Branch and Bound

( unit-v Chapter -2 )

 Backtracking is effective for subset or permutation

problems.

 Backtracking is not good for optimization problems.

 This drawback is rectified by using Branch And Bound

technique.
 Branch And Bound is applicable for only optimization
problems and will apply when greedy and dynamic
programming fails.

 Branch and Bound also uses bounding function similar to

backtracking.
 Branch and Bound
General Method:
 Here we create a state space tree and all the children of
the E-node( the live node whose children are currently being
generated) will get generated before any other live node
becomes the E-node.
For exploring the new nodes we can use either BFS or DFS
approach
The Branch and Bound technique which uses BFS (Queue)
is called as FIFO Branch and Bound.
The Branch and Bound technique which uses DFS (Stack)
is called as LIFO Branch and Bound.
Bounding functions are used to avoid the generation of the
sub tree which do not lead us to answer node.
 Terminology of tree organization

State Space :-
All paths from the root to othernodes
define the state space of the problem.

Solution Space :-
All paths from the root to solution states
define the solution space of the
problem.
 Ex :- 1
X1=1
X1=4
X1=2 X1=3

2 34 50
18

x 2= 2 3 1 3 1 2 4 1 2 3
4 4

3 8 13 19 24 29 35 40 45 51 56 61
x3=
3 4 2 4 2 3 3 4 1 4 1 3 2 4 1 4 1 2 2 3 1 3 1 2

4 6 9 11 14 16 20 22 25 27 30 32 36 38 41 43 46 48 52 54 57 59 62 64
x4=
4 3 4 2 3 2 4 3 4 1 3 1 4 2 4 1 2 1 3 2 3 1 2 1

5 7 10 12 15 17 21 23 26 28 31 33 37 39 42 44 47 9 53 55 58 60 63 65

Tree organization of the 4-queens solution space. Nodes are numbered as in depth
first search.
Problem state :-
Each node in the tree is a problem
state.
Ex:-1 2 18
and so on
Solution States :-
These are those problem
states S for which the path from the root
to S define a tuple in the solution space.

Ex:-
5 7 10 12 15 17 21 and so on
– A node which has been generated and all of whose
children have not yet been generated is called a live
node.

– The live node whose children are currently being

generated is called E-node ( Expanding node ).

– A dead node is a generated node which can not to be

expanded further or all of whose children have been
generated.
 Branch And Bound
Nodes will be expanded in three ways.
• Search the tree using a breadth-first search
(FIFO branch and bound).

• Search the tree as in a BFS, but replace the FIFO queue

with a stack (LIFO branch and bound).

• Replace the FIFO queue with a priority queue

(Least-Cost (or max priority) branch and bound).
• Least-cost branch and bound directs the search
to the parts which most likely to contain the answer.
Traveling Salesperson Problem
Let G=(V,E) be a directed graph defining
an instance of the travelling salesperson
problem.
Let cij be the cost of the edge (i,j) and
cij=∞ if (i,j)€│E(g) and let │V│=n.
Assume that the tour starts and ends at 1.
To use least cost branch and bound to
search the travelling salesperson state
space tree,we must definr a cost function
c(x) and two other functions
c^(x) <=c(x)<=u^(x).
Reduced Cost Matrix
• A Row or Column is said to be reduced if it
contains at least one zero and all remaining
entries are non-negative .
• A matrix is reduced iff every row and column is
reduced.
- if a constant t is chosen to be minimum entry in
row i (column j) will introduce a zero into a row i
(column j).
- The total amount subtracted from the columns
and rows is lower bound on the length of a
minimum cost tour and can be used as the c^(x)
value for the root of state space tree.
With every node in the state space tree,
we associate a reduced cost matrix.
Let A be the reduced cost matrix for
node R. Let S be the child of r such that
the edge (R,S) corresponds to including
edge (i,j) in the tour.if S is not a leaf
node then the reduced cost matrix for
node S can be obtained as follows.
1. Change all entries in row i column j of
A to ∞.
2. Set A(i,j) to ∞.
3. Apply row reduction and column
reduction except for rows and
columns containing ∞.
4. The total cost for node S can be
calculated as,
c^(S) =c^(R)+A(i,j)+r.
where r is the total amount
substracted in step3.
 Travelling Sales persons Problem:

