Databases Reference
In-Depth Information
Table 10.1.
First generation.
P
i
A
B
Look
Path
Cromosome
Fitness
f
i,norm
f
i,accu
P
1
4
10
2
(1,2,4)
00100 01010
0
.
0278
0
.
365
0
.
365
P
2
−
7
−
15
2
(1,5)
11101 11111
0
.
0156
0
.
205
0
.
570
P
3
8
15
2
(1,2,4)
00010 11110
0
.
0204
0
.
268
0
.
838
P
4
3
−
6
2
(1,5)
11000 01101
0
.
0123
0
.
162
1
.
0
F
t
=0
.
076
•
Single crossover.
•
Survival Probability
P
s
=0.5.
•
Mutation Probability
P
m
=0.1.
Table 10.1 shows the first generation which is randomly generated by the
testing tool. Each row in the table represents a member of the population
whose size is four. The columns in Table 10.1 have the following meanings:
•
P
i
indicates a member of the parent population;
•
A
and
B
are the values of the identifiers representing the input variables;
•
look (short for looking) gives the node number to be traversed;
•
Path
indicates which nodes have been traversed by this current test data
of A and B;
•
Cromosome
displays the bit pattern of the test data in binary-plus-sign
bit format;
•
Fitness
gives the fitness value calculated according to the test data and
the node required to be traversed;
•
f
i,norm
is the normalized fitness;
•
f
i,accu
is the accumulated normalized fitness value;
•
F
t
indicates the population total fitness value.
In Table 10.1, a 5 bit representation per input test data has been chosen.
Therefore, the cromosome size is 10 bits where the first five bits represents
the input data A and the rest five bits represents input data B. The least
significant bit is stored on the left hand side and the most significant bit
(sign bit) on the right hand side of the two substrings within the cromosome.
A large
f
i,norm
value indicates that these population members have a high
fitness value and hence have a higher probability of surviving into the next
generation.
The fitness function
f
, which calculates the test data performance based
on the condition
A
≤
B
(in the code) is given by
f
=1
/
(
|
A
−
B
|
+0
.
01)
.
2
