f) (i) Eigenvalues 0, -2 with eigenvectors and . Since there is a zero
1 1
eigenvalue, there are infinitely many equilibrium points lying along a line, and
they are all asymptotically stable as the other eigenvalue is negative.
(ii)
4
1 −2t −1
(iii) y = c1 + c2 e
1 1
3. If a mass-spring system keeps oscillating without coming to a halt, it must be an
undamped system. So the damping constant c = 0 and its oscillation is given by
second order ODE y ′′ + 4y = 0 (since m = 1 and c = 0).
(i) The ODE can be converted into the SDE
′
y1 = y2 ′ 0 1
or y = y.
y2′ = −4y1 −4 0
(ii) Since there is no damping, we know the equilibrium point is a center and is
neutrally stable (see Chapter 4 notes page 21-23).
(iii) To find the solutions of the SDE, we first find the eigenvalues and eigenvec-
0 1
tors of the underlying matrix : eigenvalues ±2i and corresponding
i −4 0
∓2
eigenvectors .
1
If we take the eigenvalue r = 2i, then a i= 0 and b =2. 1
−2 0 −2
The corresponding eigenvector v = . So p = and q = .
1 1 0
This gives
the fundamental
1 set of real 1 solutions:
0 −2 sin 2t
xRe = cos 2t − sin 2t = 2 ;
1 01 cos 2t
0 −2 − 12 cos 2t
xIm = sin 2t + cos 2t = .
1 0 sin 2t
Hence the general solution of the SDE is:
1 1
sin 2t − cos 2t
y = c1 2 + c2 2 .
cos 2t sin 2t
In addition, we are given the initial conditions: y(0) = y1 (0) = 1 and y ′ (0) =
y2 (0) = 1/2. This allows us to solve for c1 = 1/2 and c2 = −2.
The solution of this initial value problem is:
1 21 sin 2t
1
− 2 cos 2t
y= + (−2) .
2 cos 2t sin 2t
(iv) This is the case of critical damping, with the damping constant c satisfying
the condition c2 = 4mk = 16. So we have c = 4 (see Chapter 4 notes page
21-23).
5
4. (a) (i) The direction field is the one on bottom left. The matrix is the identity
matrix. Testing points (a, 0) on the positive y1 -axis yield tangent vectors
(a, 0). So all arrows there to the right. (In fact arrows in first quadrant
point up and right; those in second quadrant point up and left; those in third
quadrant point down and left; and those in fourth quadrant point down and
right.)
(ii) The trajectories are all straight lines starting from the origin and move
away from it. So the equilibrium point is a star node, and it is unstable.
(b) (i) The direction field is the one on top right. The matrix is the negative of
identity matrix. So all arrows just point in the opposite direction as those in
(a).
(ii) The trajectories are all straight lines moving towards the origin. So the
equilibrium point is a star node, and it is asymptotically stable.
(c) (i) The direction field is the one on top left. Testing points (a, 0) on the
positive y1 -axis yield tangent vectors (0, a). So all arrows there point upwards.
(ii) The trajectories are curves starting away from the origin, move towards
it and then turn away from it. So the equilibrium point is a saddle point, and
it is unstable.
(d) (i) The direction field is the one on bottom right. Testing points (a, a) in the
first quadrant yield tangent vectors (2a, 0). So all arrows along the positive
45◦ line point right.
(ii) The trajectories are curves starting away from the origin, move towards
it and then turn away from it. So the equilibrium point is a saddle point, and
it is unstable.
5. (i) The system is given by
y1′ = −0.05y1 − 0.05y1 + 0.10y2 = −0.1y1 − 0.1y2
y2′ = 0.05y1 − 0.1y2 − 0.02y2 + 0.05y3 = 0.05y1 − 0.12y2 + 0.05y3
y3′ = 0.05y1 + 0.02y2 − 0.05y3
−0.1 0.1 0
(ii) The underlying matrix is 0.05 −0.12 0.05 with eigenvalues −0.17, −0.1
0.05 0.02 −0.05
−10 −1 5
and 0, and corresponding eigenvectors 7 , 0 and 5.
3 1 7
So the general solution
is
−10 −1 5
y = c1 e−0.17t 7 + c2 e−0.1t 0 + c3 5.
3 1 7