Sei sulla pagina 1di 105

1

Nonlinear Models:
Dynamic, Goal and
Nonlinear Programming
Chapter 13
2
13.1 Introduction to Nonlinear
Programming
Most real-life situations are more
accurately depicted by nonlinear models
than by linear models.
Nonlinear models formulated in this
chapter fall into three broad categories:
Dynamic programming models (DP).
Goal programming models (GP).
General nonlinear programming models
(NLP).
3
13.2 Dynamic Programming
Dynamic programming problems can be thought
of as multistage problems, in which decisions are
made in sequence.
Known applications of dynamic programming in
business:
Resource allocation,
Equipment replacement,
Production and inventory control,
Reliability.
4
13.2 Dynamic Programming
At stage n a system is found to be in a certain
state.
A decision made at stage n takes the system to
stage n+1, and leaves it in a new state for a stage-
state related cost/profit.
The decision makers challenge is to find a set of
optimal decisions for the entire process.
5
THE U.S. DEPARTMENT OF LABOR
6
The U.S. Department of Labor has made up to
$5 million available to the city of Houston for job
creation.
Requests for funding are to be prepared by four
departments.
The Department of Labor would like to allocate
funds to maximize the number of jobs created.
THE U.S. DEPARTMENT OF LABOR
7
Data
Department Cost $ millions Jobs Created
Housinng 4 225
Employment 1 45
2 125
3 190
Highway 50 jobs per $1 million funding
Law Enforcement 2 75
3 155
4 220
Estimated new jobs created for proposal
THE U.S. DEPARTMENT OF LABOR
8
SOLUTION
The U.S. Department of Labor wants to:

Maximize the total number of new jobs.

Limit funding to $5 million.

Fund at most one proposal from each
department
9
Notation
D
j
= the amount allocated to department j, where j is:
1 - housing, 2 - Employment, 3 - Highway, 4 - Law Enforcement.

R(D
j
) = the number of new jobs created by funding department j with
$D
j
million.

The model

Max R
1
(D
1
) + R
2
(D
2
) + R
3
(D
3
) + R
4
(D
4
)
ST
D
1
+

D
2
+ D
3
+ D
4
<= 5
D
1
,

D
2
, D
3
, D
4
>= 0
Non-linear functions
SOLUTION
10
Define F
j
(X
j
) as the maximum number of new
jobs created by departments (stages) j,
j+1,, 4, given that there is $X
j
million in
funding available for departments j (state)
through 4.
The Backward Dynamic Programming
11
Stage 4: The Law Enforcement,(LED)
Start with the last stage j=4 (Law Enforcement
department, LED).
Allocate to this department funds that maximize the
number of new jobs created
Obviously, the optimal solution for the last department
is to use all the amount available at this stage).
The optimal solution for the last stage is called The
boundary condition
The Backward Dynamic Programming
12
Stage 4: The Law Enforcement table
Available Optimal Funding Maximum jobs
Funding For Stage 4 F4(X4)
0 0 0
1 0 0
2 2 75
3 3 155
4 4 220
5 5 220
States
Cost Jobs
Law Enforcement 2 75
3 155
4 220
Recall
SOLUTION
13
Stage 3: The Highway Department, (HD)
At this stage we consider the funding of both the
Highway department and the Law Enforcement
department.
For a given amount of funds available for funding of
these two departments, the decision regarding the
funds allocated to the HD affects the funds available
for the LED).
SOLUTION
14
Available Possible Remaining Max. New Jobs Optimal F3(X3)
Funding Funding Funds for When Allocating
for Stage 3,4 for Stage 3 Stage 4 D3 to Stage 3 Optimal D3
(X3) (D3) (X3-D3) R3(D3)+F4(X3-D3)
0 0 0 0 + 0 F3(0) = 0; D3(0) = 0
1 0 1 0+0
1 0 50+0 = 50 F3(1) = 50; D3 = 1
2 0 2 0+75 = 75
1 1 50+0 = 50
2 0 100+0 F3(3) = 100; D3 = 2
3 0 3 0+155 = 155
1 2 50+75 = 125
2 1 100+0 = 100
3 0 150+0 = 150 F3(3) = 155; D3 = 0
Stage 3: The Highway Department table
SOLUTION
15
Available Possible Remaining Max. New Jobs Optimal F3(X3)
Funding Funding Funds for When Allocating
for Stage 3,4 for Stage 3 Stage 4 D3 to Stage 3 Optimal D3
(X3) (D3) (X3-D3) R3(D3)+F4(X3-D3)
4 0 4 0+220 = 220
1 3 50+155 = 205
2 2 100+75 = 175
3 1 150+ 0 = 150
4 0 200+ 0 = 220 F3(4) = 220; D3 = 0
5 0 5 0+220 = 220
1 4 50+220 = 270
2 3 100+155 = 255
3 2 150+75 = 225
4 1 200+0 = 200
5 0 250+0 = 250 F3(5) = 270; D3 = 1
SOLUTION
Stage 3: The Highway Department table
16
Stage 2: The Employment Department, (ED)
At this stage we consider the funding of both the
Employment Department and the previous
departments (HD and the LED)

