optimitakon in opeYatioal YescaYch:.

clasucal opömi zabion

Umit-1: Pat-i Tnthoducion amd
technique
Ht Tntroduttion:
method of obtaining the
optimization is the
civCumStamces
vesult
wmdey given
best Cnstrucio
acbvities (design,
In engincevig design
engincevs haue to take mamy
and mairtenanc ). at Seweval
ions
amd mamagevial docis
tedmological
stuges of au Such decisions
The main amd imprttmt Joal
ov to
eauived
either to minimie the effovt
maximite the desived bemeft
iczl
iMecham ical cng ineeys des+gn mechan
Components to
achicuG minimum &mamum
Component 1ife
ot
inolved in Canstrution
Gil cai encincevs
minimum
amd othey to
huildin bvidgc8,
Oveau COSt 6Y maxîmumSatey safd
(Collecfveyknaun as
*minimitation (ov) maximí~ation
objahve All these tasks
optimi tation) of am
a tuncion of CevtGin
Cam be expressed as

(dedsfon) vav+gbcs
af
cam be defincd az the Pvoe53
optimitation
Conditions thot give the maximumn
findimq9 the
tmcion
minimum valweof
a
 Applicah ns of optimi tahon:-
difkvemt dønnarm
Typical oPpiicatiors fiom
fov
úvCvaft and acwspacc b u t w e
Design of
miimu wcht
optimal t1ajectovwes of space vehclu.
Firdim
Findimg the
cuctiol euovrs
3 Optimm design of
Selkcion of a Site fov îrdusty.

opäm+zation pYoblem-
9 tatement of am
PYoGmming
ovo mahem atical
An opimi 2ation
PYObJm Can be stated as toljaus
minimi f(x).
Find = (X1XLi. Kn)Oühich
Comstvaints
Subject to the
9;(x0J:12,. m
j:12,...P
Cx) 0
Vectov Caued design Vectoy
dimensiond
X n -

f() bjective function

9:(x) ond l;ox) ineauaui cauaiy Comstyaints

prvoblem
Cauer Constvaincd opimizaion
do not
invoive any
pvobiens
Somc oPtimitanon
Stated as
Corstyaints and Can be

M which minimi3ej f (X).

n a)
umconStvained
are caued
Cmd Such problems
amd
problem
o pimi zation
 How onc an maximi3atjon
Soluc O
RemaYk
minimr3ation p vobiem?
Pyoblem
asa
0
f) ,f)
Minimum of f(x)-

f)
maximum 0f -
+(K) .

Minimum of 4Y)=
Fi9
m+nimum
VaLue f 4X)
Covyesponds to
-point * vaue b
covvespord3 to max
ard x* also
of functi on
-

Fcx) .

maximi~at-
-ve aa maximtat-
gemevaliby, îndivectu
- Cui thaut
10ss 0f
Cam be Soved Gs a mi nimi 3akion
fn pvob Jom
of
numbey of
Pyoblen.
vaviablcs n amd the number
numbev of
*The YUHed in
p necd not bc
Constraints m amd lov
úmy y
1 Ocsign Nectov: is defAigd
ov acivity
engineevi ng 5ysbem
Any
Any Some of whch ave vCwod
u a n b teß
b a SeE of pYocss.
duvin9 the design
as yaviabies
Ked
untitieß avc wwauy
In geneval,Ccvtain
paiametrys.
ard Uese C pe aSsiyncd
Caucd
Vavigbles
Othe uamtities Cve tveated a
All the
PrOce Omd ave Caued design
in the desigh
OY deision Vaviabies xi, '=1,.n
 Vavíables ahe Collecfvely Yepresembd a
The design
o desgn yCctoy X
n
each
dimensional cavtes iam space with
Tan n-

epy esentin
a design Vaviable
CO-ovdimate axis
Consioevcd, the
space
Space S
s cauecd
Cauad

=I,2.. n) s
xiCi desgn Spare.
o
vaviobk Sfucc
the desgn
the 5pacc
dimensional design
n-
Each Point in the
js caucd a design poin vaviables in
of the impoYtant design
The choice pobem
emds
depemds
avgety dap
agely
gm te
m the

am optimi 30tion
emcE
and his eypeví
USeY
umderstamd that the
thar the
is 7mpovtomt to
HOweNCY, it techniaue
optimi3akion
amd Specd of of
0f
efficiemcy extemt. on
the numbei

to a lavgc
depemd vayiabies.
chosen design design
Chousi te des+gn VrYable,
vavobiok,

by Suacfvely
i)Thus techniquc
technique
of the optimiBation
the effcmoy
Cam be
inc eased Tnorealatian
the
GThe AYst thumb ue 0F
problom 1S
fovmularion of am optimi touhion
vavíables s Possible.