∞ 20 30 10
11
G= 15 ∞ 16 4 2
3 5 ∞ 2 4
19 6 18 ∞ 3
16 4 7 16 ∞

Row minimization:
• find the min value in each row and subtract from the entire row values :

Min values
∞ 20 30 10 11 10 ∞ 10 20 0 1
15 ∞ 16 4 2 2 13 ∞ 14 2 0
3 5 ∞ 2 4 2
3
 1 3 ∞ 0 2
19 6 18 ∞ 3 16 3 15 ∞ 0
16 4 7 16 ∞ 4 12 0 3 12
21 ∞
 Column minimization:

∞ 10 20 0 1
13 ∞ 14 2 0
1 3 ∞ 0 2
16 3 15 ∞ 0
12 0 3 12 ∞
1 0 3 0 0 =4

After Full Reduction :

∞ 20 30 10 ∞ 10 17 0 1
11 12 ∞ 11 2 0
15 ∞ 16 4 2  0 3 ∞ 0 2
3 5 ∞ 2 4 15 3 12 ∞ 0
Full reduction 11 0 0 12 ∞
19 6 18 ∞ 3
16 4 7 16 ∞

Total minimization is 21+4 = 25

Steps in dynamic reduction technique:

Draw the state space tree at the root with the optimum cost.
Calculate the cost matrix for i -> j path by making ith row and jth column
entries as ∞ . And also make M[j][i] as ∞
Calculate the Cost which includes (i->j) path is sum of optimum cost,
reduced cost , M[i][j].

a) Path from 1 to 2: so make 1st row and 2nd column as ∞, and set M[2][1]
as ∞
25
∞ 10 17 0 1 ∞ ∞ ∞ ∞ ∞
12 ∞ 11 2 0 ∞ ∞ 11 2 0 1
0 3 ∞ 0 2  0 ∞ ∞ 0 2
35
15 3 12 ∞ 0 15 ∞ 12 ∞ 0
11 0 0 12 ∞ 11 ∞ 0 12 ∞ 2

Cost = 25 + 0 + 10 = 35
Optimum reduced old value of
cost cost M[1][2]
b) Path from 1 to 3: so make 1st row and 3rd column as ∞, and set M[3][1] as ∞

25

∞ 10 17 0 1 ∞ ∞ ∞ ∞ ∞ 1
12 ∞ 11 2 0 12 ∞ ∞ 2 0
0 3 ∞ 0 2  ∞ 3
15 3
∞ 0
∞ ∞
2
0
35 53
15 3 12 ∞ 0 2 3
11 0 0 12 ∞ 11 0 ∞ 12 ∞
11

Cost = 25 + 11 + 17 = 53
Optimum reduced old value of
cost cost M[1][3]
c) Path from 1 to 4: so make 1st row and 4th column as ∞, and set M[4][1] as ∞

25

∞ 10 17 0 1 ∞ ∞ ∞ ∞ ∞ 1
12 ∞ 11 2 0 12 ∞ 11 ∞ 0
0 3 ∞ 0 2  0
∞
3 ∞
3 12
∞
∞
2
0
35 53 25
15 3 12 ∞ 0
11 0 0 12 ∞ 11 0 0 ∞ ∞ 2 3 4

Cost = 25 + 0 + 0 = 25
Optimum reduced old value of
cost cost M[1][4]
d) Path from 1 to 5: so make 1st row and 5th column as ∞, and set M[5][1] as ∞