For a given state (the amount of funds available
for funding of these three departments), the
decision regarding the funds allocated to the ED
affects the funds available for the HD and LED.
(the state at the stage j=3)
SOLUTION
17
Funding Funding Funds for When Allocating
for Stage 2,3,4 for Stage 2 Stage 3,4 D2 to Stage 2 Optimal D2
(X2) (D2) (X2-D2) R2(D2)+F3(X2-X2)
0 0 0 0 + 0 F2(0) = 0; D2 = 0
1 0 1 0+50 = 50
1 0 45+0 = 45 F2(1) = 50; D2 = 0
2 0 2 0+100 = 100
1 1 45+50 = 95
2 0 125+0 = 125 F2(2) = 125; D2 = 2
3 0 3 0+155 = 155
1 2 45+100 = 145
2 1 125+50 = 175
3 0 190+0 = 190 F2(3) = 190; D2 = 3
4 0 4 0+220 = 220
1 3 45+155 = 195
2 2 125+100 = 225
3 1 190+50 = 240 F2(4) = 240; D2 = 3
5 0 5 0+270 = 270
1 4 45+220 = 265
2 3 125+155 = 280
3 2 190+100 = 290 F2(5) = 290; D2 = 3
SOLUTION
18
Stage 1: The Housing Department

At this stage we consider the funding for the
Housing Department and all previous
departments.

Note that at stage 1 we know there is exactly $5
million to allocate
(X
1
= 5).
SOLUTION
19
Stage 1: The Housing Department submitted
only one proposal of $4 million, thus
(D
1
= 0, 4).

Available Possible Remaining Max. New Jobs Optimal F1(X1)
Funding Funding Funds for When Allocating
for Stage 1,2,3,4 for Stage 1 Stage 2,3,4 D3 to Stage 3 Optimal D1
(X1) (D1) (X1-D1) R1(D1)+F2(X1-D1)
5 0 5 0+290 = 290
4 1 225+50 = 275 F1(5) = 290; D1 = 0
Do not fund Fund the project
SOLUTION
20
Stage State
0 1 2 3 4 5
1 D1 = 0
2 D2 = 0 D2 = 0 D2 = 2 D2 = 3 D2 = 3 D2 = 3
3 D3 = 0 D3 = 1 D3 = 2 D3 = 0 D3 = 0 D3 = 1
4 D4 = 0 D4 = 1 D4 = 2 D4 = 3 D4 = 4 D4 = 5
In summary, the optimal funds allocation
to maximize the number of jobs created is:

Housing = $0
Employment = $3 million
Highways = $2 million
Law Enforcement = $0
Maximum number of jobs created = 290
SOLUTION
21
Notation Description Example
Stage Variable j A decision point The four departments
State Variable Xj The amount of resource Funds left to allocate
left to be allocated to departments
Decision Variable Dj A possible decision Funds that could have
at stage j been give to department j
Stage Return Values Rj(Dj) The stage return for Number of jobs created
making decision Dj at department j if funding Dj
Optimal Value Fj(Xj) The best cumulative Maximum number of new
return for stages j and jobs created for dept j,,4.
remaining stages at state Xj.
Boundary Condition F(XN) A set of optimal values The new jobs created from
for the last stage (N) funding Dept 4 with $X4 million.
Optimal Solution F1(T) The best cumulative The maximum total number
Value return of new jobs given $Tmillion to
allocate to all departments
Components of Dynamic Programming
22
Bellmans principle of optimality.
From a given state at a given stage, the optimal solution for the
remainder of the process is independent of any previous
decisions made to that point.

Dynamic programming is a recursive process.
The following recursive relationship describes the process for
the U.S. Department of Labor problem.
Define F
j
(X
j
) as the maximum number of new jobs created by
departments (stages) j, j+1,, 4, given that there is $X
j
million
in funding available for departments j (state) through 4.
F
j
(X
j
) = Max

{(R
j
(X
j
) + F
j+1
(X
j
- D
j
)}

all feasible X
j
Dynamic Recursive Relationship
23
The form of the recursion relation differs from
problem to problem, but the general idea is the
same:
Do the best you can for the remaining
stages with the remaining resources
available.
Dynamic Recursive Relationship
24
13.3 Computational Properties
of Dynamic Programming
Dynamic programming versus total
enumeration
Dynamic programming approach eliminates
from consideration combinations that have
no chance of being optimal.
Tricks for reducing computations further
We can reduce computational effort by
smart definition of state variable(s);
noticing the feasible range of state variables.
25
Networks can be used to model multistage decision
problems that are solved by dynamic programming
approach.
Each node represents a state variable value at each stage.
Each arc represents a possible decision at the particular
state.
The stage return is the value assigned to each arc.
The objective is to find the longest (shortest) path.
Dynamic programs as networks
26
Network Presentation-
The US Department of Labor
27

Dynamic programming problems are
typically solved using computer programs.

Since, dynamic problems differ from one
another, there is no universal code to solve
them.

