Professional Documents
Culture Documents
Welcometo
620261
620261
Introductionto
Introductionto
OperationsResearch
OperationsResearch
1
620261:
620261:
IntroductiontoOperations
IntroductiontoOperations
Research
Research
Lecturer:PeterTaylor
HeadsOffice
RichardBerryBuilding
Tel:83447887
Email:
p.taylor@ms.unimelb.edu.au
Coursedueto:MosheSniedovich
2
Schedule
Schedule
Lectures:
Mon,Wed,Friday3:15PM
Tutorial:
CheckNoticeBoardandWebsite
3
OfficeHours
OfficeHours
Monday23PM
Wednesday23PM
Friday23PM
Thesemayhavetovarysometimesseemy
assistantLisaMifsud
4
Assessment
Assessment
Assignments:10%
FinalExam:90%
5
GroupProjects
GroupProjects
Youareencouragedtostudywithfriends,
butyouareexpectedtocomposeyourown
reports.
6
Communication
Communication
Youareexpectedtorespondtoquestions
asked(bythelecturer)duringthelectures
Suggestions,comments,complaints:
Directlytolecturer
viaStudentRepresentative
Dontwaittillyouareaskedtocomplain!
7
LectureNotes
LectureNotes
OnSale(BookRoom)
Ifoutofprint,letmeknow
8
ThouShallNot
ThouShallNot
9
ThouShallNot
ThouShallNot
10
StudentRepresentative
StudentRepresentative
(SSLC)
(SSLC)
Pizza!!!!!
Twomeetings
Questionnaire
11
WebSite
WebSite
http://www.ms.unimelb.edu.au/~dmk/620261
12
ReferenceMaterial
ReferenceMaterial
LectureNotes
Bibliography(10copiesofWinstoninMaths
library,reservedShelves)
Handouts
13
ComputerLiteracy
ComputerLiteracy
_ AppliedMathematicsiscomputational.
_ Idontexpectanyspecificknowledge,butI
doexpectanopenattitudetothings
computational.
14
Perspective
Perspective
Universe Applied
maths
620-261
OR
15
WhatisOR?
WhatisOR?
Controversialquestion!
SurftheWWWforanswers
Roughly:
....Applicationsofquantitativescientificmethods
todecisionmakingandsupportinbusiness,
industrialandmilitaryorganisations,withthe
objectiveofimprovingthequalityofmanagerial
decisions.....
16
BasicCharacteristics
BasicCharacteristics
Appliesscientificmethods
Adoptsasystemsapproach
Utilisesateamconcept
Reliesoncomputertechnologies
17
ORStream
ORStream
620261:IntroductiontoOperations
Research
620262:DecisionMaking
620361:OperationsResearchMethodsand
Algorithms
620362:AppliedOperationsResearch
ProbabilityandStatisticsareusefulother
subjectstostudy
18
andmore
andmore
Honours
MSc
PhD
19
Jobs
Jobs
_ ThereisashortageofpeoplewithORskills
_ Graduateswiththeseskillsgetgoodjobs
20
Reading.....
Reading.....
AppendixA
AppendixB
AppendixE
Chapters1,2,3,4
Web
21
The OR Problem Solving Schema
Formulation Monitoring
Realization
Modelling Implementation
Analysis Solution
22
In Practice
Formulation Monitoring
Realization
Modelling Implementation
Analysis Solution
23
ImportantComment
ImportantComment
In620261:
FormulationandModelling
AnalysisandSolution
24
Chapter2:
Chapter2:
OptimizationProblems
OptimizationProblems
Generalformulation
25
Observethedistinctionbetweenfandf(x).
Notethatfisassumedtobearealvalued
functionon.
z*: opt f ( x )
x
26
Example
Example
10
z*: max n n
x
n 1
2
x 3
s. t.
10
x n 5
n 1
xn {0,1} , n 1, .. .,10
27
10
f (x): x x
2 3
n n
n1
10
: x : x n 5, x n {0,1} , n 1,..., 10
10
n1
opt max
z* unknown
28
Welet*denotethesetofoptimaldecisions
associatedwiththeoptimizationproblem.
Thatis*denotesthesubsetofwhose
elementsareanoptimalsolutiontothe
optimizationproblem.Formally,
*:={x*:x*,f(x*)=opt{f(x):x}}.
Byconstruction*isasubsetof,namely
optimalityentailsfeasibility.
29
Remarks
Remarks
Thesetoffeasiblesolution,,isusually
definedbyasystemofconstraints.
Thus,anoptimizationproblemhasthree
ingredients:
Objectivefunction
Constraints
OptimalityCriterion
30
ClassificationofOptimal
ClassificationofOptimal
Solutions
Solutions
Considerthecasewhereopt=min.Thenby
definition:
x**ifff(x*)f(x)x
Ifopt=max:
x**ifff(x*)f(x)x
Solutionsofthistypearecalledglobaloptimal
solutions.
