Lecture On: Branch and Bound

You might also like

Download as pptx, pdf, or txt
Download as pptx, pdf, or txt
You are on page 1of 19

Lecture on

Branch and Bound


V.Radhesyam
Assistant Professor
Department of IT,VRSEC
Branch & bound
• Branch & bound is a design technique used to
solve optimization problems.
• It is of 3 types: FIFO Branch & bound , LIFO
Branch & bound and LC Branch & bound
• In both FIFO and LIFO selection for next E-node is
rigid and not useful to get answer node quickly.
• In LC Branch & bound search for answer node is
done by ranking function. LC Search
• C^(x)=f[h(x)]+ g^ (x)
• h cost of reaching x from root
• g^ (x) estimation of additional effort
• f is non decreasing function
TSP by LCBB
Row and Column Minimum
State-space Tree
Node-2 Value
Make Row 1 and
Column 2 as Infinity

A(2,1) also as
infinity
Nodes-2,3,4 and 5
Cont..
4 th row and 2 nd
Column as infinity
(2,1) also infinity
State-space Tree
6,7 and 8
Solution of TSP

1-4-2-5-3-1
0/1 Knapsack problem Using LCBB
N=4 m=15
(P1,p2,p3,p4)=(10,10,12,18) C^ =-38 1
(w1,w2,w3,w4)=(2,4,6,9) U=-32

C^ ,U, Upper=∞ 3 terms sum of wts=12


Let root node as E node Total wt-sum of wts=3
(0,0,0,15) is Ubound 4th object=3/9*18=6
Include 1,2,3 C^ =-38,U=-32
C^ with fraction
10 10 12
2 4 6 Convert maximization problem to
minimization so –ve sign
sum of wts=10 C^ =-38
Total wt-sum of wts=5 U=-32 1 X1=0
X1=1 Not Include
4th object=5/9*18=10 Include
C^ =-32,U=-22
C^ =-38 2 3
U=-32
C^ =-32
10 10 12 U=-22
2 4 6

(P1,p2,p3,p4)=(10,10,12,18)
(w1,w2,w3,w4)=(2,4,6,9)
First object
sum of wts=8 C^ =-38
Total wt-sum of wts=7 U=-32 1
4th object=7/9*18=14 C^ =-32
U=-22
C^ =-36,U=-22
C^ =-38 2 3
U=-32

10 10 12
2 4 6 4 5
C^ =-38 C^ =-36
(P1,p2,p3,p4)=(10,10,12,18) U=-32 U=-22
(w1,w2,w3,w4)=(2,4,6,9)
Second object
C^ =-38
sum of wts=6 U=-32 1
Total wt-sum of wts=0
C^ =-32
4th object=9/9*18 U=-22
No fraction part
C^ =38,U=38 C^ =-38 2 3
U=-32

10 10 12 18
2 4 6 9 4
C =-38
^ 5
U=-32
C^ =-38 C^ =-36
U=-32 U=-22 Third
(P1,p2,p3,p4)=(10,10,12,18)
6 7 C^ =-38 object
(w1,w2,w3,w4)=(2,4,6,9) U=-38
C^ =-38
sum of wts=6 U=-32 1 C^ =-32
Total wt-sum of wts=0
U=-22
4th object=9/9*18
No fraction part C =-38
^ 2 3
C^ =38,U=38 U=-32

C^ =-38 4 5 C^ =-36
U=-32 U=-22

C =-38
^ 6 7 C^ =-38
U=-38
U=-32

(P1,p2,p3,p4)=(10,10,12,18) C =-38
^

(w1,w2,w3,w4)=(2,4,6,9) U=-38 8 9
Solution
• X1=1
• X2=1
• X3=0
• X4=1
• Total profit=38 and Total weight=15
• In LCBB we need explore the node with least cost
every time which gives faster result.
• FFBB also same pattern follows explore nodes in
FIFO.

You might also like