Some problems do fit a particular type, and
are presented in the next section.
Computers and dynamic programming
28
13.4 Dynamic Programming - Examples
In our examples we focus on identifying the
following elements:
The stage variable.
The state variable.
The decision variable.
The stage return or cost function(s).
The optimal value function.
The boundary conditions.
The optimal solution value (stopping rule).
The recurrence relation.
29
KNAPSACK PROBLEMS
This problem belongs to the family of resource
allocation problems.
A hiker carries a knapsack with limited space.
Each item (j) has a volume (W
i
)

and a value V
i.
Limited quantity of each item (U
i
) may be available.
There is a minimum requirement (L
i
) for the amount of each
item that is to be taken.
Fill the knapsack with items such that the total value of items
selected is maximized.
30
KNAPSACK PROBLEMS
The Integer Linear Model
. eger int is D
) j item each for ( U D
) j item each for ( L D
C D W ST
D V Maximize
i
i i
i i
i
i i
i i
s
>
s


where D
i
is the units of item j put in the knapsack.
31
The NWC firms representatives need to deliver
boxes of items the company wants to present in a
conference.
The companys small jet will be used to transport
the team. It can carry a maximum of 127 pounds of
cargo for the show.
Quantities of four different items must be carried :
4Advertising buttons 4 Laser printers
4Advertisig brochures 4 Notebook computers
NORTHWESTERN COMPUTER
32
Data
At least one box of brochures, two printers,
and two notebook computers must be
taken.
Availability, weight, and profit per item are:
Weight Profit
Item Available (lbs.) Potential ($)
Boxes of Buttons 4 4 600
Boxes of Brochures 5 9 1,100
Laser Printers 4 21 4,000
Notebook Computers 5 12 1,500
How many of each item should be loaded on the small jet
to maximize the total profit potential of the trip?
NORTHWESTERN COMPUTER
33
Stage variable= the item j under consideration.
j=1 buttons; j=2 brochures; j=3 printers; j=4 computers
State variable X
j
= The amount of weight left for
transporting item j, j+1, ,notebook computers.
Stage return function = the added value obtained by
taking D
j
additional type j items.
Optimal value function F
j
(X
j
) = the maximum total profit
potential of transporting item j through 4 (notebook
computers) if X
j
pounds remain to take these items.
Northwestern Definitions
34
Let j = 4 represent notebook computer.

F
4
(X
4
) = 0 for all values of X
4
= 0,1,2,,11; D
4
= 0

F
4
(X
4
) = 1,500 for all values of X
4
= 12, 13,,23; D
4
= 1

F
4
(X
4
) = 3,000 for all values of X
4
= 24, 25, ,35; D
4
= 2

F
4
(X
4
) = 4,500 for all values of X
4
= 36, 37, ,48; D
4
= 3
Free space = 127 - 4 lbs of buttons box - 7 lbs of brochure box
- 2(21) lbs of printer boxes - 2(12) lbs of computer boxes = 48 lbs.
Northwestern - Boundary Conditions
35
F
j
(X
j
) = Maximum{V
j
D
j
+ F
j+1
(X
j
- W
j
D
j
)

D
j
s the number of remaining items of type j
W
j
D
j
sX
j


Recursive Calculations
See a demonstration of several
recursive calculations next
36
Recursive Calculations
37
RELIABILITY PROBLEMS
Product reliability and safety are significant
concern for consumers and manufacturers.
Products must be designed with an
acceptable probability of operating
satisfactorily.
Dynamic programming can be used to
determine minimum cost product design that
ensures
minimum required level of reliability, or
highest possible degree of reliability.
38
PLAYCO TOYS, INC.
Playtco Toys manufactures the Watkins Glen
remote-control race car.
Manufacturing cost reduction must come from
three electronic components ordered from a
supplier at the following cost per unit:
Power unit $12
Sound unit $ 5
Remote control unit $ 8
$25
To remain competitive
the total cost per unit
of $25 must be reduced
to $18.
39
Other suppliers offer reduced costs for these
components, but the reliability level of their
components has been proven to be lower.






Playco needs to determine which manufacturer
component to use in order to get the highest degree
of reliability without exceeding the $18 limit.
Power Sound Remote
Manufacturer Unit Cost Reliability Unit Cost Reliability Unit Cost Reliability
DSL, Inc. 12 0.998 5 0.995 5 0.998
Karrow Industries 10 0.994 4 0.987 8 0.999
ELO Elaectronic 5 0.975 2 0.980 4 0.995
PLAYCO TOYS, INC.
40
PLAYCO TOYS - SOLUTION
Define:
P
j
(D
j
) = the probability that component j is
operational if it is purchased for $D
j
.
The reliability of the product is the probability
that all three units operate properly (the product
is functional).
Assuming independence between the
components, the reliability is calculated by :
P
1
(D
1
) P
2
(D
2
) P
3
(D
3
)
- -
41
Stage variable j = the electrical component.

State variable X
j
= the amount of money left to
purchase component j, and subsequent components.

Decision variable D
j
= the amount spent for
component j.

