73
Network Flow, Chinese Postman,Topological Sorting and Planarity Detection and Embedding Firas M Abu Hassan 20911833 CS 530: Design and Analys is of Algorithms Fall 2010 1
8/3/2019 Maxflow Problem
http://slidepdf.com/reader/full/maxflow-problem 1/73
Network Flow, ChinesePostman,Topological Sorting and
Planarity Detection and Embedding
Firas M Abu Hassan20911833
CS 530: Design and Analysis of Algorithms
Fall 2010 1
8/3/2019 Maxflow Problem
http://slidepdf.com/reader/full/maxflow-problem 2/73
Network flows Problem
Flow network: a flow network G = (V,E) is a directed graphin which each edge (u,v) ∈ E has a nonnegative capacityc(u,v) >= 0. if (u,v) ∉ E , we assume that c(u,v) = 0.
We distinguish two vertices in flow network , source s andsink t.
The source produces the material at a steady rate . The sink consumes the material at a steady rat
Objective : how much of material can be imposed throughnetwork that network can handle. (Find a maximal flow over adirected graph)
2
8/3/2019 Maxflow Problem
http://slidepdf.com/reader/full/maxflow-problem 3/73
Definition of Flow
Each edge can carry an entity called “flow”.
For flow on an edge to be valid, it must satisfythree conditions: Capacity constraint: The flow can not be greater than
the capacity of the edge f ( u,v ) <= c ( u,v )
Skew symmetry: The flow on the edge (u,v) is equal tothe negation of the flow on the edge (v,u)
Flow conservation: The total flow entering a vertex(excluding the source or sink) must equal the totalpositive flow leaving that vertex.
V vV v
t v f vs f f ),(),(||3
8/3/2019 Maxflow Problem
http://slidepdf.com/reader/full/maxflow-problem 4/73
Definition of Flow
S t
v1
v2
v3
v4
Flow network: directed graph G=(V,E)
S t
v1
v2
v3
v4
u v6/12
u v6/6
f(u,v)=6
f(u,v)=6
Flow below capacity
Maximum flow
8
3
6
8
6
6
3
3
Example: oilpipeline
4
8/3/2019 Maxflow Problem
http://slidepdf.com/reader/full/maxflow-problem 5/73
Example of a Flow:
f(v2, v1) = 1, c(v2, v1) = 4. f(v1, v2) = -1, c(v1, v2) = 10.
f(v3, s) + f(v3, v1) + f(v3, v2) + f(v3, v4) + f(v3, t) =
0 + (-12) + 4 + (-7) + 15 =
0
flowcapacity
capacity
5
8/3/2019 Maxflow Problem
http://slidepdf.com/reader/full/maxflow-problem 6/73
Residual Networks
The residual capacity of an edge (u,v) in a network with a
flow f is given by:
),(),(),( vu f vucvuc f
• The residual network of a graph G induced by a flow f is the graph including only the edges with positive
residual capacity, i.e., ( , ), wheref f
G V E {( , ) : ( , ) 0}f f
E u v V V c u v
6
8/3/2019 Maxflow Problem
http://slidepdf.com/reader/full/maxflow-problem 7/73
The basic Ford Fulkerson method
1 for each edge (u, v) E [G]
2 do f [u, v] = 0
3 f [v, u] = 0
4 while there exists a path p from s to t in theresidual network G f
5 do cf (p) = min {cf (u, v) | (u, v) is in p}
6 for each edge (u, v) in p
7do
f [u, v] = f [u, v] + cf (p)8 f [v, u] = - f [u, v]
7
8/3/2019 Maxflow Problem
http://slidepdf.com/reader/full/maxflow-problem 8/73
The basic Ford Fulkersonaugmenting paths Algorithm
S t
v1
v2
v3
v4
12
14
1 for each edge (u, v) E [G]
