Some Modeling Examples

& network modeller!

!nteger
-l!ncar

* 1-2 tave non-l!near

örnegt var
* loj!st!k , traf!k us modeller!
 Minimax Objectives
* örn : max!mum r!sk! m!n!m!ze etmek !sted!ğ!mde

* max!m!n obj olduğunda, m!n!mumu max

etmeye çalışıyoru
* örn : baz !st!yonlara m!n!mum er!ş!m! olan
yerler! max!m!ze yapmaya çalışab"l"r"z daha kapsayıcı
.
olmak !ç!n

O > bunu max!m!ze

eden
m!n!m!ze etmen
equat!on ,

em!n!m!ze
etmek
!sted!ğm!t sey! constra!ntlere
yazdık
& b!r probab!l!ty !le b!r constra!nt!n
sat!sfy olmasını
!st!yoruz . bu constra!nt kes!n
sat!sfy olucak denek
değ"ld"r bu.

Stochastic Programming

- probab!l!ty

Yükleniyor…

art!c
probab!l!ty
(
e
çonzor od!g
deklem! değ"l
!ç!n

mod!fy ed!pret
 bu endüstr!ler!n outputları başka endüstr!lerde
kullanılıyo
>
-

0 MACRO
Economic
MODEL
te
!nputa,requ!ren
>
-

demandler
.
halkın
!ht!yacı
> her s!tun
- toplanı = 1
MACRO Economic MODEL continued
>
- demandler! nasıl karşılıcaz

m
ne kader pal üret!rsem

10 Gal, %
%
50's!n!
steel endüstr!s!ne

weryorum us .

Yükleniyor…
denklenler!
RhS'a at!p

O
düzenled!k
ve
denklem!
çözüyoruz funct!ons you
.

& object!ve
yartılab!l!r
-
ama

e
laboracoallabou steel elabour e transport
 The minimum cost network flow
problem
b!r network problem! , 2 source'u O 15 tonluk mal var 2 depo düşün
.
var ve 1 , 10 ve ,
g!b!

Findthe
F!nd them!n!mum
minimum cost
cost solut!on

solution
that thatderand
sat!sf!es satisfies
supply
constra!nts M!nd the DIRECTED
demand supply
.

ARCS
constraints. Mind the
.

Supply"-" m

derad!e" _
DIRECTED ARCS
"
 The minimum cost network flow
problem model
star
0 O var
çıkış
"0" O source'tan
mal m!ktarı e
g!denler ,
yen!
O'dan 2'ye g!den
Ölen
E networke balerah
node O

rode 1 O -h!k!s eg!r!s model! yatı Caz

& &
Super
node 2

&

S
( cana
-
X'ler arclardan m!ktarı her rode !ç!n
,
geçen mal
,
flar denklem!
b!r
g!r!ş çıkışı
(sol taraf
• One flow equilibrium constraint for each node. matr!s!
bütün
• This matrix is called unimodular (Akatsayılar
-
0 -1 2'del
oluşuyor Katsayı
matrix of -1,1,and 0). If capacities
,

and
ve + .

demands are integer numbers, you get integer x(i,j) values in optimal solution.
• You can also represent multiple commodities on this network. ↓
X (! !
j)30
d!rected
Shortest PATH Problem

Source
s!nk

Moving from node 0 to node 8 in the

shortest possible travel distance. Each
arc has a distance. How do you model it
as a minimum cost network flow
problem?
her node !ç!n
y!ne flow
deklen!
yat!cal
node 0-Xo1-Xoz =
81 e b!rpat!ka çıkıcak

node 1 Xol X22-X1g 0 -

- =
transh!pment
node

ne8
-

X08 + 78 = Le b!r pat!ra g!r!cen

 traf!k ,
a!l
p!pe l!ne !ç!n kullanılır .

max

Maximum Flow Problem Kapas!te

geçeb!l!cen

summe dummy
Sending as much goods from source

Arcs have flow upper bounds. Below

↑
nodes 0 and 1 to sink nodes 5 and 6.

source s!nk
x(s0) and x(s1) are dummy arcs from
dummy node s to nodes 1 and 0. T is
dummy sink node attached to nodes 5,
·

dummy koymasak
da olur ema b!rde 6 and 7. Write one constraint for each
fazla source
olduğu
!ç!n
boydul node. Sesource Tes!nk ,
ve

s!nk

(sourcelarden)
-
dummy sourcetan s!nklere
g!den m!ktar

I
dummy source dean
hep!pment
The we
rode
.

da

soru duma
boydun
.