Optimal value function F
j
(X
j
) = the highest probability
that all the components j through 3 will function if $X
j
is
spent to purchase these components.
PLAYCO TOYS Definitions
42
Playco must spend at least $4 on remote control (stage
3).

F
3
(X
3
) = 0 for X
3
= 0, 1, 2, 3 D
3
= 0

F
3
(X
3
) = 0.995 for X
3
= 4 D
3
= 4

F
3
(X
3
) = 0.998 for X
3
= 5, 6, 7 D
3
= 5

F
3
(X
3
) = 0.999 for X
3
= 8, 9,, 18 D
3
= 8

Boundary Conditions Stage 3: Remote
43
F
j
(X
j
) = Max {P
j
(D
j
) F
j+1
(X
j
- D
j
)}
Over all values of D
j

The Recursion Stage 2: Sound unit
At least $6 is required to purchase both a
sound unit and a remote control unit. At most
$13 is required to purchase the most reliable
sound and remote control units.
Thus, the feasible range for X
2
= 3, 4, ,13.
-
44
Amount left Reliability of
for allocation to Allocation to Components 2 and 3 Optimal F2(X2)
Stage 2 and 3 (X2) Stage 2 (D2) P(D2)F3(X2 - D2) Optimal D2
0 - 5 0 ********** 0
6 2 .980(.995) 0.975100
7 2 .980(.998) 0.978040
8 2 .980(.998) 0.97804
4 .987(.995) 0.982065
9 2 .980(.998) 0.978040
4 .987(.998) 0.985026
5 .995(.995) 0.990025
10 2 .980(.999) 0.979020
4 .987(.998) 0.985026
5 .995(.998) 0.993010
Sound Remote
Unit Cost Reliability Unit Cost Reliability
5 0.995 5 0.998
4 0.987 8 0.999
2 0.980 4 0.995
Recursive Calculations
Stage 2: Sound unit
45
We only need to calculate F
1
(18)
Reliability of Components Optimal Value
Allocation to 1 through 3 and Decision
Stage 1 (D1) P(D1)F(X1-D1) F1(X1)
5 .975(.994005) 0.969155
10 .994(.982065) 0.976173
12 .998(.975100) 0.973150
Power
Unit Cost Reliability
12 0.998
10 0.994
5 0.975
Recursive Calculations Stage 1:
Power Supply
46
Amount left
for allocation to Allocation to
Stage 2 and 3 (X2) Stage 2 (D2)
0 - 5 0
6 2
7 2
8 2
4
Allocation to
Stage 1 (D1)
5
10
12
Amount left
for allocation
for stage 3 Allocation
4 4
Purchase
power supply from
Karrow Industries
Purchase
sound unit from
Karrow Industries
Purchase
remote unit from
ELO Electronic
PLAYCO TOYS Solution summary
47
PRODUCTION AND INVENTORY
PROBLEMS
During the course of a certain production run
certain parameters may vary (prices of raw
material, storage space available, etc.)

A dynamic programming modeling approach
can be found very useful in these situations.
48
CJM INDUSTRIES
CJM builds classic replicars using original
car bodies from the 1920s, 30s, 40s, and
50s.
CJM is planning a 15 cars production run of
1957 Chrysler convertibles for the months of
May, June,, September.
49
Data
Maximum production level is 3 for July, and 4 for any other
month.
Up to 2 vehicles may be stored and displayed at a cost of
$2,500 per month ($3,000 for May).







Fixed costs (due to utilities, insurance, etc.) are incurred only
when cars are been actually produced.
Month Fixed Production Cars to be
j Costs Sj ($) Costs Pj ($) Delivered Cj
May 2000 21,000 3
June 3000 16,000 2
July 4000 9,000 1
August 3000 13,000 5
September 2000 23,000 4
CJM INDUSTRIES
50
Develop a production schedule that
minimizes total cost.

CJM INDUSTRIES the Objective
51
Stage variable j: The current month.

State variable X
j
: The number of cars
in inventory at the beginning of month j.

Decision variable D
j
: The production
quantity for month j.
CJM - Definitions
52
Production costs PC
j
(D
j
) in month j are proportional to
the number of cars produced
PC
j
(D
j
) = P
j
D
j

Holding (storage) costs HC
j
(D
j
) in month j are paid for
cars that are left unsold at the end of this month.
For month j =1: HC
1
(D
1
) = 3000(X
1
+ D
1
- C
1
)= 3000D
1
9000
For months j = 2, 3, 4, 5: HC
j
(D
j
) =2500(X
j
+D
j
- C
j
)
Stage cost function:
Fixed costs FC
j
(D
j
) are incurred in month j if there is a car
production. That is
FC
j
(D
j
) = S
j
if D
j
> 0
FC
j
(0) = 0 if D
j
= 0
CJM - Definitions
Given: X
1
= 0 and C
1
= 3
53
The optimal value function F
j
(X
j
) in month j is the minimum
total cost incurred from month j through month 5
(September), given X
j
cars are inventoried at the beginning
of month j.

