Mzuzu University

Department of Mathematics

EMT 4701 - Applied Mathematics

End of Semester Examinations

19th October, 2012 Time allowed: 3 hours

Instructions

1. Attempt all questions.

2. Show all your working.
.............................................................................

1
1. Use the Runge-Kutta method with h = 0.1 to approximate the value of
the solution y = φ(t) at t = 0.1 of the differential equation
dy √
= 5t − 3 y, y(0) = 2.
dt
Give your results correct to five decimal places.

14 Marks

2. Find the Fourier Series of the function


0, −L < x < 0,
f (x) = f (x + 2L) = f (x)
L, 0 < x < L;

and use the result to show that

π 1 1 1
= 1 − + − + ···
4 3 5 7

12/4 Marks

3. Consider a vibrating string of length L = 30 that satisfies the wave equa-

tion
4uxx = utt , 0 < x < 30, , t > 0.
Assume that the ends of the string are fixed and that the string is set in
motion with no initial velocity from the initial position

x/10, 0 ≤ x ≤ 10;
u(x, 0) = f (x) =
(30 − x)/20, 10 < x ≤ 30.

Find the displacement u(x, t) of the string.

15 Marks

4. Let c be the n-vector (c1 , · · · , cn )T , b an m-vector (b1 , · · · , bm )T and

A = (aij ) be an m × n matrix. In linear programming, the standard
maximum problem is stated as follows:
Maximize cT x = c1 x1 + · · · + cn xn
Subject to Ax ≤ b
x≥0
(a) Using the notation above, state the dual of the standard maximum
problem.
2 Marks
(b) For a given standard maximum problem, what is a feasible vector x?
1 Mark

2
 (c) Suppose a vector x is feasible for the standard maximum problem
and y is feasible for its dual, show that cT x ≤ yT b.
6 Marks
(d) Consider the linear programming problem
Maximize 3x1 + 2x2 + x3
Subject to x1 − x2 + x3 ≤ 4
2x1 + x2 + 3x3 ≤ 6
−x1 + 2x3 ≤ 3
x1 + x2 + x3 ≤ 8
x1 , x2 , x3 ≥ 0
i. State the dual minimum problem.
ii. Suppose (x1 , x2 , x3 ) = (0, 6, 0) is optimal for the maximum prob-
lem. Use the Equilibrium Theorem to solve the dual problem.
3/6 Marks
(e) The following tableau were obtained while solving LP problems with
two non-negative variables x1 and x2 and two inequality constraints.
The objective function z is maximized. In each case, state with
reasons whether the linear program is unbounded or degenerate.
i.
z x1 x2 x3 x4 RHS
1 2 0 0 1 8
0 3 1 0 −2 4
0 −2 0 1 1 0
ii.
z x1 x2 x3 x4 RHS
1 0 −1 0 2 20
0 0 0 1 −2 5
0 1 −2 0 3 6
2/2 Marks
(f) Use the Simplex method to solve the following LP problem, clearly
specifying the basic and non-basic solutions at each stage.
Maximize 4x1 + x2 − x3
Subject to x1 + 3x3 ≤ 6
3x1 + x2 + 3x3 ≤ 9
x1 , x2 , x3 ≥ 0

8 Marks

5. In variational calculus, given the points (x1 , y1 ) and (x2 , y2 ) and a function
F (x, y, y ′ ), we are interested in finding the curve y = y(x) which minimizes
the integral
Zx2
I = F (x, y, y ′ )dx.
x1

3
(a) Let Y (x) = y(x) + ǫη(x) be a set of varied curves between x1 and x2
where ǫ is a parameter and η(x) is an arbitrary function such that
η(x1 ) = η(x2 ) = 0. We want
 
dI(ǫ)
= 0.
dǫ ǫ=0

Show that this holds if

d ∂F ∂F
− =0 (1)
dx ∂y ′ ∂y
12 Marks
(b) Using the Euler equation (Equation (1)), show that the straight line
gives the shortestpdistance between two points. (Hint: Minimize the
Rx
integral I = x12 1 + y ′2 dx, the arc length of y(x) from x1 and x2 ).
5 Marks
(c) In the following integrals, change the independent variable to simplify
the Euler equation and then find the corresponding first integrals.
p
R x3 1 + y ′2
i. x1 dx;
y2
R x2 p
ii. x1 y y ′2 + y 2 dx.

