Department of Mathematics and Computer Science May 8, 2010

University of Southern Denmark, Odense Marco Chiarandini

DM515 – Introduction to linear and integer

programming
Sheet 1, Spring 2010 [pdf format]

Prepare these exercises for the class of April 21.

Exercise 1
A small airline flies between three cities: Copenhagen, Aarhus, and Odense. They offer sev-
eral flights but, for this problem, let us focus on the Friday afternoon flight that departs from
Copenhagen, stops in Odense, and continues to Aarhus. There are three types of passengers:

• (a) Those traveling from Copenhagen to Odense.

• (b) Those traveling from Odense to Aarhus.

• (c) Those traveling from Copenhagen to Aarhus.

The aircraft is a small commuter plane that seats 30 passengers. The airline offers three fare
classes:

• (a) Y class: full coach.

• (b) B class: nonrefundable.

• (c) M class: nonrefundable, 3-week advanced purchase.

Ticket prices, which are largely determined by external influences (i.e., train and bus com-
petitors), have been set and advertised as follows:
Copenhagen–Odense Odense–Aarhus Copenhagen–Aarhus
Y 300 160 360
B 220 130 280
M 100 80 140
Based on past experience, demand forecasters at the airline have determined the follow-
ing upper bounds on the number of potential customers in each of the 9 possible origin-
destination/fare-class combinations:
Copenhagen–Odense Odense–Aarhus Copenhagen–Aarhus
Y 4 8 3
B 8 13 10
M 22 20 18
The goal is to decide how many tickets from each of the 9 origin/destination/ fare-class
combinations to sell. The constraints are that the plane cannot be overbooked on either of the
two legs of the flight and that the number of tickets made available cannot exceed the forecasted
maximum demand. The objective is to maximize the revenue.
Formulate this problem as a linear programming problem.

Exercise 2
Suppose that Y is a random variable taking on one of n known values:

a1 , a2 , . . . , a n

1
DM515 – Spring 2010 Assignment Sheet

Suppose we know that Y either has distribution p given by

P (Y = a j ) = p j

or it has distribution q given by

P (Y = a j ) = q j
Of course, the numbers p j , j = 1, 2, . . . , n are nonnegative and sum to one. The same is true
for the q j ’s. Based on a single observation of Y, we wish to guess whether it has distribution
p or distribution q. That is, for each possible outcome a j , we will assert with probability x j
that the distribution is p and with probability 1 − x j that the distribution is q. We wish to
determine the probabilities x j , j = 1, 2, . . . , n, such that the probability of saying the distribution
is p when in fact it is q has probability no larger than β, where β is some small positive value
(such as 0.05). Furthermore, given this constraint, we wish to maximize the probability that we
say the distribution is p when in fact it is p. Formulate this maximization problem as a linear
programming problem.

Exercise 3
Suppose that instead of minimizing the sum S = ∑in=1 | axi + b − yi | in the example in Section 2.4
of B1, we want to find a line which minimizes M = maxin=1 | axi + b − yi |. Show how to formulate
this problem as a linear programming problem.

Exercise 4
Solve the following linear programming problem by hand applying the simplex algorithm:

maximize 3x1 + 2x2

subject to x1 − 2x2 ≤ 1
x1 − x2 ≤ 2
2x1 − x2 ≤ 6
x1 ≤ 5
2x1 + x2 ≤ 16
x1 + x2 ≤ 12
x1 + 2x2 ≤ 21
x2 ≤ 10
x1 , x2 ≥ 0.

Exercise 5
Graph the feasible region for the previous exercise. Indicate on the graph the sequence of basic
solutions produced by the simplex method.

Exercise 6
What argument is used to prove that the simplex algorithm always terminates in a finite number
of iterations if it does not encounter a situation in which one of the basic variables is zero? What
may happen instead if the latter situation arises and which remedies are introduced?

Exercise 7
Solve the following problem, known as the Klee-Minty problem, using the largest coefficient
pivoting rule.

2
DM515 – Spring 2010 Assignment Sheet

maximize 100x1 + 10x2 + x3

subject to x1 ≤ 1
20x1 + x2 ≤ 100
200x1 + 20x2 + x3 ≤ 10000
x1 , x2 ≥ 0
Can you generalize the example to n variables and guess what will be the number of iterations
the simplex will do?
[You may carry out the computations online:
• Simplex Method Tool Quick-loading JavaScript-based web page that solves Simplex prob-
lems (works only with problems in two variables!!)
• Java-based interactive simplex tool hosted by Argonne National Laboratory. For the revised
simplex algorithm.]

Exercise 8
The two following LP problems lead to two particular cases when solved by the simplex algo-
rithm. Identify these cases and characterize them, that is, give indication of which conditions
generate them in general.

maximize 2x1 + x2
subject to x2 ≤ 5
− x1 + x2 ≤ 1
x1 , x2 ≥ 0
maximize x1 + x2
subject to 5x1 + 10x2 ≤ 60
4x1 + 4x2 ≤ 40
x1 , x2 ≥ 0
[Hint: one of the two cases implies an infinite number of solutions. In order to describe them all
you might recall that a convex combination of two points x1 and x2 is given by x = ∑ αi xi with
αi ≥ 0 and ∑i αi = 1.

Exercise 9
Solve the first exercise (Opgave 1) in the DM515 exam of 2008. It is available for download from
the course web-page. The theory to solve question d) will be given in the lecture of April 20.
(English speaking students might ask help to collegues or ask directly to the teacher in his office.
The text of the exam of 2010 will be written in English.)