Boundary conditions F
5
(X
5
):
F
5
(0) = 2,000 + 23,000(4) = $94,000; D
5
= 4
F
5
(1) = 2,000 + 23,000(3) = $71,000; D
5
= 3
F
5
(2) = 2,000 + 23,000(2) = $48,000; D
5
= 2

Optimal solution F
1
(0) is the minimum cost from month 1
(May) through month 5 (September) given that there is no
initial inventory.
CJM - Definitions
54
F
j
(X
j
) = Min{FC
j
(D
j
) + PC
j
(D
j
) + HC
j
(D
j
) + F
j+1
(X
j
+D
j
- C
j
)},


D
j
is feasible only if it satisfies the following conditions:
D
j
+ X
j
> C
j
;
D
3
s 3 for July; D
j
s 4 for j = 1, 2, 4, 5;
D
j
+ X
j
- C
j
s 2;
D
j
> 0

over all feasible D
j
.
The Recursion
55
Recursive Calculations Stage 4: August
X
4
Possible D
4
X
4
+ D
4
C
4
FC
4
PC
4
HC
4
F
5
(X
4
+D
4
C
4
) Total Optimal
Production Units Stored Cost Value
0 Infeasible Infeasible Infeasible
1 4 0 3 52 0 94 149.0 F
4
=149
D
4
=4
2 4 1 3 52 2.5 71 128.5 F
4
=128.5
3 0 3 39 0 94 136.0 D
4
=4

56
X
3
Possible D
3
X
3
+ D
3
C
3
FC
3
PC
3
HC
3
F
4
(X
3
+D
3
C
3
) Total Optimal
Production Units Stored Cost Value
0 2 1 4 18 2.5 149 173.5 F
3
=164.5
3 2 4 27 5.0 128.5 164.5 D
3
=3
1 1 1 4 9 2.5 149 164.5 F
3
=155.5
2 2 4 18 5 128.5 155.5 D
3
=2
2 0 1 0 0 2.5 149 151.5 F
3
=146.5
1 2 4 9 5 128.5 146.5 D
3
=1
Recursive Calculations Stage 3: July
57
13.5 Goal Programming
In real life decision situations, virtually all
problems have several objectives.
When objectives are conflicting, the
optimal decision is not obvious.
Goal programming is an approach that
seeks to simultaneously take into account
several objectives.
58
Goals are prioritized in some sense, and their level of
aspiration is stated.

An optimal solution is attained when all the goals are
reached as close as possible to their aspiration level, while
satisfying a set of constraints.

There are two types of goal programming models:
Nonpreemtive goal programming - no goal is pre-determined to
dominate any other goal.
Preemtive goal programming - goals are assigned different priority
levels. Level 1 goal dominates level 2 goal, and so on.
13.5 Goal Programming
59
A company is considering three forms of
advertising.



Goals
Goal 1: Spend no more $25,000 on advertising.
Goal 2: Reach at least 30,000 new potential
customers.
Goal 3: Run at least 10 television spots.
NONPREEMTIVE GOAL PROGRAMMING
An Advertisement Example
Cost per Ad Customers
Television 3000 1000
Radio 800 500
Newspaper 250 200
60
If these were constraints rather than goals we
would have:
3000X
1
+ 800X
2
+ 250X
3
s25,000
1000X
1
+ 500X
2
+ 200X
3
> 30,000
X
1
> 10
No feasible solution exists that satisfies all the
constraints.
When these constraints are simply goals they
are to be reached as close as possible.
An Advertisement Example
61
Detrimental variables
Ui = the amount by which the left hand side
falls short of (under) its right had side
value.
Ei = the amount by which the left side
exceeds its right had side value.
The goal equations
3000X
1
+ 800X
2
+ 250X
3
+ U
1
E
1
= 25,000
1000X
1
+ 500X
2
+ 200X
3
+ U
2
E
2
= 30,000
X
1
+ U
3
E
3
= 10
An Advertisement Example
62
The objective is to minimize the penalty of
not meeting the goals, represented by the
detrimental variables
E1, U2, U3.

An Advertisement Example
s25,000 > 30,000 > 10
63
The penalties are estimated to be as follows:

Each extra dollar spent on advertisement above
$25,000 cost the company $1.

There is a loss of $5 to the company for each
customer not being reached, below the goal of
30,000.
Each television spot below 10 is worth 100 times
each dollar over budget.
An Advertisement Example
64
It is assumed that no advantage is gained by
overachieving a goal.

Minimize 1E
1
+ 5U
2
+ 100U
3

s.t.
3000X
1
+ 800X
2
+ 250X
3
+ U
1
E
1
= 25,000
1000X
1
+ 500X
2
+ 200X
3
+ U
2
E
2
= 30,000
X
1
+ U
3
E
3
= 10
All variables are non-negative.
An Advertisement Example
The goal programming model
65
The NECC is planning next month production
of its two bicycles B2 and S10.
Data
Both models use the same seats and tires.
2000 seats are available; 2400 tires are available.
1000 gear assembly are available (used only in
the S10 model).
Production time per unit: 2 hours for B2; 3 hours
for S10.
Profit: $40 for each B2; 10$ for each S10.
PREEMTIVE GOAL PROGRAMMING -
New England Cycle Company
66
Priority 1: Fulfill a contract for
400 B2 bicycles to be delivered
next month.