25

∞ 10 17 0 1 ∞ ∞ ∞ ∞ ∞ 1
12 ∞ 11 2 0 12 ∞ 11 2 ∞ 2
0 3 ∞ 0 2  0 3
15 3
∞
12
0
∞
∞
∞
35 53 25 31
15 3 12 ∞ 0 3
11 0 0 12 ∞ ∞ 0 0 12 ∞ 2 3 4 5
5

Cost = 25 + 5 + 1 = 31
Optimum reduced old value of
cost cost M[1][4]

As the optimum cost we got at node no 4, so the next E-node will be

node no 4. now we explore the node 4 by taking the reduced matrix
that we got in the process of (1->4)
e) Path from 1 to 4 to 2: so make 1st and 4th row as ∞ and 4th and 2nd column as ∞,
and set M[2][1] as ∞
25

1
∞ ∞ ∞ ∞ ∞ ∞ ∞ ∞ ∞ ∞
12 ∞ 11 ∞ 0 ∞ ∞ 11 ∞ 0 35 53
0 3 ∞ ∞ 2  0 ∞ ∞ ∞ 2
2
25 31
∞ 3 12 ∞ 0 ∞ ∞ ∞ ∞ ∞ 3 4 5
11 0 0 ∞ ∞ 11 ∞ 0 ∞ ∞
28

2
Cost = 25 + 0 + 3 = 28
Optimum reduced old value of
cost cost M[4][2]
f) Path from 1 to 4 to 3: so make 1st and 4th row as ∞ and 4th and 3rd column as ∞,
and set M[3][1] as ∞
25

1
∞ ∞ ∞ ∞ ∞ ∞ ∞ ∞ ∞ ∞
12 ∞ 11 ∞ 0 12 ∞ ∞ ∞ 0 35 53
0 3 ∞ ∞ 2  ∞ 3 ∞ ∞ 2 2
2
25 31
∞ 3 12 ∞ 0 ∞ ∞ ∞ ∞ ∞ 3 4 5
11 0 0 ∞ ∞ 11 0 ∞ ∞ ∞
28 50
11 13
2 3
Cost = 25 + 13 + 12 = 50
Optimum reduced old value of
cost cost M[4][3]
g) Path from 1 to 4 to 5: so make 1st and 4th row as ∞ and 4th and 5th column as ∞,
and set M[5][1] as ∞
25

1
∞ ∞ ∞ ∞ ∞ ∞ ∞ ∞ ∞ ∞
12 ∞ 11 ∞ 0 12 ∞ 11 ∞ ∞ 11 35 53
0 3 ∞ ∞ 2  0 3 ∞ ∞ ∞
2
25 31
∞ 3 12 ∞ 0 ∞ ∞ ∞ ∞ ∞ 3 4 5
11 0 0 ∞ ∞ ∞ 0 0 ∞ ∞
28 50
11 36
2 3 5
Cost = 25 + 11 + 0 = 36
Optimum reduced old value of
cost cost M[4][5]

As the optimum cost we got is at node no 2, so the next E-node will

be node no 2. now we explore the node 2 by taking the reduced
matrix that we got in the process of (1->4->2)
h) Path from 1 to 4 to 2 to 3: so make 1st , 4th , 2nd rows as ∞ and 4th , 2nd ,3rd column
as ∞, and set M[3][1] as ∞
25

∞ ∞ ∞ ∞ ∞ ∞ ∞ ∞ ∞ ∞
1
∞ ∞ 11 ∞ 0 ∞ ∞ ∞ ∞ ∞
35
0
∞
∞ ∞
∞ ∞
∞
∞
2
∞
 ∞
∞
∞
∞
∞
∞
∞
∞
2
∞
2 53 25 31
2 3 4 5
11 ∞ 0 ∞ ∞ 11 ∞ ∞ ∞ ∞ 11

13 28 50 36
2 3 5
Cost = 28 + 13 + 11 = 52 52
Optimum reduced old value of
cost cost M[2][3] 3
i) Path from 1 to 4 to 2 to 5: so make 1st , 4th , 2nd rows as ∞ and 4th ,2nd , 5th column as
∞, and set M[5][1] as ∞
25