2 do f [u, v] = 0
3 f [v, u] = 0
4 while there exists a path p from s to t
in the residual network G f 5 do cf (p) = min{cf (u, v) | (u, v) p}
6 for each edge (u, v) in p
7 do f [u, v] = f [u, v] + cf (p)
8 f [v, u] = - f [u, v]
(residual) network Gf
8
8/3/2019 Maxflow Problem
http://slidepdf.com/reader/full/maxflow-problem 9/73
The basic Ford Fulkersonaugmenting paths Algorithm
S t
v1
v2
v3
v4
0 /12
0 /14
(residual) network Gf 1 for each edge (u, v) E [G]
2 do f [u, v] = 0
3 f [v, u] = 0
4 while there exists a path p from s to t
in the residual network G f 5 do cf (p) = min{cf (u, v) | (u, v) p}
6 for each edge (u, v) in p
7 do f [u, v] = f [u, v] + cf (p)
8 f [v, u] = - f [u, v]
9
8/3/2019 Maxflow Problem
http://slidepdf.com/reader/full/maxflow-problem 10/73
The basic Ford Fulkersonaugmenting paths Algorithm
S t
v1
v2
v3
v4
12
14
(residual) network Gf 1 for each edge (u, v) E [G]
2 do f [u, v] = 0
3 f [v, u] = 0
4 while there exists a path p from s to t
in the residual network G f 5 do cf (p) = min{cf (u, v) | (u, v) p}
6 for each edge (u, v) in p
7 do f [u, v] = f [u, v] + cf (p)
8 f [v, u] = - f [u, v]
10
8/3/2019 Maxflow Problem
http://slidepdf.com/reader/full/maxflow-problem 11/73
The basic Ford Fulkersonaugmenting paths Algorithm
S t
v1
v2
v3
v4
12
14
(residual) network Gf
temporary variable:cf (p) = 12
1 for each edge (u, v) E [G]
2 do f [u, v] = 0
3 f [v, u] = 0
4 while there exists a path p from s to t
in the residual network G f 5 do cf (p) = min{cf (u, v) | (u, v) p}
6 for each edge (u, v) in p
7 do f [u, v] = f [u, v] + cf (p)
8 f [v, u] = - f [u, v]
11
8/3/2019 Maxflow Problem
http://slidepdf.com/reader/full/maxflow-problem 12/73
The basic Ford Fulkersonaugmenting paths Algorithm
S t
v1
v2
v3
v4
12
14
(residual) network Gf
temporary variable:cf (p) = 12
S t
v1
v2
v3
v4
12/12
14
new flow network
G
1 for each edge (u, v) E [G]
2 do f [u, v] = 0
3 f [v, u] = 0
4 while there exists a path p from s to t
in the residual network G f 5 do cf (p) = min{cf (u, v) | (u, v) p}
6 for each edge (u, v) in p
7 do f [u, v] = f [u, v] + cf (p)
8 f [v, u] = - f [u, v]
12
8/3/2019 Maxflow Problem
http://slidepdf.com/reader/full/maxflow-problem 13/73
The basic Ford Fulkersonaugmenting paths Algorithm
S t
v1
v2
v3
v4
12
14
(residual) networkGf
S t
v1
v2
v3
v4
12/12
14
new flow network G
1 for each edge (u, v) E [G]
2 do f [u, v] = 0
3 f [v, u] = 0
4 while there exists a path p from s to t
in the residual network G f 5 do cf (p) = min{cf (u, v) | (u, v) p}
6 for each edge (u, v) in p
7 do f [u, v] = f [u, v] + cf (p)
8 f [v, u] = - f [u, v]
13
8/3/2019 Maxflow Problem
http://slidepdf.com/reader/full/maxflow-problem 14/73
The basic Ford Fulkersonaugmenting paths Algorithm
S t
v1
v2
v3
v4
12
14
(residual) network Gf
S t
v1
v2
v3
v4
12/12
14
new flow network
G
1 for each edge (u, v) E [G]
2 do f [u, v] = 0
3 f [v, u] = 0
4 while there exists a path p from s to t
in the residual network G f 5 do cf (p) = min{cf (u, v) | (u, v) p}
6 for each edge (u, v) in p
7 do f [u, v] = f [u, v] + cf (p)
