linear programming and applications graduate program in business information systems aslı sencer

33
LİNEAR PROGRAMMİNG AND APPLİCATİONS Graduate Program in Business Information Systems Aslı Sencer

Upload: gwen-parrish

Post on 22-Dec-2015

216 views

Category:

Documents


1 download

TRANSCRIPT

Page 1: LINEAR PROGRAMMING AND APPLICATIONS Graduate Program in Business Information Systems Aslı Sencer

LİNEAR PROGRAMMİNG AND APPLİCATİONS

Graduate Program in Business Information

Systems

Aslı Sencer

Page 2: LINEAR PROGRAMMING AND APPLICATIONS Graduate Program in Business Information Systems Aslı Sencer

OPERATIONS RESEARCH

BIS 517-Aslı Sencer

2

What is Operations Research? Collection of techniques used to allocate the scarce resources in the “best” –OPTIMAL – way!

Best of what? We need an “objective” to be minimized or

maximized! Profit, Cost, Utility, Delay, Distance, Flow, etc.

Page 3: LINEAR PROGRAMMING AND APPLICATIONS Graduate Program in Business Information Systems Aslı Sencer

Some applications

BIS 517-Aslı Sencer

3

Resource allocationProduction and inventory planningCapacity Planning Workers and machine schedulingInvestment planningFormulating marketing and military

strategies

Page 4: LINEAR PROGRAMMING AND APPLICATIONS Graduate Program in Business Information Systems Aslı Sencer

Some news about OR:

BIS 517-Aslı Sencer

4

A Wall Street Journal Article lists the use of LP as one of the greatest technological innovations of the past 1000 years.

1975 Nobel Prize for economics: T.C. Koopmans and L.V. Kantoprovich for their contributions in the field

1992 Nobel Prize: Harry Markowitz for his LP based research.

Page 5: LINEAR PROGRAMMING AND APPLICATIONS Graduate Program in Business Information Systems Aslı Sencer

Basic Optimization Techniques

BIS 517-Aslı Sencer

5

Linear ProgrammingInteger and Goal Programming

Transportation, Assignment Models Network Models

Nonlinear ProgrammingStochastic ProgrammingAlso Simulation

Page 6: LINEAR PROGRAMMING AND APPLICATIONS Graduate Program in Business Information Systems Aslı Sencer

LINEAR PROGRAMMING

BIS 517-Aslı Sencer

6

Most successful of all modern quantitative methods.

Program here is not a computer code! It is a plan that efficiently allocate limited resources to achive a goal.

Involves linear relationships, i.e. relations are in the form of lines, planes!

Page 7: LINEAR PROGRAMMING AND APPLICATIONS Graduate Program in Business Information Systems Aslı Sencer

Basic LP Models: Product Mix

Redwood Furniture Co.

Resource

Unit Requirements

Amount Available in

a PeriodTable Chair

Wood (ft) 30 20 300

Labor (hrs) 5 10 110

Unit profit $6 $8

BIS 517-Aslı Sencer

7

Page 8: LINEAR PROGRAMMING AND APPLICATIONS Graduate Program in Business Information Systems Aslı Sencer

What is the optimal plan to max. Profit?

BIS 517-Aslı Sencer

8

Option 1: Allocate all resources to the more profitable item.Total quantity, profit? Any resource left?

Option 2: Is it more profitable to produce less chairs and more tables?

Linear Programming

Page 9: LINEAR PROGRAMMING AND APPLICATIONS Graduate Program in Business Information Systems Aslı Sencer

Formulating a Linear Problem

Define variables:

: number of tables produced in a period

: number of chairs produced in a period

Define constraints:

Define Objective Function

labor)(110105

wood)(3002030

ct

ct

XX

XX

tX

BIS 517-Aslı Sencer

9

cX

cXMaximize 86XProfit t

)itynonnegativ(0, ct XX

tX

cX

Page 10: LINEAR PROGRAMMING AND APPLICATIONS Graduate Program in Business Information Systems Aslı Sencer

How is an LP solved?