* bu matr!s
un!modular
y!ne
LP The coz!leb!l!r
Sherarc'ın
.

upper
bound constra!n t!
Modeling the Critical Path Method
 Modeling the Critical Path Method

Yükleniyor…

Write one constraint for each arc.

 b! ürün
R!yatı yüksel!nce
# denend!
darand!
atal!r

.
orta
,
d!sonce

A nonlinear model: Price elasticity

her
zaman negat!c v!s!
olduy derad
orantık
↑ tes pr!ce
ve

change!n derand
Change!n pr!ce

Onegat!f satamıcağı
kadar yüksek
artık
S
b!r f!get

demand!
>
-
dengelemel !n

d!zg!n pr!ce sev!yes!n! ver!r ·

 Logical Models- binary
programming
Condition: C DreE sen
-AveyaByveyak!s!n!
b!rden Fret!rs!

nas!l
eb!nary
bunu > en
- at b!r! üret!l!cak
sagl!cat
-
başkan b!r
Var!able'a
e en at b!r! üret!l!cek
b
L

sağlanıyorsa
bu da cak
sağlanı
yükl!cet
-

Sa + 5 2S

!se

S + 8D + SEY, 5
 Special Ordered Sets (SOS)
sadece 1 var!able

e en fazla 2 var!able

baskate
b!n

e parkl kapas!te
katsayıları
 Disjunctive Constraints
>
- constra!ntler!n en

az y temes!n!n
sat!sfy
Here Ri are constraints. ed!lmeg!n! !st!yoruz

Where

•R1: 3x+5y <= 15…Relate constraint with binary variable:

3x+5y<=15& +BIGM (1- S 1) - 81 1b!nary
> nasıl
-
Koyer!m ?
var .

olursa BIGM
= olur Jxtsy IS onur
yok .
.

> Sz
- 0 olursa DIGM kalır 3x+
=

bound almat
3y'n!n üec!nde b!r
•R2: 3x+5y >= 3…Relate constraint
upper with binary variable:.

82 1 - B!sm yok dur 3xtSy] olur

3x+5y>=3& - BIGM (1- S 2) e = .

82 0 - lowerband you dur

(
S
=

= 0 !ç!n
And suppose other constraints exist. Relate them with S ’s. And then,
upper ve add: 812 + S 2+….
23 >= 2
+ -

lower bound

kalmayacak ve

The const. a •How do you deal with equality constraints?

nadundent olucak
•R3: 2x+y=5….. - equal!ty olursa 1 4, olarak !k! ayrı conste çer!v!r!t
ve
8 = 1 !ç!n

&
DIGM'ler
Convert R3 into 2 constraints:
2x+y <= 5+BIGM(1- )

!n tasara zayes!n 2x+y >=5 – BIGM (1- )

Reas!bleolur . aynı d!k bağlantı kurduk 52 82 farklı değ" ,
Disjunctive Variables
 Sequentially dependent constraints
When decisions made at a given time affect the ones made later

2 and 0 2 olab!l!r
all Yt !ntegere
,
1
,
Discrete Capacity Extensions
We wish to add storage capacity to a depot but have a discrete set of
capacity alternatives

kısıtı
>
-
Kaynak

-
-
-

boo b,
Zajxj
-
+

And,
And sum of
,
’s = 1 1
of S's
sum =
Maximax Objectives

> max!m!te
- the max!mum
The vehicle Routing Problem (VRP)
WE have m vehicles each with capacity Q(k). They travel from depot to customers and
satisfy each customer demand q(i). Vehicles return to depot after they satisfy several
customers on a tour. Each customer is visited once and has delivery time windows
between hours of a(i) and b(i). Goal is to minimize total travel time(d(ij) time between
customers i and j). (Similar to TSP)
-!f and only !f
eb!nary

panaana
eb!g
M
g!b!b! sey depot
M M
· depot

ed!stance
I