8 f [v, u] = - f [u, v]
14
8/3/2019 Maxflow Problem
http://slidepdf.com/reader/full/maxflow-problem 15/73
The basic Ford Fulkersonaugmenting paths Algorithm
S t
v1
v2
v3
v4
12
14
(residual) network Gf
S t
v1
v2
v3
v4
12/12
14
new flow network
G
1 for each edge (u, v) E [G]
2 do f [u, v] = 0
3 f [v, u] = 0
4 while there exists a path p from s to t
in the residual network G f 5 do cf (p) = min{cf (u, v) | (u, v) p}
6 for each edge (u, v) in p
7 do f [u, v] = f [u, v] + cf (p)
8 f [v, u] = - f [u, v]
15
8/3/2019 Maxflow Problem
http://slidepdf.com/reader/full/maxflow-problem 16/73
The basic Ford Fulkersonaugmenting paths Algorithm
(residual) networkGf
S t
v1
v2
v3
v4
12/12
14
new flow network
G
temporary variable:cf (p) = 4
S t
v1
v2
v3
v4
12
14
1 for each edge (u, v) E [G]
2 do f [u, v] = 0
3 f [v, u] = 0
4 while there exists a path p from s to t
in the residual network G f 5 do cf (p) = min{cf (u, v) | (u, v) p}
6 for each edge (u, v) in p
7 do f [u, v] = f [u, v] + cf (p)
8 f [v, u] = - f [u, v]
16
8/3/2019 Maxflow Problem
http://slidepdf.com/reader/full/maxflow-problem 17/73
The basic Ford Fulkersonaugmenting paths Algorithm
S t
v1
v2
v3
v4
12
14
(residual) networkGf
S t
v1
v2
v3
v4
12/12
new flow network
G temporary variable:cf (p) = 4
1 for each edge (u, v) E [G]
2 do f [u, v] = 0
3 f [v, u] = 0
4 while there exists a path p from s to t
in the residual network G f 5 do cf (p) = min{cf (u, v) | (u, v) p}
6 for each edge (u, v) in p
7 do f [u, v] = f [u, v] + cf (p)
8 f [v, u] = - f [u, v]
17
8/3/2019 Maxflow Problem
http://slidepdf.com/reader/full/maxflow-problem 18/73
The basic Ford Fulkersonaugmenting paths Algorithm
S t
v1
v2
v3
v4
12
10
(residual) network Gf
S t
v1
v2
v3
v4
12/12
new flow network
G
4
1 for each edge (u, v) E [G]
2 do f [u, v] = 0
3 f [v, u] = 0
4 while there exists a path p from s to t
in the residual network G f 5 do cf (p) = min{cf (u, v) | (u, v) p}
6 for each edge (u, v) in p
7 do f [u, v] = f [u, v] + cf (p)
8 f [v, u] = - f [u, v]
18
8/3/2019 Maxflow Problem
http://slidepdf.com/reader/full/maxflow-problem 19/73
The basic Ford Fulkersonaugmenting paths Algorithm
S t
v1
v2
v3
v4
12
10
(residual) network Gf
S t
v1
v2
v3
v4
12/12
new flow network
G
4
1 for each edge (u, v) E [G]
2 do f [u, v] = 0
3 f [v, u] = 0
4 while there exists a path p from s to t
in the residual network G f 5 do cf (p) = min{cf (u, v) | (u, v) p}
6 for each edge (u, v) in p
7 do f [u, v] = f [u, v] + cf (p)
8 f [v, u] = - f [u, v]
19
8/3/2019 Maxflow Problem
http://slidepdf.com/reader/full/maxflow-problem 20/73
The basic Ford Fulkersonaugmenting paths Algorithm
S t
v1
v2
v3
v4
12
10
(residual) networkGf
S t
v1
v2
v3
v4
12/12
new flow network
G
4
1 for each edge (u, v) E [G]
2 do f [u, v] = 0
3 f [v, u] = 0
4 while there exists a path p from s to t
in the residual network G f 5 do cf (p) = min{cf (u, v) | (u, v) p}
6 for each edge (u, v) in p
7 do f [u, v] = f [u, v] + cf (p)
8 f [v, u] = - f [u, v]
20
8/3/2019 Maxflow Problem
http://slidepdf.com/reader/full/maxflow-problem 21/73
The basic Ford Fulkersonaugmenting paths Algorithm
