DBA5103: Operations Research and Analytics Semester I, 2020/2021, NUS

Assignment 1: Due on Sep.4, 2020

The solution is for your own reference only. Do not CIRCULATE.

1. (10’) Consider the production problem of n products labeled by j = 1, 2, ..., n. The unit selling price
for the j-th product is denoted by pj . All products can be produced from m raw materials. The
production of product j requires aij units of raw material i, i = 1, ..., m. The i-th raw material can
be procured at a unit price ci . The company has a total budget of B that can be used to procure the
raw materials. Naturally, we assume that all the parameters here are nonnegative. The problem is to
find the production plan that maximizes the total profit (revenue from selling the products minus the

m
procurement cost).

er as
(a) (7’) Let xj be the production quantities of product j, j = 1, ..., n and yi be the procurement

co
eH w
quantities of raw material i, i = 1, ..., m.
(i) Formulate a linear programming model for the problem.

o.
(ii) Let
rs e 
a11 a12 . . . a1n

ou urc
 a21 a22 . . . a2n 
A= .
 
.. .. ..
 ..

. . . 
am1 am2 . . . amn
o

denote the bill-of-material matrix and

aC s
vi y re

       
p1 x1 c1 y1
 p2   x2   c2   y2 
p= . , x =  , c =  , y =  .
       
.. .. ..
 ..   .   .   . 
pn xn cm ym
ed d

Express your model in matrix form.

ar stu

(b) (3’) Formulate a model for the problem without introducing the procurement decisions yi .

Solution:
is

(a) With the decisions variables xj and yi , we can write the model as
Th

Pn Pn
max pj xj − i=1 ci yi
Pj=1
n
s.t. aij xj = yi , i = 1, ..., m
Pj=1
m
i=1 i yi ≤ B
c
xj ≥ 0, j = 1, ..., n, yi ≥ 0, i = 1, ..., m.
sh

Using matrix notation, we can equivalently write the above formulation as

max p 0 x − c0 y
s.t. Ax = y
c0 y ≤ B
x ≥ 0, y ≥ 0.

1-1

This study source was downloaded by 100000759652805 from CourseHero.com on 08-20-2021 20:28:42 GMT -05:00

https://www.coursehero.com/file/68950859/Assignment-1-Solutionpdf/
 Assignment 1: Due on Sep.4, 2020 1-2

Note: the constraint y ≥ 0 can be omitted since all the parameters are nonnegative, x ≥ 0 implies that
y ≥ 0 as well.
(b)
Pm Observe that the total procurement cost incurred P from producing one unit of product j is simply:
n Pm
i=1 ci aij and the total cost from the production is: j=1 ( i=1 ci aij ) xj . Therefore, we have
Pn Pm
max (pj − i=1 ci aij )xj
Pj=1
n Pm
s.t. j=1 ( i=1 ci aij ) xj ≤ B
xj ≥ 0, j = 1, ..., n.

Using matrix notation, this becomes quite clear by directly eliminating the variables y in the problem
using the constraints Ax = y:
max (p0 − c0 A)x
s.t. c0 Ax ≤ B
x ≥ 0.

Note: we have omitted the constraint y ≥ 0 in the above formulation. Adding the constraints Ax ≥ 0

m
is correct as well.

er as
co
eH w
2. (10’) Consider each of the following feasible set and answer the corresponding questions.

o.
(a) (4’) Let a be a nonnegative parameter. The set of all (x, y) ∈ <2 satisfying the constraints:

rs e √
a2 x + ay ≤ 1,
ou urc
x ≥ 0,
y ≥ 0.
o

Is it a polyhedron? Draw a sketch of the feasible set for a = 1.

aC s

(b) (4’) The set of all (x, y) ∈ <2 satisfying the constraints:
vi y re

x2 + y 2 ≤ 1,
x ≥ 0,
y ≥ 0.
ed d

Is it a polyhedron? Draw a sketch of the feasible set.

ar stu

(c) (2’) {x ∈ <|x2 − 8x + 15 ≤ 0}. Is it a polyhedron?