Choose
á5 Fau design
to
I-4 OeSign constraintS

vaviabics hac to
The, seiecded de sotisíy
cevtoin Speified
fumctiona omd othr 1aurercri
Knouwn Gs ConStaints (vcStictions)-

These that must he Satísficd to

constyoins Psiuurg
Collecti yely Caued
Gm cceptoble desgn a e
design carstoints.

11mrtarions on the
that mepvesemt
ConstaTrts
ane
of the 3y3ttm
behauiouN OY PeYfommam ce
turcional corstiaints"
behawiouY oY
ttimed
time
[imitations
Constyaints that yepyeRmt physical
vaviables auaitabiriy,fubricobiri
such a5
design
0i* khawn us geomehyic
and tYamspoytabi liby
oY Side Constyanks"

USUoluy tuwo
ef types of
-Mathemoticauytheve arc
Constraínts: EQuatiby o nequgy Constiaints
elakioships
heuality corstaints StGte that te
grcatey tham.
vaviables ave eihey
Cumong design
amOng Vadue.
uad to a esouNcc
Smauey than ov
velati on Ships
Constvaints State that the
Euauby
veSoUYce value.
Shaud exactly match a
to
Constraints ave uauy move diffiCuut
Eauaiby
amd theefovc, Yecd to be avojded wheveuer
hamdle
vuc In te
Possible. Thus the Scond thumb
tbhe
is that the
tovmuation of opömization pvoblem
 CorstaintS Shoud be
Umbe 0f Complex eAyalby
Kpt GS lo as Possible
5 CorstYaint Suvface
pvoblem
with only
optimization
Coside Cum
inequaiy Costraints 9;(L0
the cauobion
ofX that Satisf
LThe s t of
vauws
SuNtace in the desian
hYPEY
9 (-o. fovms
a
Constiairt y
comStiairt
urfars
fare
is caued o
Space nd
diides
the Spare
the design spare
Constyaint
Suvface
This the othov
the othe
ame is 9;(1)4o and
into
nto buo egins:

is 9;()>0 uheve ()>0 e

in the vegicn
T h e oints IyTng
infeosiobie oy Umorreptob le
ie roicbe
the Yegim
uheve g; (x) 40
poTnts Iying în
0Y CCceptahie
constvaint SAfucez
COecton of au the
The
*The
hich Sepavates the
acptabe
0, J=1,2..m, nfare.
H() OTstyoint
Comosite
egion 1s caued the
tnam
tham
point that
líes on one or move

A design SUvtacc is caud a boumd Point

burd
gne conmStiaint canStrairt is caued am
assodated
and the
amd
Cctive ConStrairnt
amy' constioirt
constiairt
iie on ay
ie
that do not
points
fiee points
Design
knaun
aye
Sunfacc
 pepemdig on Jhethey a pavticuaY design Tpoint
moc eptobie onion
to thc acceptable
Ov
belongs
belongs
cam bc demtiticod ov dassifed onc d the
it
folouir fouv types

Fvee amd atceptable poi nt

Free ad umacceptabiePoint
Goumd amd acceptabe poirt
fBoumd amd umaccptable Poimt
C m t a i n t J O

KL 3ehawiay COrsbairt Ih=o

infeasiab
giM LulIULULL
pornt
E Behawíc Constroint 9y:0
ehawas

JTTITTNTI
COTStain Bound un-
oCceptab1eA
pont Side Consbaint I, =o
Fvee
wmacceptabe \Y\nNN IITITITTT
Boumd uTacceptab Point
Point