S t
v1
v2
v3
v4
12
10
(residual) networkGf
S t
v1
v2
v3
v4
12/12
new flow network
G
4
temporary variable:cf (p) = 7
1 for each edge (u, v) E [G]
2 do f [u, v] = 0
3 f [v, u] = 0
4 while there exists a path p from s to t
in the residual network G f 5 do cf (p) = min{cf (u, v) | (u, v) p}
6 for each edge (u, v) in p
7 do f [u, v] = f [u, v] + cf (p)
8 f [v, u] = - f [u, v]
21
8/3/2019 Maxflow Problem
http://slidepdf.com/reader/full/maxflow-problem 22/73
The basic Ford Fulkersonaugmenting paths Algorithm
S t
v1
v2
v3
v4
12
10
(residual) networkGf
S t
v1
v2
v3
v4
12/12
new flow network
G
4
temporary variable:cf (p) = 7
1 for each edge (u, v) E [G]
2 do f [u, v] = 0
3 f [v, u] = 0
4 while there exists a path p from s to t
in the residual network G f 5 do cf (p) = min{cf (u, v) | (u, v) p}
6 for each edge (u, v) in p
7 do f [u, v] = f [u, v] + cf (p)
8 f [v, u] = - f [u, v]
22
8/3/2019 Maxflow Problem
http://slidepdf.com/reader/full/maxflow-problem 23/73
The basic Ford Fulkersonaugmenting paths Algorithm
(residual) networkGf
S t
v1
v2
v3
v4
12/12
new flow network
G
S t
v1
v2
v3
v4
12
3
11
1 for each edge (u, v) E [G]
2 do f [u, v] = 0
3 f [v, u] = 0
4 while there exists a path p from s to t
in the residual network G f 5 do cf (p) = min{cf (u, v) | (u, v) p}
6 for each edge (u, v) in p
7 do f [u, v] = f [u, v] + cf (p)
8 f [v, u] = - f [u, v]
23
8/3/2019 Maxflow Problem
http://slidepdf.com/reader/full/maxflow-problem 24/73
The basic Ford Fulkersonaugmenting paths Algorithm
(residual) network Gf
S t
v1
v2
v3
v4
12/12new flow network
G
S t
v1
v2
v3
v4
12
3
11
Finally we have:| f | = f (s, V) = 23
1 for each edge (u, v) E [G]
2 do f [u, v] = 0
3 f [v, u] = 0
4 while there exists a path p from s to t
in the residual network G f
5 do cf (p) = min{cf (u, v) | (u, v) p}
6 for each edge (u, v) in p
7 do f [u, v] = f [u, v] + cf (p)
8 f [v, u] = - f [u, v]
24
8/3/2019 Maxflow Problem
http://slidepdf.com/reader/full/maxflow-problem 25/73
Analysis of the Ford Fulkerson algorithm
The running time depends on how the augmenting path p in line 4 isdetermined.
1 for each edge (u, v) E [G]
2 do f [u, v] = 0
3 f [v, u] = 0
4 while there exists a path p from s to t
in the residual network G f 5 do cf (p) = min{cf (u, v) | (u, v) p}
6 for each edge (u, v) in p
7 do f [u, v] = f [u, v] + cf (p)
8 f [v, u] = - f [u, v]
25
8/3/2019 Maxflow Problem
http://slidepdf.com/reader/full/maxflow-problem 26/73
Analysis of the Ford Fulkerson algorithm
O(|E|)
O(|E|) O(|E||fmax|)
running time: O ( |E| |fmax|)with fmax as maximumflow
1 for each edge (u, v) E [G]
2 do f [u, v] = 0
3 f [v, u] = 0
4 while there exists a path p from s to t
in the residual network G f 5 do cf (p) = min{cf (u, v) | (u, v) p}
6 for each edge (u, v) in p
7 do f [u, v] = f [u, v] + cf (p)
8 f [v, u] = - f [u, v]
O(|E|)
26
8/3/2019 Maxflow Problem
http://slidepdf.com/reader/full/maxflow-problem 27/73
Ford Fulkerson – cuts of flow networks
A cut (S,T) of a flow network G=(V,E) is a partiton of V into Sand T = V \ S such that s S and t T.