Solution:
√
is

(a) Note that a is a parameter and the nonlinear terms a2 , a do not affect the fact that the constraints
are linear in (x, y). Hence it is a polyhedron. For example, when a = 1, it is shown in the figure below.
Th

1
sh

𝑥
1

This study source was downloaded by 100000759652805 from CourseHero.com on 08-20-2021 20:28:42 GMT -05:00

https://www.coursehero.com/file/68950859/Assignment-1-Solutionpdf/
 Assignment 1: Due on Sep.4, 2020 1-3

𝑥
1

(b) The constraint x2 + y 2 ≤ 1 cannot be expressed as linear constraints. This is not a polyhedron.
Its sketch is shown in the figure.
(c) The constraint can be equivalently expressed as x ≥ 3 and x ≤ 5. Hence, it is a polyhedron.

m
3. (5’) Consider the problem

er as
2x1 + 3|x2 − 10|

co
min

eH w
s.t. |x1 + 2| + |x2 | ≤ 5,

o.
and reformulate it as a linear programming problem.

rs e
Solution: Note that |x1 + 2| + |x2 | = max{x1 + 2 + x2 , x1 + 2 − x2 , −x1 − 2 + x2 , −x1 − 2 − x2 }, and
ou urc
we have
min 2x1 + 3x3
s.t. x1 + 2 + x2 ≤ 5,
o

x1 + 2 − x2 ≤ 5,
aC s

− x1 − 2 + x2 ≤ 5,
vi y re

− x1 − 2 − x2 ≤ 5,
x2 − 10 ≤ x3 ,
− x2 + 10 ≤ x3 .
ed d

An alternative (and better) formulation is

ar stu

min 2x1 + 3x3

s.t. y1 + y2 ≤ 5,
x1 + 2 ≤ y1 ,
is

− x1 − 2 ≤ y1 ,
Th

x2 ≤ y2 ,
− x2 ≤ y2 ,
x2 − 10 ≤ x3 ,
− x2 + 10 ≤ x3 .
sh

4. (10’) The Primo Insurance Company is introducing two new product lines: special risk insurance and
mortgages. The expected profit is $5 per unit on special risk insurance and $2 per unit on mortgages.
Management wishes to establish sales quotas for the new product lines to maximize total expected
profit. The work requirements are as follows:

This study source was downloaded by 100000759652805 from CourseHero.com on 08-20-2021 20:28:42 GMT -05:00

https://www.coursehero.com/file/68950859/Assignment-1-Solutionpdf/
 Assignment 1: Due on Sep.4, 2020 1-4

Work-Hours per Unit

Department Special Risk Mortgage Working-Hours Available
Underwriting 3 2 2400
Administration 0 1 800
Claims 2 0 1200

(a) (4’) Formulate a linear programming model for this problem.

(b) (4’) Use the graphical method to solve this model. Numerically verify your solution using the
software you prefer.
(c) (2’) Identify the two equations in the constraints, whose solution gives the optimal solution.
Solution:
(a) Let x1 be the number of units on special risk insurance and x2 be the number of units on mortgages.

max 5x1 + 2x2

m
er as
s.t. 3x1 + 2x2 ≤ 2400
x2 ≤ 800

co
2x1 ≤ 1200

eH w
x1 ≥ 0, x2 ≥ 0.

o.
(b) Optimal solution is (x∗1 , x∗2 ) = (600, 300) and the optimal profit is 3600. The figure below shows
rs e
the feasible region and the contour line of the objective function.
ou urc
o
aC s
vi y re
ed d
ar stu

(c) The relevant two equations are 3x1 + 2x2 = 2400 and 2x1 = 1200, from which one can solve
is

(x∗1 , x∗2 ) = (600, 300).

