Topic 1: Introduction · F.-Javier Heredia This work is licensed under the Creative Commons...
Transcript of Topic 1: Introduction · F.-Javier Heredia This work is licensed under the Creative Commons...
F.-Javier Heredia
This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivs 3.0 Unported License. To view a copy of this license, visit http://creativecommons.org/licenses/by-nc-nd/3.0/ or send a letter to Creative Commons, 444 Castro Street, Suite 900, Mountain View, California, 94041, USA.
MEI
O/U
PC-U
B :
NET
WO
RK
FLO
WS
Network Flows UPCOPENCOURSEWARE number 34414
Topic 1: Introduction
F.-Javier Heredia http://gnom.upc.edu/heredia
MEI
O/U
PC-U
B :
NET
WO
RK
FLO
WS
intr
oduc
tion
Introduction-2
Introduction Introduction • Definitions:
– Minimum shortest path problems. – Maximum flow problems. – Minimum cost flow problems.
• Network flow problems (NF) as a class of linear programming problems (LP): – Standard form of the minimum flow cost problem. – Nodes-arcs incidence matrix. – Formulation of the minimum shortest path problem. – Formulation of the maximum flow problem. – Formulation of the transportation and assignment problem. – Transformation to the standard form.
• Applications.
F.-Javier Heredia http://gnom.upc.edu/heredia
MEI
O/U
PC-U
B :
NET
WO
RK
FLO
WS
intr
oduc
tion
Introduction-3
Shortest Path Problems
1
2 4
3 5
6
10
25
35
20
15
35
40
30
20
s t
Shortest Path Problems
1
2 4
3 5
6
10
25
35
20
15
35
40
30
20
s t
• Applications: – Minimum cost flow problem. – Minimum time flow problem. – Equipment replacement problems.
Shortest Path Problems • To Identify the shortest path between a source node
(“s”) and the sink node (“t”).
1
2 4
3 5
6
10
25
35
20
15
35
40
30
20
s t
F.-Javier Heredia http://gnom.upc.edu/heredia
MEI
O/U
PC-U
B :
NET
WO
RK
FLO
WS
intr
oduc
tion
Introduction-4
Maximum Flow Problems • To seek how to send the maximum possible amount of
flow from a source node (“s”) to sink node (“t”) in a capacitated network.
3 5
3
1
2 4
6
6
4
2
2
5
6
4
s t 5
4 2
2
3
5
4
2
Maximum Flow Problems
• Applications: determining the maximum steady –state flow of: – Petroleum products in a pipeline network. – Cars in a road network. – Messages in a telecommunication network. – Electricity in an electrical network.
Maximum Flow Problems
F.-Javier Heredia http://gnom.upc.edu/heredia
MEI
O/U
PC-U
B :
NET
WO
RK
FLO
WS
intr
oduc
tion
Introduction-5
Minimum Flow Cost Problems • We wish to determie a least shipment cost of a single commodity
through a network in order to satisfy comsuption at demand nodes with the production of the supply nodes.
• Applications: – Logistics(warehouses to retailers). – Automobile routing in an urban traffic network. – Routing of calls through a telephone system.
1
2
3
4
5
6
7 10 10
-10
-15
5
3 6
4
3 1 2
2
4
6
5 0
10
5
0
F.-Javier Heredia http://gnom.upc.edu/heredia
MEI
O/U
PC-U
B :
NET
WO
RK
FLO
WS
intr
oduc
tion
Introduction-6
Standard form of the minimum cost flow problem (MCNFP) (I)
• Let G=(N,A) be the directed graph defined by the set N of n nodes (|N| = n) and the set A of m directed arcs (|A| = m) .
• Demand/supply vector: bj , j=1,2,...,n: b=[5,10,0,-15,0,5,10]’
• Cost vector: cij , (i,j)∈A : c=[5,3,4,...]’ • Flow: xij , (i,j)∈A : amout of commodity to be sent between node i
and node j through arc (i,j).
1
2
3
4
5
6
7 b7 = +10
-10
b4 = -15
5
3 6
4
3 1 c74 = 2
2
4
6
5 0
10
5 (7,4)
n=7 ; m=11 ; N={1,2,3,4,5,6,7} ; A={ (1,2), (1,3), (2,3),...}
F.-Javier Heredia http://gnom.upc.edu/heredia
MEI
O/U
PC-U
B :
NET
WO
RK
FLO
WS
intr
oduc
tion
Introduction-7
Standard form of the minimum cost flow problem(II)
• Mathematic formulation:
Standard form of the minimum cost flow problem(II)
Standard form of the minimum cost flow problem(II)
– We assume the network is balanced: 01
=∑=
n
iib
• Matrix notation:
≥=
′
(1.2c)0(1.2b):s.a.(1.2a)min
xbxT
xc
Standard form of the minimum cost flow problem (II)
{ } { }
∈∀≥
∈∀=−
=
∑∑
∑
∈∈
∈
(1.1c)
(1.1b):s.a.