S t
v1
v2
v3
v4
12/12
11/14
TS
For example:S = {s,v1,v2) , T = {v3,v4,t}Net flow f(S ,T) = f(v1,v3) + f(v2,v4) + f(v2,v3)
= 12 + 11 + (-0) = 23
Max Flow = Min Cut = 23
27
8/3/2019 Maxflow Problem
http://slidepdf.com/reader/full/maxflow-problem 28/73
Edmonds-Karp Algorithm
Edmonds-Karp = Ford-Fulkerson using breadth-first search inline 4 to find an augmenting path,i.e., the augmenting path found is shortest in terms of thenumber of edges.
Edmonds-Karp runs in O(VE2) time… 28
8/3/2019 Maxflow Problem
http://slidepdf.com/reader/full/maxflow-problem 29/73
Analysis of Edmonds-Karp
If the Edmonds-Karp algorithm is run on a flow networkG = (V, E) with source s and sink t, then the total numberof flow augmentations performed by the algorithm is
O(VE).
Since, breadth-first search can be implemented in O(E)time, the running time of Edmonds-Karp is O(VE * E) =
O(VE2).
29
8/3/2019 Maxflow Problem
http://slidepdf.com/reader/full/maxflow-problem 30/73
A BFS Example
g
s
d
b
f
e
c
a
L = s 0
30
8/3/2019 Maxflow Problem
http://slidepdf.com/reader/full/maxflow-problem 31/73
s
d
b
g
f
e
c
a
L = s 0
L = a, c 1
A BFS Example
31
8/3/2019 Maxflow Problem
http://slidepdf.com/reader/full/maxflow-problem 32/73
s
d
b
g
f
e
c
a
L = s 0
L = a, c 1
L = d, e, f 2
A BFS Example
32
8/3/2019 Maxflow Problem
http://slidepdf.com/reader/full/maxflow-problem 33/73
s
d
b
g
f
e
c
a
L = s 0
L = a, c 1
L = d, e, f 2
L = b , g 3
A BFS Example
33
8/3/2019 Maxflow Problem
http://slidepdf.com/reader/full/maxflow-problem 34/73
s
d
b
g
f
e
c
a
L = s 0
L = a, c 1
L = d, e, f 2
L = b , g 3
A BFS Example
34
8/3/2019 Maxflow Problem
http://slidepdf.com/reader/full/maxflow-problem 35/73
s
d
b
g
f
e
c
a
0
1
1
2
2
2
3
3
d (b ): shortest distance from s to b
A BFS Example
35
8/3/2019 Maxflow Problem
http://slidepdf.com/reader/full/maxflow-problem 36/73
History
36
Discoverer Method Big-Oh Year Ford, Fulkerson Augmenting path mF*1955
Edmonds-Karp Shortest path m2n1970
Dinitz Shortest path mn2 1970
Edmonds-Karp, Dinitz Capacity scaling m2 logn1972
Dinitz-Gabow Capacity scaling mn logn1973
Karzanov Preflow-push n3 1974
Sleator-Tarjan Dynamic trees mn log n1983
Goldberg-Tarjan FIFO preflow-push mn log (n2 / m)1986
. . . . . . . . .. . .
Goldberg-Rao Length functionm3/2 log (n2 / m) log n
mn2/3 log (n2 / m) log n1997
8/3/2019 Maxflow Problem
http://slidepdf.com/reader/full/maxflow-problem 37/73
CHINESE POSTMAN ALGORITHM
• This is used when we are trying to find the shortest
route round a network such all the arcs are travelled
along at least once.
37
8/3/2019 Maxflow Problem
http://slidepdf.com/reader/full/maxflow-problem 38/73
Edmond’s Algorithm
procedure FloydWarshallWithPathReconstruction ()
for k := 1 to n
for i := 1 to n
for j := 1 to nif path[i][k] + path[k][j] < path[i][j] then
path[i][j] := path[i][k]+path[k][j];
next[i][j] := k;
Run time = O(n 3 )
38
Add thi t th t k
8/3/2019 Maxflow Problem
http://slidepdf.com/reader/full/maxflow-problem 39/73
A postman starts at the depot T and has to deliver mailalong all the streets shown in the network below. Find aroute that means he travels the shortest possible distance.
