ISE 330

Introduction to Operations Research: What is Linear Programming?

Deterministic Models
‡ Linear Programming provides methods for allocating limited
resources among competing activities in an optimal way.
www-scf.usc.edu/~ise330/2007
„ Linear → All mathematical functions are linear
„ Programming → Involves the planning of activities
August 29, 2007 ‡ Any problem whose model fits the format for the linear
Lecture 2 programming model is a linear programming problem.

Linear Programming
(Sec. 3.1 – 3.3)

Example 3.1 – Wyndor Glass Co. Example 3.1 – Wyndor Glass Co.
‡ The company produces glass products and owns 3 plants. ‡ Production time available each week
Management decides to produce two new products. „ Plant 1: 4 hours
‡ Product 1 „ Plant 2: 12 hours
„ 1 hour production time in Plant 1 „ Plant 3: 18 hours
„ 3 hours production time in Plant 3
„ $3,000 profit per batch
‡ Product 2
„ 2 hours production time in Plant 2
„ 2 hours production time in Plant 3
„ $5,000 profit per batch

1
Example 3.1 – Wyndor Glass Co. Example 3.1 – Wyndor Glass Co.
Production Time per Batch,
Hours Maximize Z = 3x1 + 5x2
Production Time
Product Available per Week,
Plant 1 2 Hours Subject to:
1 1 0 4 x1 ≤ 4
2x2 ≤ 12
2 0 2 12 3x1 + 2x2 ≤ 18
3 3 2 18
Profit per batch $3,000 $5,000 and
x1 ≥ 0, x2 ≥ 0
Maximize
3x12x+x2Z12x≤=2412
3x
≤ 18
1 + 5x2

Example 3.1 – Wyndor Glass Co. General Linear Programming Problems

x2 x1 = 4 Optimal: ‡ Allocating resources to activities
Maximize Z = 3x1 + 5x2 x1 = 2
x2 = 6 Example General
Subject to: Production capacities of plants Resources
x1 ≤ 4 2x2 = 12 3 plants m resources
2x2 ≤ 12 Production of products Activities
3x1 + 2x2 ≤ 18 Z = 36 2 Products n activities
Production rate of product j, xj Level of activity j, xj
and
Profit Z Overall measure of performance Z
x1 ≥ 0, x2 ≥ 0
3x1 + 2x2 = 18
Z = 10 x1
Z=0

2
General Linear Programming Problems General Linear Programming Problems
Objective Function
Z = c1x1 + c2x2 + … + cnxn Z = Value of overall measure of performance Resource Usage per Unit of Activity
xj = Level of activity j Amount of
= Decision variables
Activity Resource
Constraints
cj = Increase in Z resulting from increase in j Resource 1 2 ... n Available
a11x1 + a12x2 + … + a1nxn ≤ b1 = Parameters
bi = Amount of available resources
1 a11 a12 ... a1n b1
a21x1 + a22x2 + … + a2nxn ≤ b2
... = Parameters 2 a21 a22 ... a2n b2
aij = Amount of resource i consumed by each unit of j
am1x1 + am2x2+ … + amnxn≤ bm = Parameters
... ... ... ... ... ...
Functional m am1 am2 ... amn bm
and Constraints Contribution to Z per c1 c2 ... cn
x1 ≥ 0, x2 ≥ 0, ..., xn ≥ 0 Non-negativity unity of activity
Constraints

Other Forms of Linear Programming Problems Linear Programming Solutions

‡ Minimize rather than maximize objective function ‡ Solution – Any specification of values for the decision
„ Minimize Z = c1x1 + c2x2 + ... + cnxn variables (xj)
‡ Some function constraints with greater-than-or-equal-to (≥) ‡ Feasible solution – A solution for which all constraints are
„ ai1x1 + ai2x2 + ... + ainxn ≥ bi for some value of i satisfied
‡ Some functional constraints in equation form ‡ Infeasible solution – A solution for which at least one
„ ai1x1 + ai2x2 + ... + ainxn = bi for some value of i constraint is violated
‡ Deleting non-negativity constraints ‡ Feasible region – The collection of all feasible solutions
„ xj unrestricted in sign for some value of j ‡ Optimal solution – A feasible solution that has the most
favorable value of the objective function

3
Linear Programming Solutions Linear Programming Solutions
x2 x1 = 4 x1 = 9 x2 x1 = 4

2x2 = 12 2x2 = 12

3x1 + 2x2 = 18 3x1 + 2x2 = 18

x1 x1
No Feasible Solution Multiple Optimal Solutions

Linear Programming Solutions Linear Programming Solutions

x2 x1 = 4 x2 x1 = 4

2x2 = 12 2x2 = 12

3x1 + 2x2 = 18 3x1 + 2x2 = 18

x1 x1
No Optimal Solution Corner-point Feasible (CPF) Solution

4
Linear Programming Assumptions In-class Example
‡ Proportionality – The contribution of each activity to Z or a ‡ 3.1-6 – The Whitt Window Company is a company with only three
constraint is proportional to the level of activity xj employees which makes two different kinds of hand-crafted windows: a
wood-framed and an aluminum-framed. They earn $60 profit for each
„ 0 Z = 3x1 – 1 + 5x2
wood-framed window and $30 for each aluminum-framed window. Doug
‡ Additivity – Every function is the sum of the individual makes the wood frames, and can make 6 per day. Linda makes the
contributions of the activities aluminum frames, and can make 4 per day. Bob forms and cuts the glass,
„ 0 Z = 3x1 + 5x2 + x1x2 and can make 48 square feet of glass per day. Each wood-framed
window uses 6 square feet of glass and each aluminum-framed window
‡ Divisibility – Decision variables are allowed to have any value, uses 8 square feet of glass.
including non-integer values
‡ Certainty – The value assigned to each parameter is assumed How many windows of each type should be produced per day to maximize
profit?
to be a known constant

Homework (Due September 5, 2007)

‡ 3.1-1 - 10 points
‡ 3.1-2 - 5 points
‡ 3.1-3 - 5 points
‡ 3.1-4 - 10 points
‡ 3.1-7 - 15 points
‡ 3.1-9 - 15 points
‡ 3.1-13 - 15 points
‡ 3.2-2 - 10 points
‡ 3.2-5 - 5 points
‡ 3.3-2 - 10 points