Th

5. (5’) Find all extreme points in the following polyhedron:

(a) (2’) P = {(x1 , x2 , x3 )|x1 + x2 + x3 ≤ 1, x1 , x2 , x3 ≥ 0}.
sh

(b) (3’) P = {(x1 , x2 , x3 , x4 )|x1 + x2 + 21 x3 ≤ 1, x1 , x2 , x3 , x4 ≥ 0}.

Solution:
(a) (0, 0, 0), (0, 0, 1), (0, 1, 0), (1, 0, 0);
(b) (0, 0, 0, 0), (1, 0, 0, 0), (0, 1, 0, 0), (0, 0, 2, 0).

This study source was downloaded by 100000759652805 from CourseHero.com on 08-20-2021 20:28:42 GMT -05:00

https://www.coursehero.com/file/68950859/Assignment-1-Solutionpdf/
 Assignment 1: Due on Sep.4, 2020 1-5

Note: The problem, say, part (a) is asking for the extreme points of P = {(x1 , x2 , x3 )|x1 + x2 + x3 ≤
1, x1 , x2 , x3 ≥ 0} not the extreme point of Q = {(x1 , x2 , x3 , x4 )|x1 + x2 + x3 + x4 = 1, x1 , x2 , x3 , x4 ≥ 0}
after standardization, which are two different polyhedrons.

6. Investment under Taxation: (10’) An investor has a portfolio of n different stocks. He has bought
si shares of stock i at price pi , i = 1, ..., n. The current price of one share of stock i is qi . The investor
expects that the price of one share of stock i in one year will be ri . If he sells shares, the investor
pays transaction costs at the rate of 1% of the amount transacted. In addition, for each stock, the
investor pays taxes at the rate of 30% on capital gains (there is no tax if there is no capital gain).
For example, suppose that the investor sells 1, 000 shares of a stock at $50 per share. He has bought
these shares at $30 per share. Upon selling, he receives 1, 000 × 50 = $50, 000. However, he owes
0.30 × (50, 000 − 30, 000) = $6, 000 on capital gain taxes and 0.01 × 50, 000 = $500 on transaction
costs. So, by selling 1, 000 shares of this stock he nets 50, 000 − 6, 000 − 500 = $43, 500. On the
other hand, if the current price of the stock is $20 per share instead of $50. Then, there is no
capital gain on this stock and hence no tax to be paid. In this case, by selling 1,000 shares, he nets

m
20, 000 − 0.01 × 20, 000 = $19, 800.

er as
(a) (6’) Formulate the problem of selecting how many shares the investor needs to sell in order to raise

co
an amount of money at least K, net of capital gain taxes and transaction costs, while maximizing

eH w
the expected value of his (remaining) portfolio next year.

o.
(b) (4’) Using the data for the portfolio in investment.csv, solve the problem for K = $9, 000 and

rs e
attach your code.
ou urc
Solution: (a) Let xi be the number of shares of stock i to be sold. The amount of money raised from
selling xi shares of stock i is then qi xi − 0.3 × max{qi − pi , 0}xi − 0.01qi xi ). The LP formulation of the
problem is then
o
Pn
max Pi=1 ri (si − xi )
n
i=1 (qi xi − 0.3 max{qi − pi , 0}xi − 0.01xi qi ) ≥ K
aC s

s.t.
0 ≤ xi ≤ si , ∀i = 1, ..., n
vi y re

(b) Please refer to Assignment 1 Code.ipynb for the code. The maximum expected value of the portfolio
in the next year is $15356.8 and the optimal selling strategy is
ed d

S1 S2 S3 S4 S5
ar stu

Number of shares 1,000 1,000 1,000 118.762 0

is
Th
sh

This study source was downloaded by 100000759652805 from CourseHero.com on 08-20-2021 20:28:42 GMT -05:00

https://www.coursehero.com/file/68950859/Assignment-1-Solutionpdf/
Powered by TCPDF (www.tcpdf.org)