T
HG
FE
DCB
A
0.8
1.7
1.01.0
2.0 0.750.2
0.4 0.6
0.5 1.5
1.70.6
0.9
1.6
0.9
0.3
Odd vertices are B, G, H & T Possible pairings:
BG & HT
Route: BEG & HECT
Length: 1.7 + 1.8 = 3.5
BH & GT Route: BEH & GECT
Length: 1.5 + 2.0 = 3.5
BT & GH
Route: BCT & GH
Length: 0.8 + 0.9 = 1.7
The shortest extra route is:
Add this to the network
Possible route is:
TABEAGHEGHFDFECDTCBCT
39
T
8/3/2019 Maxflow Problem
http://slidepdf.com/reader/full/maxflow-problem 40/73
A
B C
D
EF
GH
Route is:
T A B E A G H E G H F D F E C D T C B C T
Total length of alloriginal paths is16.45 km
Add on the extraroutes BCT and GH= 1.7 km
Shortest length =16.45 + 1.7 = 18.15 km
40
8/3/2019 Maxflow Problem
http://slidepdf.com/reader/full/maxflow-problem 41/73
Topological Sort
it should be possible to perform all the tasks insuch an order as to satisfy all of thedependencies
Such as order is called a topological sort of the
vertices
The requirement is to list all of the vertices suchthat in that list, vi precedes v j whenever a path
exists from vi to v j
If this is not possible, this would imply the existenceof a cycle.
Theorem:41
8/3/2019 Maxflow Problem
http://slidepdf.com/reader/full/maxflow-problem 42/73
Topological Sort
int topologicalOrderTraversal( ){
int numVisitedVertices = 0;
while(there are more vertices to be visited){
if(there is no vertex with in-degree 0)
break;else{
select a vertex v that has in-degree 0;
visit v;
numVisitedVertices++;
delete v and all its emanating edges;
}}
return numVisitedVertices;
} Running time: O(V + E) =O(n)
42
8/3/2019 Maxflow Problem
http://slidepdf.com/reader/full/maxflow-problem 43/73
Example
Let us define the table of in-degrees
VertexIn-
degree
1 02 1
3 3
4 3
5 2
6 0
43
8/3/2019 Maxflow Problem
http://slidepdf.com/reader/full/maxflow-problem 44/73
Example
And a queue into which we can insert vertices 1and 6
1 6
Queue
VertexIn-
degree
1 02 1
3 3
4 3
5 2
6 0
44
8/3/2019 Maxflow Problem
http://slidepdf.com/reader/full/maxflow-problem 45/73
Example
We dequeue the head (1), decrement the in-degree of all adjacent vertices: 2 and 4
2 is decremented to zero: enqueue 2
6 2
QueueSort1
VertexIn-
degree
1 02 0
3 3
4 2
5 2
6 0
45
8/3/2019 Maxflow Problem
http://slidepdf.com/reader/full/maxflow-problem 46/73
Example
We dequeue 6 and decrement the in-degree of alladjacent vertices
Neither is decremented to zero
2
QueueSort1, 6
VertexIn-
degree
1 02 0
3 2
4 2
5 1
6 0
46
8/3/2019 Maxflow Problem
http://slidepdf.com/reader/full/maxflow-problem 47/73
Example
We dequeue 2, decrement, and enqueue vertex 5
5
QueueSort1, 6, 2
VertexIn-
degree
1 02 0
3 1
4 1
5 0
6 0
47
8/3/2019 Maxflow Problem
http://slidepdf.com/reader/full/maxflow-problem 48/73
Example
We dequeue 5, decrement, and enqueue vertex 3
3
QueueSort1, 6, 2, 5
VertexIn-
degree
1 02 0
3 0
4 1
5 0
6 0
48
8/3/2019 Maxflow Problem
http://slidepdf.com/reader/full/maxflow-problem 49/73
Example
We dequeue 3, decrement 4, and add 4 to thequeue
4
QueueSort1, 6, 2, 5, 3
VertexIn-
degree
1 02 0
3 0
4 0
5 0
6 0
49
8/3/2019 Maxflow Problem
http://slidepdf.com/reader/full/maxflow-problem 50/73
Example
We dequeue 4, there are no adjacent vertices todecrement the in degree
QueueSort1, 6, 2, 5, 3, 4
VertexIn-
degree
1 02 0
3 0
4 0
5 0
6 0
50
8/3/2019 Maxflow Problem
http://slidepdf.com/reader/full/maxflow-problem 51/73
Example
The queue is now empty, so a topological sort is1, 6, 2, 5, 3, 4
51
8/3/2019 Maxflow Problem
http://slidepdf.com/reader/full/maxflow-problem 52/73
Planar Graphs
Can be drawn on the plane without crossings
We will :
Testing if a graph is planarFinding a nice drawing of a planar graph
52
8/3/2019 Maxflow Problem
http://slidepdf.com/reader/full/maxflow-problem 53/73
Testing if a graph is planar
Euler’s Formula
- region : is a part of the plane completelydisconnected off from other parts of the planeby the edges of the graph.