Fg:constyaint SUHace
Tn 2Dspace

-6: Objocëe fumdtion: isto

task in the foY mulaion pvocedue
The thivd
Obdis io
in tevms of the design
oY
find a fundtion
and othey problem paromckers twíth
vavia bles
the design is op timied This
YeSPc to which

tunction is lonoun as the obyechive fumcion

Guthe choice of the optimization objefve fmcion
te Pvoblem
TS govevn ed by the natuye of
the objective tunction is n e dh the
C Sucdtion of
tamt dorsfons Tn the whole optimum
most impor
design pvOCeSS
theve may bc more tham
Gi In Some situatiors,
tumtion to be Satis fed imultameous
Ore
One objecb
muliPIC
A n optömizafon pvobiem inuolving
is known a3 mulh-oljectivc
objecive fumctios
Pgramming Problem

ex f(K) a,f(X) to A(X)

:

() Od fal) - bu objecrie fumctioms

f(x)> new objecive funcion fov optimi totion

Cgnstrants whose vaue indicate the
a, a OYe

nelahve împortomC of gne objechve timctim

velative tu the othey
Tn gneval, objective fum ctian Com be of 2 bype)

eithey the obfetive fumction s tó be manimited

Ov to be minimi zed. ut the opbmitotion nthud
Ove USUaluy eithey tov minimization pYoblen ov

fUY moximization pvoblom

Fovtunat ethe dupícate pvicipic* heps u
auauing the Some method obe used fov
by
minimitGtion 0Y mayimtcion (with minov Chage

in the objetive
tuntion inStead of a change
 in emtive design

T f the method/design
is doueoped fY SoM n9
Yoblemr E Cam also he wed to
a minmitOtim

Solve a aximizution pYohlom STmPy

murying the obiecive fmtrion-(minus) aned
Vice vEvSa

L objective tumtion Surtage-

all pojnts saisfying f{X)=c=Constamt
The 1ocus of
Space, and
fovms a hypey Suyfare in the design
a difkVent
ToY Cach vawe o c theve CoYYesponds
membey 0fa fimity of Suvfaces
These Suface caued objectivc furstim SUfaccs can
be shoun ina uu dimensional design SPace in

folluing f

optimum
T
Poin TTIT TTTTTT
Mce the Objectivc tumcion Surface ave dawn alon
the opmum Point
with the COTStaint Sufaces, 3ur
dlateimintd withoat much di ffulh
Con be
hat: whoneeY the nurmbey
the muin pYoblem is
the comsQTnt
thvee,
vaviab les eXCeelh tun
a
design eNn
fumuion Sufaes become ComPiCX
d objective has to be SoNed
ha Sokvcd
n d the Problem
visuali3otion
toY pvobicmm
mothemutical,
a
Prely
optimitatiom pvoblem fomugiOn:
bhe
mathematical mode d
USed to create
optimal problsm
Neec to
Optimi2cution)

IchoSedesign
va i@ble
Pomuate crstvaints

ond Jov vovigbie koumds)

UZFonmulaie ohjccive
fumcbon
choose om Opömzaion |2
techniue

->obtain Soluhin
 a t Step Ts to Teala the med fov usire
The
opimization in a s P a hc doign pvoblem

to hoose the impõkan

Then the desighey needs
design vaiablea a Ssoúarcd with the poblem
The fovmulation opbmization Pvoblem. inuolves dthey
cOmsidevations suh 0 Cor5tVaints,ojechvc fmikian

Tpical optimization PYoblem mathemmakical

madel iSas follous:
MOximitc oY minimjte (ohjectic funaism)
SUbject toC CCmSbraints)
Cov)
optimite(objectivc furrction)
Subjct to ( conStroints)

Problems:-
Classi fication of oPtim zotion
m the <xistane of constraints
( clossificatiom based
As dieused, optim tatian Com bc classificd
a corstraincd oY Uncanstyajncd dlapending
On whcthey or not Constairts exist in thc
PYoblem
i) Classifiation based an the Natue be desgn
yaviobes