∞ ∞ ∞ ∞ ∞ ∞ ∞ ∞ ∞ ∞
1
∞ ∞ 11 ∞ 0 ∞ ∞ ∞ ∞ ∞
35
0
∞
∞ ∞
∞ ∞
∞
∞
2
∞
 0
∞
∞
∞
∞
∞
∞
∞
∞
∞
53 25 31
2 3 4 5
11 ∞ 0 ∞ ∞ ∞ ∞ 0 ∞ ∞

28 50 36
2 3 5
Cost = 28 + 0 + 0 = 28 52 28
Optimum reduced old value of
cost cost M[2][5] 3 5

As the optimum cost we got is at node no 5, so the next E-node will

be node no 5. we need to explore the node. But node 5 will get only
option of going to node 3. so here no need to do any calculations.
Hence the optimal path is 1->4 -> 2 -> 5 -> 3 -> 1
 0/1 knapsack problem

Nodes will be expanded in three ways.

• Search the tree using a breadth-first search
(FIFO branch and bound).

• Search the tree as in a BFS, but replace the FIFO queue

with a stack (LIFO branch and bound).

• Replace the FIFO queue with a priority queue

(Least-Cost (or max priority) branch and bound).
 0/1 knapsack problem ^

LC Branch and Bound Solution: Cost Function= c

Upper number = u
EX:- n=4, ( p1,p2,p3,p4 )= (10,10,12,18 ) 32
( w1,w2,w3,w4 ) = ( 2, 4, 6, 9 ), m=15 1 38

^ 32 2 22
Process: The calculation of U and C is as 3 32
follows. 38

U(1) - Scan through the objects from left to 32

right and put into the knapsack until 38 4 5 22
the first object that does not fit is 36
encontered.
32 6 38
^ 38 7 38
C(1) – Similar to U(1) except that it also
considers a fraction of the first object that
does not fit the knapsack.
38 8 9 20
38 20
Answer node
Continue this process until an answer node is
found.
FIFO Branch and Bound EX:- n=4, ( p1,p2,p3,p4 )= (10,10,12,18 )
( w1,w2,w3,w4 ) = ( 2,4,6,9 ), m=15
^
32
38
1
Cost Function = c
32 22 Upper Bound = u
2
38 3 32

32
22
38 4 5 36 6 22 7 30
32 30

32
8 38
38 9 38

50 32
50 10 11 12 38 13 20
32 38 20
Answer
node
EX:- n=4, ( p1,p2,p3,p4 )= (45,30,45,10 )
( w1,w2,w3,w4 ) = ( 3,5,9,5 ), m=16
C=75 w=8
u=115
1
X1 =1
X1 =0
75 w=8 75 w=14
115 2 3 79

X2 =1 X2 =0

75 w=8
90 w=12
115 4 5 98
X3 =1
X3 =0 X3 =1 X3 =0

120 w=17 75 w=8 90 w=12 45 w=3

115 85
8 9 55
98
6 7

100 w=17 90 w=12

8 100 9 90
Answer node
 Algorithm UBound(cp,cw,k,m)
// cp is current total profit, cw is the current weight total, k is the index of the
last removed item, m is the knapsack size.

{
b=cp;
fori=k+1 to n do
{
if(c+w[i]<=m) then
{

c=c+w[i];
b=b-p[i];
}
}
return b;
}
Home Work :-
1. Draw the portion of the state space tree generated by
LCBB and FIFIBB for the following knapsack problem.

a) n=5, ( p1,p2,p3,p4,p5 ) = ( 10,15,6,8,4 ),

( w1,w2,w3,w4,w5 )= ( 4,6,3,4,2 ) and
m=12

b) n=5, ( p1,p2,p3,p4,p5 )=
( w1,w2,w3,w4,w5 )=
( 4,4,5,8,9 ) and m=15.