31
f(x) Global max
Global min
X
32
Question:
Howdowesolveoptimizationproblemsof
thistype?
Answer:
Therearenogeneralpurposesolution
methods.Themethodsusedarevery
muchproblemdependent.
33
Suggestion
Suggestion
Trytothinkaboutoptimizationproblemsin
termsoftheformat:
Z*:=optf(x)
s.t.
constraints
34
Thus.........
Thus.........
Modelling=
Modelling=
opt=?
f(x)=?
Constraints=?
35
Tip
Tip
Youmayfinditusefultoadoptthefollowing
approach:
Step1:Identifyandformulatethedecision
variables.
Step2:Formulatetheobjectivefunctionand
optimalitycriterion.
Step3:Formulatetheconstraints.
Butdonotbedogmaticaboutit!!!!!
36
Example2.4.2
Example2.4.2
FalseCoinProblem
FalseCoinProblem
Ncoins
N1havethesameweight(good)
1isheavier(false)
Findthebestweighingschemeusinga
balancebeam.
37
Observations
Observations
Itdoesnotmakesensetoputadifferent
numberofcoinsoneachsideofthescale.
Theresultofanynontrivialweighingmust
fallintoexactlyoneofthefollowingcases:
Falsecoinisonthelefthandside
Falsecoinisontherighthandside
Falsecoinisnotonthescale
38
Theschemeshouldtelluswhattodo
ateachtrial,i.e.howmanycoinsto
placeoneachsideofthescale,
dependingonhowmanycoinsarestill
tobeinspected.
ThetermBestneedssome
clarification:
39
Best=???
Best=???
Best=fewestnumberofweighings
isnotwelldefinedbecauseaprioriwedontknow
howmanyweighingswillbeneededbyagiven
scheme.
Thisissobecausewedonotknowwherethe
falsecoinwillbeplaced.
Thebottomline:whodecideswherethefalse
coinwillbeasweimplementtheweighing
scheme?
40
Weneedhelp!!!
Weneedhelp!!!
Manyofthedifficultiesarenicelyresolvedif
weassumethat
MotherNatureAlwaysPlaysAgainstUs!
Ofcourse,ifyouareanoptimistyoumay
prefertoassumethat
MotherNatureAlwaysPlaysinOurFavour!
41
Assumption
Assumption
MotherNatureAlwaysPlaysAgainstUs!
Observethatthisassumptionresolvesthequestion
ofwherethefalsecoinwillbe.
Naturewillalwaysselectthelargestof(nL,nR,no)
nL nR
no
42
Solution
Solution
Let
n:=Numberofweighingsrequiredto
identifythefalsecoin.
xj:=Numberofcoinsplacedoneachside
ofthescaleinthejthweighing
(j=1,2,3,...,n)
43
Thus,ourobjectivefunctionis
f(x1,x2,...,xn):=n
and
opt=min.
Tocompletetheformulationoftheproblemwe
havetodetermine.
44
Constraints
Constraints
Let
sj:=Numberofcoinsleftforinspectionafter
thejthweighing(j=0,1,2,...,n)
Thenclearly,
s0:=N(Allcoinsareyettobe inspected)
sn:=1(Onlyfalsecoinisleftfor inspection)
45
xj{0,1,2,...,[sj1/2)]}
where
[z]:=Integerpartofz.
46
Dynamics=????
Dynamics=????
Wehavetospecifythedynamicsofthe
process:howthe{sj}arerelatedtothe
{xj}.
Thisisnotdifficultbecauseweassume
thatNaturePlaysAgainstUs:
sj=max{xj,sj12xj}
xj xj
sj-1-2xj
47
(j-1) weighing:
sj-1
coins left
xj xj
j-th weighing:
sj-1-2xj
49
CompleteFormulation
CompleteFormulation
z * : min n
s.t.
so N
sn 1
sj max{x j , s j1 2x j}, j 1, 2,..,n
sj 1
x
2
x j {0,1, 2,...}
50
ExamplesofORProblems
ExamplesofORProblems
Example2.4.6TowersofHanoi
Task:Movethediscsfromlefttoright
Rules:
Onediscatatime
Nolargedisconasmallone
51
Example2.4.4TravellingSalesman
Problem
VisitNcities,startingthetourand
terminatingitinthehomecitysuchthat:
Eachcity(exceptthehomecity)isvisited
exactlyonce
Thetourisasshortaspossible.
Question:Whatistheoptimaltour?
52
Remark:
Thereare(N1)!distincttours.Thismeans
thatfor11citiesthereare3,628,800possible
toursandforN=21citiesthereare2x1018
possibletours!!!!!!!!!!!!!!!
53
Ifwetrytoenumerateallthefeasible
toursforN=21usingasuperfast
computercapableofenumerating
1,000,000,000tourspersecond,wewill
completetheenumerationofallthe
feasibletoursinapproximately800
years.
Thisphenomenonisknownas
TheCurseofDimensionality!
54