NECC Prioritized Goals
Priority 4:
At least 200 tires left
over at the end of the
month.
At least 100 gear
assemblies left over at
the end of the month.
Priority 2: Produce at least
1000 total bicycles during
the month.

Priority 3:
Achieve at least
$100,000 profit for the
month.
Use no more than 1600
labor-hours during the
month.
67
Management wants to determine the
production schedule that best meets its
prioritized schedule.
New England Cycle Company Example
68
NECC - SOLUTION
Decision variables
X
1
= The number of B2s to be produced next month
X
2
= The number of S10s to be produced next month
Functional / nonnegativity constraints
0 X , X
Tires 2400 X 2 X 2
assemblies Gear 1000 X
Seats 2000 X X 2
2 1
2 1
2
2 1
>
s +
s
s +
69
Goal constraints
Priority 1 (goal 1): Production of at least 400 B2s
X
1
+ U
1
- E
1
= 400

Priority 2 (goal 2): Production of at least 1000 total cycles
X
1
+ X
2
+ U
2
- E
2
= 1000

Priority 3 (goal 3) Profit of at least $100,000
.04X
1
+ .10X
2
+ U
3
- E
3
= 100 (in $1000)
Priority 3 (goal 4) Use a maximum of 1600 labor hours
2X
1
+ 3X
2
+ U
4
- E
4
= 1600

Priority 4 (goal 5) At least 200 leftover tires
2X
1
+ 2X
2
+ U
5
- E
5
= 2200
Priority 4 (goal 6) At least 100 leftover gear assembly
X
2
+ U
6
- E
6
= 900
NECC - SOLUTION
70
Priority level objectives

Priority 1: Underachieving a production of 400 B2s:
Minimize U
1


Priority 2: Underachieving a total production of 1000:
Minimize U
2


NECC - SOLUTION
71
Priority level objectives
Priority 3: Underachieving a $100,000 profit
Using more than 1600 labor-hours
Minimize 30U
3
+ E
4



NECC - SOLUTION
Each $1,000 short of the $100,000 goal is considered
30 times as important as utilizing an extra labor-hour.
72
Priority level objectives
Priority 4: Using more than 2200 tires
Using more than 900 gear assemblies
Minimize E
5
+ 2E
6

NECC - SOLUTION
Each leftover gear assembly is deemed
twice as important as leftover tire.
73
Solve the linear goal programming for priority 1 objective, under
the set of regular constraints and goal constraint as shown below
Minimize U
1

ST
NECC - The solution procedure
X
1
+ U
1
- E
1
= 400
Tires 2400 X 2 X 2
Gear 1000 X
Seats 2000 X X 2
2 1
2
2 1
s +
s
s +
X
1
= 400, thus
U
1
= 0, and
priority 1 goal is
fully achieved.
74
Solve the linear goal programming for priority 2 level objective,
under the set of original constraints plus the constraint X
1
> 400
(maintain the level of achievement of the priority 1 goal).
Minimize U
2

ST
NECC - The solution procedure
Every point that
satisfies X
1
+ X
2
>
1000 yields U
2
= 0,
and therefore,
priority 2 goal is fully
achieved.
Tires 2400 X 2 X 2
Gear 1000 X
Seats 2000 X X 2
2 1
2
2 1
s +
s
s +
X
1
> 400
X
1
+ X
2
+ U
2
- E
2
= 1000

75
Solve the linear goal programming for priority 3
level objective, under the set of original constraints
plus the constraint X
1
> 400 (maintain the level of
achievement of the priority 1 goal), plus the
constraint X
1
+ X
2
> 1000 (maintain the level of
achievement of the priority 2 goal).

Every point in the range X
1
= 400 and 600 s X
2
s 800 is
optimal for this model; 30U
3
+ E
4
= 1720 is the level of
achievement for the priority 3 goal.

NECC - The solution procedure
76
Solve the linear goal programming for priority 4
level objective, and notice that after the previous
step the feasible region is reduced to a segment of
a straight line between the points (400,600) and
(400,800).

X1 =400; 600 s X2 s 700 and E5 + 2E6 = 0
NECC - The solution procedure
77
In summary NECC should produce
400 B2 model
Between 600 and 700 S10 model
NECC - Solution Summary
78
13.6 Nonlinear Programming
A nonlinear programming problem (NLP) is one in
which the objective function, F, and or one or more
constraint functions, G
i
, possess nonlinear terms.

There is no a universal algorithm that can find the
optimal solution to every NLP.