G pavomckr(ov) Static optimi3ation

The pyobjam is to find vaues to a set of
desgn
desgn aYamEkks that make Some presevibecd
fumction or these Pavamctb mimimum ulyed
to cevtain Carstain3
Gi Tajectdy (oiy dynomuc opbmimtion of
s to find valuos tua set iign
The pvoblem
Contbmous fumctions
PavamekYS uhich ae alu
that minimßej a n objectue
ctheY pavamcty
Some CorstvcintsS
a Set of
tunchon ubject to
ficoion bosed on
the Phy5ical stuctuse c
Classi
the problemn:
Ls oplimal Contoj ond non optümal Conto

Pyobloms
conbvo) Pyobjem 5 a math ematicck
An optimal

pvogyommig pvabiem invoivirg numbey of

euolve fom the
Stage,wheve each stoge
Stage ina pvesci bed mamney
pyeceedlimg
vaviable) dcscvibej it those ne
Ly Two

L Comthoi (daigh)-> Jovenn the enolubon

L State Vay+able)

Clasi fications toscd on bhe nature cb the eauation;

(iy
inuolved
L 1ineaY,nonlineay, geometiic and quadsatc
pvogyanming vobl ems
Nan inea On 3traint famction is
non lincaYthen pyo blem
Caued non- lircav
Pogvammir.
emetic - CmstYaints eypye ssed a3
ave
Posyngmialy in X.
 A function h(X) is caud Poynomia ib h can

expveSSCd a) t h SumoF PouueY tems ech d

he
the fovm
CXh .... Xin

in a nomlincay
uacd yatic - i t is G Pvoblemm

pvoblen with a uabatic objcctive

Pvogyammmig
function and nar Contraints
ConStainby ahe
Lineay poqyammming pvoblam:-

incay fumcrioYs ot bhe design vaviabe

. which minimites

f0-c
JEl,2,..m
ay Xi =bj,
,2 -.,n
classification based on fermissib le'valuès ck csign
avableS:
vawes pevmitted fov the design vaviable,
opimi3aion probloms cam lbe (lassi-ficd as integer-"
amd eal valued" Pregramming pYobleTns "

design Vaviables X,X ¥n of m oPimizatiom

Pvob ems ve cstYicke to take on Only intogoY
Cdiscrebe) values, the pvobam is Caued am
inkag
PTOgyammîrg PYoblom
 vaviable
-on the othev hamd; if au the design
take any eal vauue, the optimtobion
ve peymitted to
-valuwed rogiamming prdttem.
PrYoblem is caluod aeal
bhe deieyminsbtt matove d the
Classi fication baerd on
Vaualies
noune ok the Vaiaieu
om the detevminstic
Basec
pvobtlems Cam he lagificd
inNolved,OPtimítation

03 a dete1mi Hstic md
Stochaitic pYOgíammi
ProblmS
opti is am mRtation
- A Stochastic
programmig Poblom
whicdh Some oy au d
khe paYometers
pvoblem in
Cdesign vovigbe ond /o preuigned pavamettYs)
dve prdbabiistic (nan deermintite 8 Stochastic)
poblem type3
-Accoydig to thB defiriion,the
pveviouy oe detminstic Prgrammirg
Considerd
PYoblems
the timctior
boaad an the sepoYbiniby d
Vid classi Acation
Cam be Clapifiecd as Sepavabe
ems
opti mitabon pvobl
based
md man Separabie pyogi omming problems
Cnstaint
on the Sepavabiliey oh the objetiwe amd
fimckions h
Serrable pNogyamring
pvoblem dhe 15
A
@nd bhen canstaints
fmction
whch the 0bjective
said to be
ae Scpavobe. A fimcion f()is
iF it ambe expveSSd the Sum
Separa bie
obr n Sumak vaialdt fomcians f.C),S.(K)..
fntn) ie
 fC)-FY)
cosSificahion ha Sr on t e nmhn OF 0bject tmchin
Ni

Depondimg On tr rimber of abjatiw funtions

to be minimilcd, optim z0tion problems (om be
dassifod as Single- ( mult.Ohiccti PYgoming