BIS 517-Aslı Sencer

10

Graphical Method: Applicable to a maximum of two decision variables.

Simplex Method: Applicable to all LP.Takes long to implement manually.

Use softwares based on simplex and other techniques.

Page 11: LINEAR PROGRAMMING AND APPLICATIONS Graduate Program in Business Information Systems Aslı Sencer

Graphical Solution

BIS 517-Aslı Sencer

11

Constraint 1

Constraint 2

11

15

10 22

Optimal Solution: Xt=4 tables, Xc=9 ChairsProfit*=$96

(4,9)

Xt

Xc

Page 12: LINEAR PROGRAMMING AND APPLICATIONS Graduate Program in Business Information Systems Aslı Sencer

Basic LP Models: Feed Mix

Two types of seeds are mixed to formulate the wheat of wild birdseed.

Nutritional Item

Proportional Content

Total RequirementBuckwheat

Sunflower wheat

Fat .04 .06 ≥480 lb

Protein .12 .10 ≥1200 lb

Roughage .10 .15 ≤1500

Cost per lb $.18 $.10

BIS 517-Aslı Sencer

12

Page 13: LINEAR PROGRAMMING AND APPLICATIONS Graduate Program in Business Information Systems Aslı Sencer

LP Formulation

BIS 517-Aslı Sencer

13

vity)(nonnegati 0, X

(roughage) 150015.00.10X

(protein) 120010.00.12X

(fat) 48006.00.04X toSubject

10.00.18XCost Minimize

mixture in (lb)sunflower of Amount :

mixture in (lb)buckweat of Amount :

b

b

b

b

b

s

s

s

s

s

s

b

X

X

X

X

X

X

X

Page 14: LINEAR PROGRAMMING AND APPLICATIONS Graduate Program in Business Information Systems Aslı Sencer

Graphical Solution to Feed Mix Problem

BIS 517-Aslı Sencer

14

Fat RoughageProtein

(15000,0)(10000,0)

(3750,7500)

Xb

Xs

Optimal Solution:Xb*=3750 lb, Xs*=7500 lbCost*=$1425

Page 15: LINEAR PROGRAMMING AND APPLICATIONS Graduate Program in Business Information Systems Aslı Sencer

Types of Feasible Regions

BIS 517-Aslı Sencer

15

Bounded Feasible Region

Unbounded FeasibleRegion

Page 16: LINEAR PROGRAMMING AND APPLICATIONS Graduate Program in Business Information Systems Aslı Sencer

Cases in an LP: Infeasible Solution

BIS 517-Aslı Sencer

16

No feasible region

If an LP has no feasible region, then the solution is INFEASIBLE!

Do all LP has an optimal Solution?

Page 17: LINEAR PROGRAMMING AND APPLICATIONS Graduate Program in Business Information Systems Aslı Sencer

Cases in an LP: Multiple Optima

0, X

4553X

7248X

6065X toSubject

12X01P Maximize

21

21

21

21

21

X

X

X

X

X

7

24

7

66

)1(

7

36

7

24

*2

*1 X

X

10

BIS 517-Aslı Sencer

17

Optimal Solutions:Point(1): X1*=4 2/7, X2*=6 3/7Point(2): X1*=6 6/7, X2*=4 2/7P*=$120

(1)

(2)

Infinite number of optimal solutions exist in the form

Page 18: LINEAR PROGRAMMING AND APPLICATIONS Graduate Program in Business Information Systems Aslı Sencer

Cases in an LP: Unbounded Optimal Solution

BIS 517-Aslı Sencer

18

Optimal Solution:X1*=, X2*=P*=

Page 19: LINEAR PROGRAMMING AND APPLICATIONS Graduate Program in Business Information Systems Aslı Sencer