r = |E | - |V | + 2
c = r - |E | + |V | c must equal 2 for planar graphs.
53
8/3/2019 Maxflow Problem
http://slidepdf.com/reader/full/maxflow-problem 54/73
54
Face-Edge Handshaking
DEF: The degree of a region F is the number ofedges at its boundary, and is denoted by deg(F ).
THM: Let G be a planar graph with region set R .
Then:
RF
F E )deg(21 ||
Ex: (4+4+4+4+4+4)/2 = 12
8/3/2019 Maxflow Problem
http://slidepdf.com/reader/full/maxflow-problem 55/73
55
Proving that Q4 is Non-Planar
Now we have enough invariants to prove thatthe 4-cube is non-planar.
1) Count the number of vertices and edges:
|V | = 16 (twice the number for 3-cube)
|E | = 32 (twice the number for 3-cube plusnumber of vertices in 3-cube)
2) Suppose 4-cube were planar so by Euler’sformula number of regions would be:
r = 32-16+2=18
8/3/2019 Maxflow Problem
http://slidepdf.com/reader/full/maxflow-problem 56/73
56
Proving that Q4 is Non-Planar
3) Calculate the degree of each region
In our case, this give |E | ½·18·4=36
contradicting |E | = 32 !
Thus 4-cube cannot be planar. •
RF
F E )deg(2
1 ||
8/3/2019 Maxflow Problem
http://slidepdf.com/reader/full/maxflow-problem 57/73
Testing if a graph is planar
Circle-Chord Method
Step One: Find a circuit that contains all thevertices of the graph.
Step Two: Draw this circuit as a large circle.
a b c d
e hgfgh
f
bc
e
d
a
57
8/3/2019 Maxflow Problem
http://slidepdf.com/reader/full/maxflow-problem 58/73
Testing if a graph is planarStep Three: Choose one chord, and decide to draw it either inside or outsidethe circle. If chosen correctly, it will force certain other chords to be drawnopposite to the circle. (Inside if the first chord was drawn outside, and viceversa.)
gh
f
bc
e
d
a
Since the chords could be drawn withoutcrossing, this graph is planar.
