Scribd
Upload
13 views

ODM2022 Tutorial-4

This document contains 4 exercises related to optimal decision making and linear programming: 1) Determine if two given basic solutions are optimal by performing row operations and checking reduced costs. 2) Calculate that it would take 10^15 seconds (over 300,000 years) to solve a linear program with 100 variables and 30 constraints by checking all possible index sets. 3) Show that a given linear program has feasible solutions but no basic feasible solutions, then transform it into standard form. 4) Prove that a basic feasible solution is uniquely optimal if the reduced cost of every nonbasic variable is positive, and vice versa if the solution is non-degenerate.

Uploaded by

ME
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
13 views

ODM2022 Tutorial-4

This document contains 4 exercises related to optimal decision making and linear programming: 1) Determine if two given basic solutions are optimal by performing row operations and checking reduced costs. 2) Calculate that it would take 10^15 seconds (over 300,000 years) to solve a linear program with 100 variables and 30 constraints by checking all possible index sets. 3) Show that a given linear program has feasible solutions but no basic feasible solutions, then transform it into standard form. 4) Prove that a basic feasible solution is uniquely optimal if the reduced cost of every nonbasic variable is positive, and vice versa if the solution is non-degenerate.

Uploaded by

ME
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 2

1

Optimal Decision Making MGT-483


Tutorial 4 — Exercises
RAO

In this tutorial you determine the basic representation of a basic solution, which allows
you to check whether this solution may be optimal. Afterwards, you examine the worst-
case time needed to solve a linear program with the simplex algorithm, which corresponds
to checking all basic solutions. Next, you verify that the existence of basic feasible solutions
depends on the particular representation of a linear program. Finally, you will prove that
the reduced cost of a non-degenerate basic feasible solution can be used to assess its
optimality.

Exercise 1 (Recognizing optimality). Consider the following linear program:

max x1 + 2x2 (1)


s.t. x1 + x3 = 1
x2 + x4 = 1
x1 + x2 + x5 = 2
x1 , x2 , x3 , x4 , x5 ≥ 0.

(a) Derive the basic representation for the index set I = {1, 2, 5} using elementary
row operations (which you might know as Gaussian Elimination). What is the
corresponding basic solution? Determine whether this basic solution is optimal.

(b) Derive the basic representation for the index set I = {1, 2, 4} using elementary row
operations. What is the corresponding basic solution? Determine whether this basic
solution is optimal.

(c) Compute the basic directions for all non-basic variables with negative reduced costs
in part (b). Verify graphically whether or not the current BFS can be improved by
moving along these basic directions.

Exercise 2 (A naive algorithm for solving LPs). Imagine you are supposed to solve a linear
program in standard form with n = 100 variables and m = 30 constraints by checking
the basic solutions that correspond to all possible index sets I ⊆ {1, . . . , n} with |I| = m.
You have a computer that can check 106 index sets per second. In the worst case, how
much time do you need to solve this linear program?
2

Exercise 3 (Every LP has at least one BFS?). Show that the following LP problem

max x1 + 7x2
s.t: − 3x1 ≤ −7
x1 ≥ 0

has feasible solutions but no basic feasible solution. Next, transform the problem into
standard LP form. How does the above statement change? Explain your observations.

Exercise 4 (Conditions for a unique optimum). Let x be a basic feasible solution associated
with some basis matrix B. Prove the following:

(a) If the reduced cost of every nonbasic variable is positive, then x is the unique optimal
solution.

(b) If x is the unique optimal solution and is non-degenerate, then the reduced cost of
every nonbasic variable is positive.

You might also like