Elchanan Mossel (UC Berkeley) CS 170:spring 2014: April 3, 2014 1 / 16

CS 170:Algorithms
Prof Elchanan Mossel.

Based on: Algorithms by Dasgupta Papadimitriou and Vazirani.
Slides edited from a version created by Prof. Satish Rao.
For UC-Berkeley CS170 Spring 2014 students use only.
Do not re-post or distribute
Elchanan Mossel (UC Berkeley) CS 170:Spring 2014 April 3, 2014 1 / 16
CS 170: Algorithms
S
H
H
H
H
.
.
.
.
.
Linear Programming today!!!!
No laptops please.
Thank you !!!
Outline
Maximum Flow.
Problem.
Linear Program
Upper bound: Cut!
Residual Network
Algorithm
Minimum Cut.
Maximum ow
Flow network G = (V, E), source s, sink t V, capacities c
e
>0.
a
c
d
e
s
t
4
4
3
1
4
2
1
1
1
3
1
2
2
2
1
Find Flow: f
e
1
0 f
e
c
e
. Capacity constraints. 3 =f
s,c
c
s,c
= 4.
2
If u is not s or t
(w,u)E
f
wu
=
(u,w)E
f
uw
. 3 =f
s,c
=f
c,d
+f
c,e
= 2+1.
Maximize: size(f ) =
(s,u)E
f
su
. f
sa
+f
sc
= 1+3 = 4
Optimal? c
ad
+c
cd
+c
et
= 1+1+2 = 4.
The capacity of any s t cut gives an upper bound.
S-T cut.
An s t cut is a partition of V into S and T where s S and t T. Its
capacity is the total capacity of edges from S to T.
Do you know the denition?
Find Flow: f
e
1
0 f
e
c
e
. Capacity constraints.
2
If u is not s or t
(w,u)E
f
wu
=
(u,w)E
f
uw
.
Valid or Invalid?
s
b
c
d
t
3
2
1
1
2
2
1
1
2
3
2
3
s
b
c
d
t
3
2
1
1
2
2
1
1
2
2
2+1 = 2
Algorithms.
FindFlow: f
e
1
0 f
e
c
e
. Capacity constraints.
2
(w,u)E
f
wu
=
(u,w)E
f
uw
.
3
maximize
su
f
su
.
Linear program!
Variables f
e
, linear constraints, linear optimization function.
Cool!
Note...
Integer? (Given integer capacities.)
Yes. There is an integer vertex solution!
Constraint matrix has every subdeterminant being 1, 0, 1.
Vertex solution to linear program must be integral!
Ford-Fulkerson.
Simplex method.
Find s to t path with remaining capacity.
Add to ow variables along path.
Update remaining capacity.
Repeat.
a
c
d
e
s
t
4
4
3
1
4
2
1
1
4 4 X 3
1
1 1 X 0
1
1 1 X 0 1
2 2 X 1
1
Uh oh...
Residual Capacity.
Add to ow along path. Or reduce ow on reverse edge.
Update remaining capacity.
Reduce r
e
=c
e
f
e
and add reverse r
uv
=f
vu
Repeat.
s
a
b
t
1
1
1
1
1
1 1 X 0
1
1 1 X 0
1
1 1 X 0 1
1
1
1
1
1
1 X 0
1 X 0
1
1
No remaining path. Uh oh! Optimal is 2! (At most 2 due to cut.)
Add reverse arcs to indicate reverse capacity.
Bigger Example.
Add to ow along path.
Update residual capacities: : r
e
=c
e
f
e
; r
uv
=f
vu
.
Repeat.
a
c
d
e
s
t
4
4
3
1
4
2
1
1
4 4 X 3
1 1
1 1 X 0
11
1 1 X 0 1 1
2 2 X 1
1 1
4 4 X 3
1 1
3 3 X 2
11
2 X 1 1 X 0
12 2
4 4 X 3
1 1
1 X 0 0 X 1 1
4 X 3 3 X 2
12
3 X 2 2 X 1
12
1 1 X 0
11
3 X 2 2 X 1
23 3
4 X 3 3 X 2
12 2
Check Result...
a
c
d
e
s
t
4
4
3
1
4
2
1
1
1
3
1
2
2
2
1
Find Flow: f
e
1
0 f
e
c
e
. Capacity constraints. 3 =f
s,c
c
s,c
= 4.
2
If u is not s or t
(w,u)E
f
wu
=
(u,w)E
f
uw
. 3 =f
s,c
=f
c,d
+f
c,e
= 2+1.
Maximize: size(f ) =
(s,u)E
f
su
. f
sa
+f
sc
= 1+3 = 4
Optimal? c
ad
+c
cd
+c
et
= 1+1+2 = 4.
Any s t cut gives an upper bound.
Correctness.
1. Capacity Constraints: 0 f
e
c
e
.
Only increase ow to c
e
.
Or use reverse arcs decrease to 0.
Flow values to be between 0 and c
e
.
