AM 121: Homework # 1 (solution)

1. 3.1-5

Solution: By the graphical method, every point in the shaded region below stands
for an allocation (x1 , x2 ) that satisfies all the constraints.

50

100

150

200

250

300

350

400

450

100 200 300 400 500 600

Which point in the feasible region is optimal? To find out, we draw lines of constant

10x1 + 20x2 =constant

These are the lines with slope -1/2, hence they are parallel to each other. All points
(x1 , x2 ) on each of these lines give the same objective function value. Draw the line of
slope -1/2 through point A, B, C, D respectively. This gives four parallel lines among
which point B gives the greatest value. In other words, the point B is an optimal
solution; indeed, in this case point B is the unique optimal solution. It is not difficult
to see that B = (3, 9), and the optimal objective function value is

Z = 10x1 + 20x2 = 10 3 + 20 9 = 210

As a check, note A = (0, 7.5) and C = (4.5, 7.5), and they make Z equal 150 and 195
respectively. They are both suboptimal.

2. 3.1-12

Solution: Note that point B : (x1 , x2 ) = (2, 3) is always on the line kx1 + x2 = 2k + 3
where k 0, so the only situation point B could be the optimal solution is drawn as

1
 follows roughly.

50

100

150

200

250

300

350

400

50 100 150 200 250 300 350 400 450 500 550

If point B is actually optimal, there is no better neighboring point, i.e point A =

(1, 3),C = ( 2k+3
k
, 0) are both suboptimal, which means

Z(B) = 2 + 2 3 = 8 > Z(A) = 1 + 2 3 = 7 (always true!)

2k + 3 2k + 3
Z(B) = 8 > Z(C) = +20= k > 1/2.
k k
Therefore, as long as k > 1/2, point B : (x1 , x2 ) = (2, 3) is optimal.

3. 3.3-2

Solution:

(a) TRUE. When we study the intuitive idea behind Simplex Algorithm in the notes,
we know the fact that a vertex is an optimal solution if there is no better neigh-
boring vertex. Here the objective function value of the point (3,3) beats those of
point (0,2) and point (6,3), so (3,3) must be an optimal solution.
(b) TRUE. If there are multiple optimal solutions, then at least two must be adjacent
cornerpoint feasible solution. In our case, either (0,2) or (6,3) must be an optimal
solution, but cant both!!
(c) FALSE. Just check the case that if we want to maximize Z = x1 x2 in the
feasible region, point (0,0) is the optimal solution (all the other points in the
feasible region correspond to negative objective function values)

4. 3.4-13

Solution: Let

2
 x1F = number of tons of cargo type 1 stowed in the front compartment
x1C = number of tons of cargo type 1 stowed in the center compartment
x1B = number of tons of cargo type 1 stowed in the back compartment
x2F = number of tons of cargo type 2 stowed in the front compartment
x2C = number of tons of cargo type 2 stowed in the center compartment
x2B = number of tons of cargo type 2 stowed in the back compartment
x3F = number of tons of cargo type 3 stowed in the front compartment
x3C = number of tons of cargo type 3 stowed in the center compartment
x3B = number of tons of cargo type 3 stowed in the back compartment
x4F = number of tons of cargo type 4 stowed in the front compartment
x4C = number of tons of cargo type 4 stowed in the center compartment
x4B = number of tons of cargo type 4 stowed in the back compartment

Maximize P = 320x1F + 320x1C + 320x1B + 400x2F + 400x2C + 400x2B +

360x3F + 360x3C + 360x3B + 290x4F + 290x4C + 290x4B
subject to

x1F + x2F + x3F + x4F 12

x1C + x2C + x3C + x4C 18
x1B + x2B + x3B + x4B 10
x1F + x1C + x1B 20
x2F + x2C + x2B 16
x3F + x3C + x3B 25
x4F + x4C + x4B 13
500x1F + 700x2F + 600x3F + 400x4F 7000
500x1C + 700x2C + 600x3C + 400x4C 9000
500x1B + 700x2B + 600x3B + 400x4B 5000

1 1 1 1 1 1 1 1
x1F + x2F + x3F + x4F x1C x2C x3C x4C = 0
12 12 12 12 18 18 18 18
1 1 1 1 1 1 1 1
x1F + x2F + x3F + x4F x1B x2B x3B x4B = 0
12 12 12 12 10 10 10 10

x1F 0, x1C 0, x1B 0, x2F 0, x2C 0, x2B 0

x3F 0, x3C 0, x3B 0, x4F 0, x4C 0, x4B 0.

Extra problem

Solution: The problem is equivalent to

3
 Min Z = y1 + 2y2 + 8y3

s.t |3x1 + 4x2 7| y1

|2x1 + 3x2 5| y2
| x1 + 4x2 9| y3

Why this is true? By noticing that |3x1 + 4x2 7| + 2 |2x1 + 3x2 5| + 8 | x1 + 4x2 9|
is the lower bound of y1 + 2y2 + 8y3 , but the minimization of the former is the same as the
minimization of the latter.

Note |3x1 + 4x2 7| y1 is equivalent to

3x1 + 4x2 7 y1
3x1 + 4x2 7 y1

Therefore, the original problem can be stated as the following linear programming:

Minimize y1 + 2y2 + 8y3

s.t 3x1 + 4x2 y1 7

3x1 + 4x2 + y1 7
2x1 + 3x2 y2 5
2x1 + 3x2 + y2 5
x1 + 4x2 y3 9
x1 + 4x2 + y3 9
yi 0, i = 1, 2, 3