4/4 Marks

4
Solutions
1. We first calculate k1 , k2 , k3 and k4 .

k1 = f (t0 , y0 )
√
=5×0−3 2
= −4.242640687.

2 Marks

k2 = f (t0 + h/2, y0 + k1 h/2)

= f (0.05, 2 + (−4.242640687) × 0.1/2)
= f (0.05, 1.787697966)
√
= 5(0.05) − 3 1.787697966
= −3.7613354

3 Marks

k3 = f (t0 + h/2, y0 + k2 h/2)

= f (0.05, 2 + (−3.7613354) × 0.1/2)
= f (0.05, 1.81193323)
√
= 5(0.05) − 3 1.81193323
= −3.788242077

3 Marks

k4 = f (t0 + h, y0 + hk3 )
= f (0.1, 2 + 0.1 × (−3.788242077))
= f (0.1, 1.621175792)
√
= 5(0.1) − 3 1.621175792
= −3.31976205162

3 Marks

Hence
0.1
φ(0.1) ≈ 2 + [−4.242640687 + 2(−3.7613354) + 2(−3.788242077) − 3.31976205162]
6
= 1.622307372

5
 2/1 Marks

2. We have 
0, −L < x < 0,
f (x) = f (x + 2L) = f (x)
L, 0 < x < L;
Its Fourier Series is
∞ ∞
a0 X nπx X nπx
f (x) = + an cos + bn sin
2 n=1
L n=1
L

where:
ZL
1
a0 = f (x)dx
L
−L
ZL
1
= Ldx
L
0
1
= [Lx]L
0
L
= L,

2 Marks

ZL
1 nπx
an = f (x) cos dx
L L
−L
ZL
1 nπx
= L cos dx
L L
0
ZL
nπx
= cos dx
L
0
 L
L nπx
= sin
nπ L 0
= 0,

3 Marks

6
and
ZL
1 nπx
bn = f (x) sin dx
L L
−L
ZL
1 nπx
= L sin dx
L L
0
ZL
nπx
= sin dx
L
0
 L
L nπx
= − cos
nπ L 0
 
L L
= − cos nπ +
nπ nπ
L
= [1 − cos nπ]
nπ
 2L
nπ n odd
= .
0 n even

5 Marks

Hence
L 2L X 1 nπx
f (x) = + sin
2 π n L
n odd
∞
L 2L X 1 (2m − 1)πx
= + sin
2 π m=1 2m − 1 L

2 Marks

π
Let L = π and x = 2. Then
π  ∞
π X 1 (2m − 1)π
f = +2 sin =π
2 2 m=1
2m − 1 2
∞
X 1 (2m − 1)π π
⇒2 sin =
m=1
2m − 1 2 2
∞
π X 1 (2m − 1)π
⇒ = sin
4 m=1
2m − 1 2
π 1 1 1
i.e., = 1 − + − + · · · as required.
4 3 5 7

7
 4 Marks

3. The solution is
∞
X nπx nπat
u(x, t) = cn sin cos
n=1
30 30
where
30
2 nπx
Z
cn = f (x) sin dx
30 0 30
10 30
1 nπx 1 nπx
Z Z
= x sin dx + (30 − x) sin dx
150 0 30 300 10 30

2 Marks

In the first integral, let u = x and dv = sin nπx

30 dx. Then du = dx and
30
v = − nπ cos nπx
30 .

2 Marks

Hence
10
30 10
 
1 nπx 1 30x nπx nπx
Z Z
x sin dx = − cos + cos dx
150 0 30 150 nπ 30 nπ 0 30
 10
x nπx 6 nπx
= − cos + 2 2 sin dx
5nπ 30 n π 30 0
2 nπ 6 nπ
=− cos + 2 2 sin
nπ 3 n π 3
3 Marks

In the second integral, let u = 30 − x and dv = sin nπx

30 dx. Then du = −dx
30
and v = − nπ cos nπx
30 .