a b c d
e hgf
58
8/3/2019 Maxflow Problem
http://slidepdf.com/reader/full/maxflow-problem 59/73
a
e
f d
c
bh
g
a
f
e
d
c
bh
g
Testing if a graph is planar
• Kuratowski theoremA graph is planar if and only if it does not contain the K5
or the K3,3
59
S i Al ith
8/3/2019 Maxflow Problem
http://slidepdf.com/reader/full/maxflow-problem 60/73
Sugiyama Algorithm
Remove cycles
Generate levelling
Sugiyama 1979
Running time O(|V||E|\log|E)
11
3
8
1
5
10 7
4
12
6
9
2
60
S i Al ith
8/3/2019 Maxflow Problem
http://slidepdf.com/reader/full/maxflow-problem 61/73
Sugiyama Algorithm
Algorithm
Remove cycles
Generate levelling
11
85
10 7
12
6
9
31 421
61
Sugiyama Algorithm
8/3/2019 Maxflow Problem
http://slidepdf.com/reader/full/maxflow-problem 62/73
Sugiyama Algorithm
Algorithm
Remove cycles
Generate levelling
11
8 10
129
31
5 7
4
6
21
2
62
Sugiyama Algorithm
8/3/2019 Maxflow Problem
http://slidepdf.com/reader/full/maxflow-problem 63/73
Sugiyama Algorithm
Algorithm
Remove cycles
Generate levelling
1112
3
8
1
5
10
7
42
9
6
1
2
3
63
Sugiyama Algorithm
8/3/2019 Maxflow Problem
http://slidepdf.com/reader/full/maxflow-problem 64/73
Sugiyama Algorithm
Algorithm
Remove cycles
Generate levelling
Split long edges
11
3
8 9
1
5
10
7
42
12
6
1
2
3
4
64
Sugiyama Algorithm
8/3/2019 Maxflow Problem
http://slidepdf.com/reader/full/maxflow-problem 65/73
Sugiyama Algorithm
Algorithm
Remove cycles
Generate levelling
Split long edges
Reduce crossings11
6
3
8 9
1
5
10
7
42
12
1
2
3
4
65
Sugiyama Algorithm
8/3/2019 Maxflow Problem
http://slidepdf.com/reader/full/maxflow-problem 66/73
Sugiyama Algorithm
Algorithm
Remove cycles
Generate levelling
Split long edges
Reduce crossings11
6
3
8 9
1
5
10
7
42
12
1
2
3
4
66
Sugiyama Algorithm
8/3/2019 Maxflow Problem
http://slidepdf.com/reader/full/maxflow-problem 67/73
Sugiyama Algorithm
Algorithm
Remove cycles
Generate levelling
Split long edges
Reduce crossings11
6
3
89
1
5
10
7
42
12
1
2
3
4
67
Sugiyama Algorithm
8/3/2019 Maxflow Problem
http://slidepdf.com/reader/full/maxflow-problem 68/73
Sugiyama Algorithm
Algorithm
Remove cycles
Generate levelling
Split long edges
Reduce crossings11
6
3
89
1
5
10
7
42
12
1
2
3
4
68
Sugiyama Algorithm
8/3/2019 Maxflow Problem
http://slidepdf.com/reader/full/maxflow-problem 69/73
Sugiyama Algorithm
Algorithm
Remove cycles Generate levelling
Split long edges
Reduce crossings11
6
3
89
1
5
10
7
42
12
1
2
3
4
69
Sugiyama Algorithm
8/3/2019 Maxflow Problem
http://slidepdf.com/reader/full/maxflow-problem 70/73
Sug ya a go t
Algorithm
Remove cycles Generate levelling
Split long edges
Reduce crossings11
6
3
89
1
5
10
7
42
12
1
2
3
4
70
Sugiyama Algorithm
8/3/2019 Maxflow Problem
http://slidepdf.com/reader/full/maxflow-problem 71/73
g y g
Algorithm
Remove cycles Generate levelling
Split long edges
Reduce crossings11
6
3
89
1
5
10
7
42
12
1
2
3
4
71
Sugiyama Algorithm
8/3/2019 Maxflow Problem
http://slidepdf.com/reader/full/maxflow-problem 72/73
g y g
Algorithm
Remove cycles Generate levelling
Split long edges
Reduce crossings
Join long edges 11
6
3
89
1
5
10
7
42
12
1
2
3
4
72
Sugiyama Algorithm
8/3/2019 Maxflow Problem
http://slidepdf.com/reader/full/maxflow-problem 73/73
g y g
Algorithm
Remove cycles Generate levelling
Split long edges
Reduce crossings
Join long edges
Basic properties
2 level crossing minimisation
Minimum edge set whoseremoval eliminates crossings
11
6
3
89
1 42
10
12
57
1
2
3
4
![MaxFlow Revisited: An Empirical Comparison of Maxflow ...dbatra/papers/vb_bmvc12.pdfVERMA, BATRA: MAXFLOW REVISITED 3. compared BK, PRL and HPF but not IBFS. Goldberg et al. [21]](https://static.documents.pub/doc/80x56/612637b918708e260d7c059c/maxflow-revisited-an-empirical-comparison-of-maxiow-dbatrapapersvbbmvc12pdf.jpg)