Probems
The problCms casideved prveui ouly are Singk
pvobloms.
ohi activc PYogromming
Problm cm be
A muh objctive
A prusrN OTmi
Stated o8 fonow
f.(x), F(X) A)
Find X hich minimites
1,1,...m
Subject tu ()40 j
be
-> donote okjectie fundion to
t,ffk
minimited Simutoncau
Pat-2 -Cdossical optimizatiom technique-
Ux diffkremtial
Classical optimtaion techniques
to dekmine points of maxima ad
Calauus
diffrontiabe fumctiang
Ond
minima of Contimous
tov
L optimum Solubion
Singie-varviable ftumction
funchon with
o onSbain
muli-variabie
-a
with Caualiby .Cmd
G meutivaviabie tumcbon
cqualiy ConStiants

2 S i n g e vaviable Optmizcibn!

furchion
tochniue) of singk vaviabe
optimizakion
easiey tounderStmd Since
Simpe
ond
Ove 0ong one unvialte.
n e vavialle
involve
funcins
vaiab!e
Singe is
optimizaticon problem
A s i g -viabe in inknaloo
in intval
uhich he vaue
of X=x
one in
f(),uhere flr)
Suh that x* minimite
the objectve fümtin a'x is a Yoal vaviabie
find a Sduion
*puipose af
optmkation is to
fCX is minimum
consideved
LGquúvalomtdual problem
-f(K), ad
mînimited
it to be
minimum:
*Relotiue(o)lOcal i d to
) s
o e vaiable
A funcion-of
A
X=Xt 16
a eldhve lOcl mini mum at
hawe positivc
sufticimtuy Small
f f{xAH) fOv au

value) ob h-
amd ncJative
 moiimum:
Relatu O LOa
Soid to hONe a Yelakiwe ov locad maximun
C fs
at x:
if f(x)>f(x>+h) fo ay sufficierky

Smou vauwe of h doe to Zen

minimunm:
Gobal (or) akSone
not Just fur au x
X ,and
f(x*) fx) to
au

(ie: itewal Cgntved at x, (o)

Close to x* f()
ovey uhich
neihbahood of x), fn the domain
is defncd
Ga1obal (o) 0h50ue maimum;
amd not ust fo u x
f(x*)2f() .fot auw
Cemtved at yt,()
lose to x* Cie interuad
domain which
neighbanhood uf x* ) fn the
ovey

fC is defined
camSboints
PrOblem withou
optimizaHon
YSingie vaviahe fh' (x¥)=o, but
tx): 4(x*)=
. .

Let
ís
#o, then f(x*)
f (x*) is
if f n ) > o
amd n
C)
i a minimum of
Cwen n is
if f (x*J0and
moximum of f()
a

nminmAm ib ' n is odd

maxkunm
GinuheT
Gii nuithey

Cuherc x* is a staionay Point

Prabep- amo minimum auue
Delevminc bhe maj mum
CK) 12x45X* 440x3+5
-0
Givm f() n . 45x
440x3 +5
Sol
difwvt to *
( x - 12(5x)-45(4K3) tho(3x9

3+12ox2
f U - 60xL180
oliff wYt x
x*t21o% -

G.
fUx)= 24ox SHo

To find Stakionay
points
equale f'UX) =0

Gox 8oK +120K*

=o

6oXx+3x+2]:0.
x - o( L3x+20

Xzo(o*=1,2

Case-

(x)= 2Ho-540t 2HO

-60 <0
at ¥ -
FC hos. maximum

from 12
fC= 12-45 +40+5=

tmax12

tHC-
Put X=2 in cG.
(2) 240[8) -510l) + 24o(1)
) =2400
f(x has min um Ot 1:2.
 fvom
f(2) n3) -
+501444o(8)+5

tmin-
COsC-3 Put X-o in eg.
(o)o
neatheY
moximum (moy minimumm

tumtion
Dekrmine eyhame points of the
fcx 3-3Xt6
given fX) x-3x+6.