One class of NLPs Convex Programming Problems
can be solved by algorithms that are guaranteed to
converge to the optimal solution.
79
The objective is to maximize a concave
function or to minimize a convex function.
The set of constraints form a convex set.
Properties of Convex Programming
Problems
80
A smooth function (no sharp points, no
discontinuities)
One global maximum (minimum).
A line drawn between any two points on the
curve of the function will lie below (above) the
curve or on the curve.
A One Variable Concave (Convex)
Function
X
A Concave function
A convex function
X
81
An illustration of a two variable
convex function
82
If a straight line that joins any two points in
the set lies within the set, then the set is
called a convex set.
Convex set
Non-convex set
Convex Sets
83
In a NLP model all the constraints are of
the less than or equal to form G
i
(X) s B.
If all the functions G
i
are convex, the set
of constraints forms a convex set.
NLP and Convex Sets
84
13.7 Unconstrained Nonlinear
Programming
One-variable unconstrained problems are
demonstrated by the Toshi Camera
problem.

The inverse relationships between
demand for an item and its value (price)
are utilized in this problem.

85
TOSHI CAMERA
Toshi camera of Japan has just
developed a new product, the Zoomcam.

It is believed that demand
for the initial product will
be linearly related to the
price.
Price Estimated
P ($) Demand (X)
100 350,000
150 300,000
200 250,000
250 200,000
300 150,000
350 100,000
86
Unit production cost is estimated to be $50.
What is the production quantity that
maximizes the total profit from the initial
production run?

SOLUTION

Total profit = Revenue - Production cost

F(X) = PX - 50X

TOSHI CAMERA
87
From the Price / Demand table it can be verified
that P = 450 - .001X

The Profit function becomes
F(X) = (450 - .001X)X = 400X - .001X
2



This is a concave function.
400,000 0
TOSHI CAMERA
88
To obtain an optimal solution (maximum profit),
two conditions must be satisfied:
A necessary condition dF/dX = 0
A sufficient condition d
2
F/dX
2
< 0.
The necessary condition is satisfied at:
dF/dX = 400 - 2(.001)X = 0; X = 200,000.
The sufficient condition is satisfied since
d
2
F/dX
2
= -.002.
The optimal solution:
Produce 200,000 cameras.
The profit is F(200,000) = $40,000,000.
TOSHI CAMERA
89
If a function is known to be concave or
convex at all points, the following condition is
both a necessary and sufficient condition for
optimality:

The point X* gives the maximum value for a
concave function, or the minimum value for a
convex function, F(X), if at X*
dF/dX = 0
Optimal solutions for concave/convex
functions with one variable
90
Determining whether or not a multivariate
function is concave or convex requires analysis
of the second derivatives of the function.
A point X* is optimal for a concave (convex)
function if all its partial derivatives are equal to
zero at X*.
For example, in the three variable case:
c
c
c
c
c
c
F
X
F
X
F
X
1
1
2
2
3
3
0 0 0 = = = ; ; ;
Optimal solutions for concave/convex
functions with more than one variable
91
13.8 Constrained Nonlinear
Programming Problems one variable
The feasible region for a one variable
problem is a segment on a straight line
(X > a or X s b).

When the objective function is nonlinear
the optimal solution must not be at an
extreme point.
92
TOSHI CAMERA - revisited
Toshi Camera needs to determine the
optimal production level from among the
following three alternatives:
150,000 s X s 300,000
50,000 s X s 175,000
150,000 s X s 350,000

93
400 0
X > 250,000
X s 350,000
X* = 250,000
400
0
400 0
Maximize F(X) = 400X - .001X
2
X > 150,000
X s 300,000
150
X > 50,000
X s 175,000
50 175
X*=200,000
X* = 175,000
The objective function does not change:
TOSHI CAMERA solution
94
Constrained Nonlinear Programming
Problems m variables
m n 2 1
2 n 2 1
1 n 2 1
n 2 1
B ) X ..., , X , Gm(X
.
.
.
B ) X ..., , X , G2(X
B ) X ..., , X , G1(X
ST
) X ..., , X , X ( F Maximize
s
s
s
Let us define Y
1
, Y
2
, ,Y
m

as the instantaneous improvement
in the value of F for one unit
increase in B
1
, B
2
, B
m
respectively.
95
This is a set of necessary conditions for optimality of
most nonlinear problems.
If the problem is convex, the K-T conditions are also
sufficient for a point X* to be optimal.
|
.
|

\
|
c
c
+ +
|
.
|

\
|
c
c
+
|
.
|

\
|
c
c
=
c
c
|
.
|

\
|
c
c
+ +
|
|
.
|

\
|
c
c
+
|
.
|

\
|
c
c
=
c
c
= = =
>
m
m
2 1
1
m
m 2
1
1
1
m m 2 2 1 1
m 2 1
X
Gm
Y ...
X
2 G
2 Y
X
1 G
Y
Xn
F
.
.
.
X
Gm
Y ...
X
2 G
Y
X
1 G
Y
X
F
. 4
0 S Y ..., , 0 S Y , 0 S Y . 3
0 Y ..., , Y , Y . 2
. feasible is * X . 1
2
S1, S2, ,Sm
are defined as
the slack variables
in each constraint.
Kuhn-Tucker optimality conditions
96
PBI INDUSTRIES
PBI wants to determine an optimal production
schedule for its two CD players during the month
of April.
Data
Unit production cost for the portable CD player = $50.