2. Conservation Constraints:
ow into v = ow out of v (if not s or t .)
Algorithm adds ow, say f , to path from s to t .
Each internal node has f in, and f out.
Optimality: upper bound.
s-t Cut: V =ST and s S and t T.
s
t
Lemma: Capacity of any s t cut is an upper bound on the ow.
C(S, T) - sum of capacities of all arcs from S to T
C(S, T) =
e=(u,v):uS,VT
c
e
For valid ow:
Flow out of (S) = Flow out of s.
Flow into (T) = Flow into t .
For any valid ow, f : E Z+, the ow out of S (into T)
eST
f
e
eTS
f
e
eST
c
e
eTS
0 =C(S, T).
The value of any valid ow is at most C(S, T)!
Optimality: max ow = min cut.
At termination of augmenting path algorithm.
No path with residual capacity!
Depth rst search only starting at s does not reach t .
s
t
f
e
=c
e
f
e
= 0
S be reachable nodes.
No arc with positive residual capacity
leaving S
= All arcs leaving S are full.
= No arcs into S have ow.
Total ow leaving S is C(S, T).
Valid ow = all that ow from source.
Value of ow equals value of C(S, T). and Optimal is C(S, T).
Flow is maximum!!
Cut is minimum s t cut too!
any ow any cut and this ow = this cut.
Maximum ow and minimum s t cut!
Celebrated max ow -minimum cut theorem.
Theorem: In any ow network, the maximum s-t ow is equal to the
minimum cut.
Celebrate!
Back to business: Algorithm Terminates?
It will!!
Flow keeps increasing.
How long?
One more unit every step!
O(mF) time where F is size of ow.
Efciency.
s
a
b
t
100000
1
100000 100000
100000 99999
1
99999
1
1
99999
1
99999
1
99999
1
99999
1
1
Edmonds-Karp
Augment along shortest path.
Breadth rst search!
O(|V||E|) augmentations.
Analysis idea.
d(v) is distance to sink.
Only route ow on (u, v) if d(u) d(v).
Only reverse ow on (u, v) if d(v) d(u).
Maximum d(v) is |V|.
Distances only go up. (To prove!)
Every augment removes edge at a distance.
O(|V||E|) removals.
O(|V||E|
2
) time.
Bipartite Matching
Given a bipartite graph: B = (L, R, E) where E LR.
Cindy
Jin
Shelby
Katya
Chen
Ollie
Ivan
Jacque
Find largest subset of edges (matches) which are one to one.
Bipartite Matching
Algorithm by Reduction.:
From matching problem produce ow problem.
From ow solution produce matching solution.
C
J
S
K
C
O
I
J
s
t
Max ow = Max Matching Size.
Flow is not integer necessarily....
Augmenting path algorithm gives integer ow.
Review.
Maximum ow.
Greedy augment path...
Except reverse old decisions ..
Reverse residual capacities.
Optimality?
Upper bound on optimal: s t cut!
Find ow and s t cut with equal value!
Optimal plus proof of optimality.
Application to bi-partite matchings.

Elchanan Mossel (UC Berkeley) CS 170:spring 2014: April 3, 2014 1 / 16

Caricato da

Informazioni sul documento

Descrizione originale:

Titolo originale

Copyright

Formati disponibili

Condividi questo documento

Condividi o incorpora il documento

Opzioni di condivisione

Hai trovato utile questo documento?

Questo contenuto è inappropriato?

Copyright:

Formati disponibili

Elchanan Mossel (UC Berkeley) CS 170:spring 2014: April 3, 2014 1 / 16

Caricato da

Copyright:

Formati disponibili

CS 170:Algorithms

Prof Elchanan Mossel.

Potrebbero piacerti anche