f'a) 3 3 =0
duff wvt tox ogain
fX)= 6x
euaie f')=o
to find Stahoray points
3x-30
3(x)6

atx=I, 66: f"(K) >0 -JminimBePoint

0k Y-1, -6 f°Cx) 40 - Maximum Point

:2 MUli-vaviable optimi3afon with no Cumstoints:-

COnside the matix

23
Og 33 g
 The leodimg pvînipe
minovs ne givonby

A: lal
A an Qn

A On O Os
031

Ay A
makix:.
NatuYe of
uae

PYiniple
minovsOf a Posiövc
be 1eadumg PosTtve
pos+ve deAnite.
deAnite.
Tf u is
squae
matiix 'a'
them YC aLenatively
auenatively
minos
Piniple
If bhe Caduirg thøn 'A' is
itieLstone-ve),

amd pos
megative
definite.
be ncgatic A ae
Said to minof of A aYC
pvincipe
au the leocurg SQuae
the suaR
the
IF to 0. then
oy eual
tham
grealeY
övC defioi te
mabix A' is Senmi posi ane
of A
Pinopleminovs

( If au lading
the
Said to be
then A is
'A'
ueinatiwely0 ano 20,

Semi 9ative dc Anit

then
then
a none oF bhe
a bove byp.
ype
Ifit s
(51 is in de Arite.
SQueVe
matsúx
CFind the natwve of Squane matix A = i A 3s

SThe leadi Pinple minors 0F A ave

A=3 So.

AL 9-1=8>o
Az:3
3-
1 -| 3(14)-i(4)-1(:)

42 -4-2

36>0

minors ae0
Qu
eadime Pinipk
Ts Positive
deinite
Toture of mabiy

Hessian matviy (J):

mabix of Second
of f) is tbe
mahix
The Hess am
devivatives of f(). wheve
odey pavtial
T
X = (M,,X . . . ¥n)

Hessam
marix ) =

dMmdy d¥mð¥. dMm3

XmOtn
COse- Tf x - (x.,} then

them
Case-il Tf X#(,M73¥,

YsdY J¥302 3
Condition, toY
Suffient
Suffoemt
TecessoYy and objocuie ftuntb
tunch on
The objoctuie
Optimizakion af
multi- vayiabic
without Cgnstvajnts:-

minima 0f t (x)-
ind the maima (ov)
hop1 To
- where X= (¥,m.Nn
Poinbs ybyequaki
auaking
upa Find the Staionay
0.0,0 0
ian matix () at ench
the Hass
Fovm
Stup3
stabony point Positvc definite
oeñinl"
(J) is positivc
NakiDe d the maGüx
Step: Tf
Then fCx) has minimum
Sep5 Tf the Naune o the makix (J) is ngatie
defirie. then fC has mamum
 Nabe of the moBri () 1s idefi nite
Step6: he
no e x t r e m e poins
hos
then fY)

Probrem: ( ) find
of the fumcion
cytleme point
(0 Find the 1,46
minima 0f +( )=Y,+723+2Y244
the moxima (ar)

Sol Given
f ( x ) = X 3 - + 5 42 % 2 + 4 1 4 6 . 0

wYt to .
Pbauy
diff ea0

3+4%
wYt x.
diff eg
f 6%,+
wYt x?
diff ea

3f 0
Patiy w t to 'x?
diff ea0
f- 3+8%,*0-

t to ¥
df epG pavtialy

3 6, +8
wYt toX
Pavbauy
diFf e
 nts
To find Stabionaay Poi

equote 20. 0
3X-+8 20

3tH0
X T3x )=o 3t8o
10 ( ¥=-8/3 .

o Cor) x, =/3.

ne
Points
The Statigndy
(o.).(0,-8).(and
mabix (J)
-FYom the He sian

J 6t4
6

(o,0) inJ.
Case-i put (x,,y)

+4 o
J :6
Ght?
JLE 32>0

i s tvc defnite
fhas minimum atio,o)

6 hom eq0]
tmin
Case-i Put(x,a) (o,~8/3) =

40 40
J 61t4
61+8 : o Gt8 o-8

eodima PYinipe =
J,:4 >o

-3220
matv J is indehrite
minimum
noY
4has nathey maxîmum

=,)innJ
CaSe-: put (x,,}1)
J-

leaduig PYindpie J,
-4 20
J, -3 40
mautvix J s indefnite
+ha neithey maimum
(mo) minimum

case-w put (X.x.) =

(3 inJ
J:6%tH
L o 6X2t8
-C Lo -8J

: -4<0.
Pviniple
J,
eaci
definite.
Maix is -ve

maxîmum at (/3,-8/3)
f has
+ay4 4x1+6.
No, rax N+

may=16-69