Solving Linear Programs with a Spreadsheet

BIS 517-Aslı Sencer

19

Write out the formulation tablePut the formulation table into a spreadsheetUse Excel’s Solver to obtain a solution

Page 20: LINEAR PROGRAMMING AND APPLICATIONS Graduate Program in Business Information Systems Aslı Sencer

Solution in the Excel Solver

BIS 517-Aslı Sencer

20

Page 21: LINEAR PROGRAMMING AND APPLICATIONS Graduate Program in Business Information Systems Aslı Sencer

Applications of LP:Transportation Models

BIS 517-Aslı Sencer

21

Sporting goods company

Capacity Plants Warehouses Demand

Juarez

Seoul

Tel Aviv

Yokohama

Phoenx

NY

Frankfurt100

300

200

150

100

200

150

Page 22: LINEAR PROGRAMMING AND APPLICATIONS Graduate Program in Business Information Systems Aslı Sencer

LP:Transportation Models (cont’d.)

From Plant

Destination

Frankfurt NY Phoenix Yokohama

Juarez $19 $7 $3 $21Seoul 15 21 18 6

Tel Aviv 11 14 15 22

BIS 517-Aslı Sencer

22

Shipping Costs per pair of skis

What are the optimal shipping quantities from the plants to the warehouses, if the demand has to be met by limited capacities while the shipping cost is minimized?

Page 23: LINEAR PROGRAMMING AND APPLICATIONS Graduate Program in Business Information Systems Aslı Sencer

LP:Transportation Models (cont’d.)

Xij: Number of units shipped from plant i to warehouse j. i=1,2,3 and j=1,2,3,4.

Minimize shipping costs=19X11+7X12+3X13+21X14

+15X21+21X22+18X23+6X24

+11X31+14X32+15X33+22X34

From Plant

Destination

CapacityFrankfurt NY Phoenix Yokohama

Juarez X11 X12 X13 X14 100

Seoul X21 X22 X23 X24 300

Tel Aviv X31 X32 X33 X34 200

Demand 150 100 200 150 600

BIS 517-Aslı Sencer

23

Page 24: LINEAR PROGRAMMING AND APPLICATIONS Graduate Program in Business Information Systems Aslı Sencer

LP:Transportation Models (cont’d.)

BIS 517-Aslı Sencer

24subject to#shipped from a plant can not exceed the capacity:

X11+X12+X13+X14≤100 (Juarez Plant)

X21+X22+X23+X24≤300 (Seoul Plant)

X31+X32+X33+X34≤200 (Tel Aviv Plant)

#shipped to a warehouse can not be less than the demand:

X11+X21+X31≥150 (Frankfurt)

X12+X22+X32≥100 (NY)

X13+X23+X33≥200 (Phoenix)

X14+X24+X34≥150 (Yokohama)Nonnegativity

Xij ≥0 for all i,j.

Page 25: LINEAR PROGRAMMING AND APPLICATIONS Graduate Program in Business Information Systems Aslı Sencer

LP:Transportation Models (cont’d.)

BIS 517-Aslı Sencer

25

Capacity Plants Warehouses Demand

Juarez

Seoul

Tel Aviv

Yokohama

Phoenx

NY

Frankfurt100

300

200

150

100

200

150

Optimal Solution: Optimal cost=$6,250

100

50

100

150

100

100

Page 26: LINEAR PROGRAMMING AND APPLICATIONS Graduate Program in Business Information Systems Aslı Sencer

LP: Marketing Applications

How to allocate advertising budget between mediums such as TV, radio, billboard or magazines?

Ex: Real Reels Co. Allocated ad. Budget=$100,000

Playboy True Esquire

Readers 10 million 6 million 4 million

Significant Buyers

10% 15% 7%

Cost per ad $10,000 $5,000 $6,000

Exposures per ad

1,000,000 900,000 280,000

BIS 517-Aslı Sencer

26

•No more than 5 ads in True and at least two ads in Playboy and Esquire

Page 27: LINEAR PROGRAMMING AND APPLICATIONS Graduate Program in Business Information Systems Aslı Sencer

LP: Marketing Applications (cont’d.)