2 Marks

Hence
Z 30
30 30
 
1 nπx 1 −30(30 − x) nπx nπx
Z
(30 − x) sin dx = cos − cos dx
300 10 30 300 nπ 30 nπ 10 30
  30
1 −30(30 − x) nπx 30 30 nπx
= cos − sin
300 nπ 30 nπ nπ 30 10
 30
−(30 − x) nπx 3 nπx
= cos − 2 2 sin
10nπ 30 n π 30 10
2 nπ 3 nπ
= cos + 2 2 sin
nπ 3 n π 3

8
 3 Marks

Adding the two expressions, we obtain

9 nπ
cn = sin .
n2 π 2 3
1 Mark

Hence ∞
X 9 nπ nπx nπt
u(x, t) == sin sin cos
n=1
n2 π 2 3 30 15

2 Marks

4. (a) The standard minimum problem is

Maximize y T b = b1 y1 + · · · + bm ym
Subject to yT A ≥ c
y≥0
2 Marks
(b) A vector x is feasible for a given maximum problem if it satisfies the
corresponding constraints.
1 Mark
(c) Suppose x is feasible for the standard maximum problem. Then

Ax ≤ b. (2)

If y is also feasible for the standard minimum problem then

yT A ≥ cT . (3)

Since y ≥ 0, we multiply Equation (2) by yT to obtain

yT Ax ≤ yT b. (4)

Similarly, multiplying Equation (3) by x we obtain

yT Ax ≥ cT x. (5)

Combining Equations (4) and (5), we obtain

cT x ≤ yT Ax ≤ yT b.

Hence cT x ≤ yT b.
6 Marks

9
(d) The dual problem is
Minimize 4y1 + 6y2 + 3y3 + 8y4
Subject to y1 + 2y2 − y3 + y4 ≥ 3
−y1 + y2 + y4 ≥ 2
y1 + 3y2 + 2y3 + y4 ≥ 1
y1 , y2 , y3 , y4 ≥ 0
3 Marks
(e) If (x1 , x2 , x3 ) = (0, 6, 0) is optimal for the maximum problem then

cT x = 3x1 + 2x2 + x3
= 3(0) + 2(6) + 0
= 12

1 Mark
Evaluating the constraints of the maximum problem at (0, 6, 0), we
obtain

x1 − x2 + x3 = −6 < 4
2x1 + x2 + 3x3 = 6 = 6
−x1 + 2x3 = 0 < 3
x1 + x2 + x3 = 6 < 8.

2 Marks (1/2 Marks each constraint

By the Equilibrium Theorem, we conclude that

(y1 , y2 , y3 , y4 ) = (0, y2 , 0, 0).

Since yT b = cT x if both x and y are optimal for their respective

problems, we must have yT b = 12 ⇒ 6y2 = 12 ⇒ y2 = 2.
3 Marks
(f) i. This is degenerate because the basic variable x3 has the value 0.
ii. This is unbounded because the ratio of RHS to the entering
variable coefficient is either undefined or negative. We can hence
increase x2 without bound.
2/2 Marks
(g) Adding slack variables, we obtain the following equations:

z − 4x1 − x2 + x3 = 0
x1 + 3x3 + x4 = 6
3x1 + x2 + 3x3 + x5 = 9
x1 , x2 , x3 , x4 , x5 ≥ 0

10
 2 Marks - 1/2 marks each constraint
The corresponding simplex tableau is

z x1 x2 x3 x4 x5 RHS
Row 0 1 −4 −1 1 0 0 0 basic x4 = 6, x5 = 9
Row 1 0 1 3 0 1 0 6 non-basic x1 = x2 = x3 = 0
Row 2 0 3 1 3 0 1 9 z=0

1 Marks
Now chose x1 as the pivot column. To chose the pivot row, calculate
the rations RHS/x where x is the “entering variable coefficient”. The
ratios are 6/1 = 6 and 9/3 = 3. Since 3 < 6, we choose Row 2 as the
pivot row and 3 as the pivot element. An iteration of the simplex
method gives:

z x1 x2 x3 x4 x5 RHS
1 4
Row 0 1 0 3 5 0 3 12 basic x1 = 3, x4 = 3
8
Row 1 0 0 3 −1 1 − 13 3 non-basic x2 = x3 = x5 = 0
1 1
Row 2 0 1 3 1 0 3 3 z = 12