(1.1a)
Ajix
Nibxx
xcz
ij
iAijj
jiAjij
ij
A(i,j)ijij
),(0
min
),(:),(:
Balance equations
F.-Javier Heredia http://gnom.upc.edu/heredia
MEI
O/U
PC-U
B :
NET
WO
RK
FLO
WS
intr
oduc
tion
Introduction-8
Node-arc incidence matrix • The associated node-arc
incidence matrix T is:
i (cij ,uij) j
1
2 4
3 5
(15,40)
45,1
0)
(25,20)
(35,
50)
(45,
60)
−−−
−−−−−
=
1110000010110100010110100000110100000011
54321
)4,5()3,5()5,4()3,4()2,3()4,2()3,1()2,1(
T
• Properties: – 2m non-zero elements among nm. – Only 2 elements ≠0 per column. – Every non-zero elements is +1 or -1. – Rank(T)= n-1.
• Consider the following network:
F.-Javier Heredia http://gnom.upc.edu/heredia
MEI
O/U
PC-U
B :
NET
WO
RK
FLO
WS
intr
oduc
tion
Introduction-9
Shortest path problem formulated as a MCNFP
• b1= +1 ; b6= -1 ; bj= 0 ∀j ≠1,6
1
2 4
3 5
6
10
25
35
20
15
35
40
30
20
+1 -1
i cij j
F.-Javier Heredia http://gnom.upc.edu/heredia
MEI
O/U
PC-U
B :
NET
WO
RK
FLO
WS
intr
oduc
tion
Introduction-10
Maximum flow problem formulated as a MCNFP
• Artificial arc xts with cts=-1 and uts= +∞. • cij=0 ∀(i,j)≠(t,s)
3 5
(0,3)
1
2 4
6
(0,5)
s t
i (cij ,uij) j
(-1,+∞)
F.-Javier Heredia http://gnom.upc.edu/heredia
MEI
O/U
PC-U
B :
NET
WO
RK
FLO
WS
intr
oduc
tion
Introduction-11
Transportation problem formulated as a MCNFP
• Properties: – N = N1∪ N2:
N1: production nodes; N2: demand nodes. – ∀(i,j)∈A : i ∈ N1 ; j ∈ N2
1
2
3
4
5
6
7
F.-Javier Heredia http://gnom.upc.edu/heredia
MEI
O/U
PC-U
B :
NET
WO
RK
FLO
WS
intr
oduc
tion
Introduction-12
Assigment problem formulation
• Properties: – N = N1∪ N2 ; |N1 |= |N2| – A⊆ N1× N2.
– bj=+1 ∀j ∈N1 ; bj=-1 ∀j ∈N2
1
2
3
4
5
6
+1
+1
+1
-1
-1
-1
F.-Javier Heredia http://gnom.upc.edu/heredia
MEI
O/U
PC-U
B :
NET
WO
RK
FLO
WS
intr
oduc
tion
Introduction-13
Transformation to the standard form (I)
• Undirected arcs:
– Contribution to the objective function: cij xij +cij xji
– cij ≥ 0 ⇒ at the optimal solution xij > 0 or xji > 0
i j xji
xij
i j xji
xij
– Transformation: each directed arc {i,j} is replaced by two directed arcs (i,j) and (j,i) with cost cij :
F.-Javier Heredia http://gnom.upc.edu/heredia
MEI
O/U
PC-U
B :
NET
WO
RK
FLO
WS
intr
oduc
tion
Introduction-14
Transformation to the standard form (II)
• Arcs with non-zero lower bounds: – xij ≥ lij
– xij is replaced by x’ij+lij in the problem’s formulation:
Bounds: xij ≥ lij ⇒ x’ij+lij ≥ lij ; x’ij ≥ 0
Balance equations: bi → bi - lij ; bj → bj + lij
Objective function: cij xij → cij (x’ij + lij ) = cij x’ij + cij lij
cte.
F.-Javier Heredia http://gnom.upc.edu/heredia
MEI
O/U
PC-U
B :
NET
WO
RK
FLO
WS
intr
oduc
tion
Introduction-15
Transformation to the standard form (III) • Arcs with negative costs:
– Let be xij with cij<0. – Let uij a trivial upper bound of the arc (i,j). – xij is replaced by x’ij = uij – xij.