BIS 517-Aslı Sencer

27

million 11.36Exposure Optimal

ads. 2 Xads, 5 Xads, 3.6 X:Solution Optimal

0,,

2

2

5

)(000,100000,6000,5000,10

28.09.0Exposure TotalMax

Esquire in adds#:

True in adds#:

Playboy in adds#:

ETp

EPT

E

P

T

ETp

ETp

E

T

p

XXX

X

X

X

budgetXXX

XXX

X

X

X

Not integer?

Page 28: LINEAR PROGRAMMING AND APPLICATIONS Graduate Program in Business Information Systems Aslı Sencer

LP: Assignment Models

Assignment of a set of workers to a set of jobs

Individual

Time required to complete one job

Drilling Grinding Lathe

Ann 5min 10min 10min

Bud 10 5 15

Chuck 15 15 10

BIS 517-Aslı Sencer

28

Page 29: LINEAR PROGRAMMING AND APPLICATIONS Graduate Program in Business Information Systems Aslı Sencer

LP: Assignment Models (cont’d.)

333231

232221

131211

101515

15510

10105Jobs of Time TotalMin

3,2,1,

j job toassignednot is i worker if,0

j job toassigned is i worker if,1

XXX

XXX

XXX

ji

X ij

.min20Time Total

1 X,1 X,1X

:Solution Optimal

332211

0

)(1

)(1

)(1

worker single a by performed is job Each

)(1

)(1

)(1

job single a to assigned be canworker A

332313

322212

312111

333231

232221

131211

ijX

LatheXXX

GrindingXXX

DrillingXXX

ChuckXXX

BuddXXX

AnnXXX

BIS 517-Aslı Sencer

29

Page 30: LINEAR PROGRAMMING AND APPLICATIONS Graduate Program in Business Information Systems Aslı Sencer

LP: Diet Problem

BIS 517-Aslı Sencer

30

How much to use of each ingredient so that the nutritional requirements are met in the cheapest way?

Ex: Feed Mix problem given at the beginning of the lecture

Page 31: LINEAR PROGRAMMING AND APPLICATIONS Graduate Program in Business Information Systems Aslı Sencer

LP:Labor Planning

BIS 517-Aslı Sencer

31Addresses staffing needs over a specific time

period.

Hong Kong Bank of Commerce: 12 Full time workers available, but may fire some. Use part time workers who has to work for 4

consequtive hours in a day. Luch time is one hour between 11a.m. and 1p.m.

shared by full time workers. Total part time hours is less than 50% of the day’s

total requirement. Part-timers earn $4/hr (=$16/day) and full timers

earn $50/day.

Page 32: LINEAR PROGRAMMING AND APPLICATIONS Graduate Program in Business Information Systems Aslı Sencer

LP:Labor Planning (Cont’d.)

Time Period Minimum labor required

9a.m.-10a.m. 10

10a.m.-11a.m. 12

11a.m.-noon 14

Noon-1p.m. 16

1p.m.-2p.m. 18

2p.m.-3p.m. 17

3p.m.-4p.m. 15

4p.m.-5p.m. 10

BIS 517-Aslı Sencer

32

Page 33: LINEAR PROGRAMMING AND APPLICATIONS Graduate Program in Business Information Systems Aslı Sencer

LP:Labor Planning (cont’d.)

BIS 517-Aslı Sencer

33

0,

)10...141210(5.04

12

10

15

17

18

16 5.0

14 5.0

12

10

16$50$

.5...,2,1,slot at time start work who tellersPart time # :

dayper tellers timeFull # :

5

54

543

5432

4321

321

21

1

i

i

i

i

PF

P

F

PF

PPF

PPPF

PPPPF

PPPPF

PPPF

PPF

PF

PFCostPersonnelDailyMin

iiP

F