Unit production cost for the deluxe table player = $90.

There is additional intermix cost of $0.01(the
number of portable CDs)(the number of deluxe
CDs).
97
Forecasts indicate that unit selling price for
each CD player is related to the number of
units sold as follows:
Portable CD player unit price = 150 - .01X1
Deluxe CD player unit price = 350 - .02X2
PBI INDUSTRIES
98
Resource usage
Each portable CD player uses 1 unit of a particular
electrical component, and .1 labor hour.
Each deluxe CD player uses 2 units of the
electrical component, and .3 labor hour.
Resource availabilty
10,000 units of the electrical component units;
1,500 labor hours.
PBI INDUSTRIES
99
PBI INDUSTRIES SOLUTION
Decision variables
X1 - the number of portable CD players to produce
X2 - the number of deluxe CD players to produce
The model
Max (150-.
X X X X X
ST
X X
X
01X1)X1+(350-.02X2)X2 - 50X1- 90X2-.01X1X2 =
-.01X1

.1X1 +.3X2 1,500
- X1
2
+ +
+ s
s
s
s
. .
,
02 2 01 1 2 100 1 260 2
1 2 2 10 000
0
2 0
2
Production cannot be negative
Resource constraints
100
For a point X1, X2 to be optimal, the K-T conditions
require that:
Y1S1 = 0; Y2S2 = 0; Y3S3 = 0; Y4S4 = 0, and
) 1 ( Y ) 0 ( Y ) 3 (. Y ) 2 ( Y 260 2 X 04 . X 01 .
or
X
G
Y
X
G
Y
X
G
Y
X
G
Y
X
F
) 0 ( Y ) 1 ( Y ) 1 (. Y ) 1 ( Y 100 X 01 . X 02 .
or
X
G
4 Y
X
G
Y
X
G
Y
X
G
Y
X
F
4 3 2 1 1
2
4
4
2
3
3
2
2
2
2
1
1
2
4 3 2 1 2 1
1
4
1
3
3
1
2
2
1
1
1
1
+ + + = +
|
.
|

\
|
c
c
+
|
.
|

\
|
c
c
+
|
.
|

\
|
c
c
+
|
.
|

\
|
c
c
=
c
c
+ + + = +
|
.
|

\
|
c
c
+
|
.
|

\
|
c
c
+
|
.
|

\
|
c
c
+
|
.
|

\
|
c
c
=
c
c
PBI INDUSTRIES SOLUTION
101
Finding an optimal production plan.
Assume X1>0 and X2>0.
The assumption implies S3>0 and S4>0.
Thus, Y3 = 0 and Y4 = 0.

Add the assumption that S1 = 0 and S2 = 0.
From the first two constraints we have X1 = 0 and
X2 = 5000.
X1= 0
A contradiction X1>0
AS A RESULT THE SECOND ASSUMPTION CANNOT BE TRUE
PBI INDUSTRIES SOLUTION
102
Change the second assumption. Assume that S1 = 0
and S2 > 0.
As before, from the first assumption S3 > 0 and S4 > 0.
Thus, Y3 = 0 and Y4 = 0.
From the second assumption Y2 = 0.
Substituting the values of all the Ys in the partial derivative
equations we get the following two equations:
-.02X1 - .01X2 + 100 = Y1
-.01X1 - .04X2 + 260 = 2Y1
Also, since S1 = 0 (by the second assumption)
X1 + 2X2 = 10,000
Solving the set of three equations in three unknowns we get:
PBI INDUSTRIES SOLUTION
103
X1 = 1000, X2 = 4500, Y1 = 35

This solution is a feasible point (check the constraints).
X1 and X2 are positive.
1X1 + 2X2 <= 10,000 [1000 + 2(4500) = 10,000]
.1X1 + .3X2 <= 1,500 [.1(1000) + .3(4500) = 1450]
This problem represents a convex program since
It can be shown that the objective function F is concave.
All the constraints are linear, thus, form a convex set.

The K-T conditions yielded an optimal solution
PBI INDUSTRIES SOLUTION
104
Portable Deluxe Total Profit
April Production 1000 4500 810000
Portable Deluxe Used Available
Electrical Components 1 2 10000 10000
Labor Hours 0.1 0.3 1450 1500
PBI INDUSTRIES
PBI INDUSTRIES Excel SOLUTION
105
Copyright 2002 John Wiley & Sons, Inc. All rights reserved.
Reproduction or translation of this work beyond that named in
Section 117 of the United States Copyright Act without the express
written consent of the copyright owner is unlawful. Requests for
further information should be addressed to the Permissions
Department, John Wiley & Sons, Inc. Adopters of the textbook are
granted permission to make back-up copies for their own use only,
to make copies for distribution to students of the course the
textbook is used in, and to modify this material to best suit their
instructional needs. Under no circumstances can copies be made
for resale. The Publisher assumes no responsibility for errors,
omissions, or damages, caused by the use of these programs or
from the use of the information contained herein.

Potrebbero piacerti anche