0 ≤ x’ij ≤ uij
Objective function: z = ...+ cij xij +...→ z’= ...+ cijuij – cij x’ij +...
cst >0
i j x’ij
-cij ,uij
bi - uij bj + uij
i j xij
cij ,uij
bi bj
Balance equations:
F.-Javier Heredia http://gnom.upc.edu/heredia
MEI
O/U
PC-U
B :
NET
WO
RK
FLO
WS
intr
oduc
tion
Introduction-16
Transformation to the standard form (IV)
• Arcs with capacity: – Let 0 ≤ xij ≤ uij. – Network transformation:
i j xij
cij ,uij
bi bj i k
xik cik ,∞
bi - uij
j bj + uij xjk
0 ,∞
The network is equivalent to the original: xij ≡ xik ; xik + xjk = uij ⇒ xik = xij ≤ uij
The objective function id not modified: cik ≡ cij
F.-Javier Heredia http://gnom.upc.edu/heredia
MEI
O/U
PC-U
B :
NET
WO
RK
FLO
WS
intr
oduc
tion
Introduction-17
Transformation to the standard form (V)
• Unbalanced network:
01
>∑=
n
jjb
– Excess of production ( ) :
2
3
4 +8
-8
-7
1 +10
2
3
4 +8
-12
-10
1 +10
01
<∑=
n
jjb
– Excess of demand ( ) :
5 -3
A dummy demand node n+1 is added linking all production nodes through uncapacitated - null cost arcs.
5 +4
A dummy supply node n+1 is added, linking all demand nodes through uncapacitated – null cost arcs.
F.-Javier Heredia http://gnom.upc.edu/heredia
MEI
O/U
PC-U
B :
NET
WO
RK
FLO
WS
intr
oduc
tion
Introduction-18
Transformation to the standard form (VI)
• Example 1: (Solution)
1
2 4
3 5
(-2,10,0)
(-1,
20,0
)
(10, ∞,5)
(2, ∞
,0)
-10
0 10
i (cij ,uij ,lij ) j
bi bj
-15
20
F.-Javier Heredia http://gnom.upc.edu/heredia
MEI
O/U
PC-U
B :
NET
WO
RK
FLO
WS
intr
oduc
tion
Introduction-19
Applications
• Solve the applications examples from book
chapter 1.3 of AMO (1 week):
– Problem description.
– Network formulation and objective function
associated to the problem.
– Problem classification.
• Problems assigment:
F.-Javier Heredia http://gnom.upc.edu/heredia
MEI
O/U
PC-U
B :
NET
WO
RK
FLO
WS
intr
oduc
tion
Introduction-20
Transformation to the standard form Example (1/4)
1. Negative cost elimination
1
2 4
3 5
(2,10,0) (1
,20,
0)
(10, ∞,5) (2
, ∞,0
)
-10+10 = 0
0 10+20 = +30
i (cij ,uij ,lij ) j
bi bj
-15-10-20= -45
20 z=z’+20+20=z’+40
F.-Javier Heredia http://gnom.upc.edu/heredia
MEI
O/U
PC-U
B :
NET
WO
RK
FLO
WS
intr
oduc
tion
Introduction-21
Transformation to the standard form Example (2/4)
2. Lower bounds elimination ≠0
1
2 4
3 5
(2,10,0) (1
,20,
0)
(10, ∞,0) (2
, ∞,0
)
0
0+5 = 5 30-5 = -25
i (cij ,uij ,lij ) j
bi bj
-45+5 = -40
20-5 = 15 z=z’+40+25+50=z’+115
F.-Javier Heredia http://gnom.upc.edu/heredia
MEI
O/U
PC-U
B :
NET
WO
RK
FLO
WS
intr
oduc
tion
Introduction-22
Transformation to the standard form Example (3/4)
3. Capacity elimination
1
2 4
3 5
(0, ∞,0)
(10, ∞,0) (2
, ∞,0
)
0
5 -25
i (cij ,uij ,lij ) j
bi bj
-40+20+10 = 30
15
z=z’+115
6 -20
7
-10
(2, ∞,0)
F.-Javier Heredia http://gnom.upc.edu/heredia
MEI
O/U
PC-U
B :
NET
WO
RK
FLO
WS
intr
oduc
tion
Introduction-23
Transformation to the standard form Example (4/4)
4. Unbalance network
1
2 4
3 5
(0, ∞,0)
(10, ∞,0)
(2, ∞
,0)
0
5
-25
i (cij ,uij ,lij ) j
bi bj
30
15
z=z’+115
6 -20
7
-10
(2, ∞,0)
7 -25