Since all coefficients of Row 0 are non-negative, this solution is opti-

mal.
5 Marks
5. (a) We have Z x2
I(ǫ) = F (x, y, y ′ )dx.
x1

Using the chain rule, we obtain

Zx2  
dI(ǫ) ∂F ∂Y ′ ∂F ∂Y
= + dx (6)
dǫ ∂Y ′ ∂ǫ ∂Y ∂ǫ
x1

Since Y (ǫ) = y(x) + ǫη(x), we have

∂Y
= η(x)
∂ǫ
and
∂Y ′
= η ′ (x).
∂ǫ
Substituting in Equation (6), we get
Zx2  
dI(ǫ) ∂F ′ ∂F
= η (x) + η(x) dx (7)
dǫ ∂Y ′ ∂Y
x1

11
 and
  Zx2  
dI(ǫ) ∂F ′ ∂F
= η (x) + η(x) dx = 0. (8)
dǫ ǫ=0 ∂y ′ ∂y
x1
∂F ′ du d ∂F
In the first integral, let u = ∂y ′ and dv = η (x). Then dx = dx ( ∂y ′ )
and v = η(x). Hence
Z x2 Z x2  
∂F ′ ∂F d ∂F
η (x)dx = η(x) − η(x) dx
x1 ∂y ∂y ′ dx ∂y ′
′
x1
Z x2  
d ∂F
=− η(x) dx
x1 dx ∂y ′
Substituting in Equation (8), we get
  Zx2    
dI(ǫ) d ∂F ∂F
= − + η(x)dx = 0.
dǫ ǫ=0 dx ∂y ′ ∂y
x1

Since η(x) is arbitrary, we must have

 
d ∂F ∂F
− + = 0,
dx ∂y ′ ∂y
i.e.,  
d ∂F ∂F
− = 0,
dx ∂y ′ ∂y
as required.
12 Marks
(b)
p
F (x, y, y ′ ) = 1 + y ′2 dx
∂F 1
⇒ = · 2y ′ · (1 + y ′2 )1/2
∂y ′ 2
y′
= p .
1 + y ′2
∂F
Since = 0, the Euler equation is
∂y
   
d ∂F d y′
= =0
dx ∂y ′ dx 1 + y ′2
y′
⇒ = c1
1 + y ′2
⇒ y ′ = c2 , a constant.
Hence y(x) is a straight line.

12
 5 Marks
(c) i. We have
x2
p Z y2 √
1 + y ′2 1 + x′2
Z
2
dx = dy.
x1 y y1 y2
∂F
Since = 0 and
∂x
1 ′ 1
∂F + x′2 ) 2
2 x (1
=
∂x′ y2
x′
= √ ,
y 2 1 + x′2
the first integral is

x′
√ = const.
y 2 1 + x′2
4 Marks
ii. We have
Z x2 p Z p
y y + y dx = y 1 + x′2 y 2 dy.
′2 2
x1

∂F
Since = 0 and
∂x
∂F 1 1
= y · 2x′ · (1 + x′2 y 2 )− 2
∂x′ 2
yx′
= p ,
1 + x′2 y 2

the first integral of the Euler equation is

yx′
p = const.
1 + x′2 y 2

4 Marks

13
Useful formulae
1. The fourth-order Runge-Kutta method to approximate the solution of the
initial value problem

y ′ = f (t, y), y(t0 ) = y0

is given by
h
yn+1 = yn + (k1 + 2k2 + 2k3 + k4 )
6
where
k1 = f (tn , yn )
1 1
k2 = f (tn + h, yn + hk1 )
2 2
1 1
k3 = f (tn + h, yn + hk2 )
2 2
k4 = f (tn + h, yn + hk3 )

2. The Fourier Series of a periodic function f (x) with period 2L is given by

∞
a0 X  mπx mπx 
+ am cos + bm sin
2 m=1 L L
where
L
1 mπx
Z
am = f (x) cos dx, m = 0, 1, 2, · · ·
L −L L
Z L
1 mπx
bm = f (x) sin dx, m = 1, 2, · · ·
L −L L

3. The displacement u(x, t) of a string of length l disturbed from equilibrium

and released with zero velocity at time t = 0 to vibrate freely satisfies the
wave equation

a2 uxx = utt , 0 < x < l, t > 0;

the boundary conditions

u(0, t) = 0, u(l, t) = 0, t ≥ 0;

and the initial conditions

u(x, 0) = f (x), ut (x, 0), 0 ≤ x ≤ l;

14
where f is a given function. The displacement u(x, t) is given by
∞
X nπx nπat
u(x, t) = cn sin cos
n=1
L L

2 RL nπx
where cn = f (x) sin dx
L 0 L

15