Information Technology Reference
In-Depth Information
1
Start
2
t:=0;
3
Initialize parental population
P
o
(
t
)withsex
o
;
4
Initialize parental population
P
c
(
t
)withsex
c
;
5
Repeat
6
For
k=1
To
λ
c
+
λ
o
Do
7
Choose one parent from
P
o
(
t
)andonefrom
P
c
(
t
)
8
recombination step sizes;
9
recombination objective variables;
10
mutation step sizes;
11
mutation objective variables;
12
fitness of a
k
:=
f
(
x
k
);
13
If
k
≤
λ
o
Then
14
sex(
a
k
)=
o
;
15
Else
16
sex(
a
k
)=
c
;
17
Add a
k
to offspring population
O
(
t
);
18
Next
19
Select parental population
P
o
(
t
+1)
20
from
O
(
t
) considering the fitness function;
21
Select parental population
P
c
(
t
+1)
from
O
(
t
) considering the constraint violation;
22
23
t
:=
t
+1;
24
Until
termination condition
25
End
Fig. 7.6.
Pseudocode of the TSES
These modifications lead to promising results on the test functions, see next
section 7.5.3. The sex ratio with a majority of
λ
c
≈
85
c
-individuals and only
λ
o
15
o
-individuals show that the diversity within the
o
-individuals may
not exceed a certain level. Otherwise, the population would be able to reach
the region of the unconstrained optimum resulting in an explosion of mean
step sizes. The survival possibility for the most successful individuals over up
to
κ
reproduction cycles emphasizes the important role of the
c
-individuals.
Figure 7.6 shows the pseudocode of the TSES.
≈
7.5.3
Experimental Analysis
An experimental analysis shows the properties of the TSES under various
conditions.
Parameter Analysis of the TSES
First, we show the runs of the TSES under various parameter settings for
μ
o
,μ
c
,λ
o
,λ
c
and
κ
[72], [75]. Table 7.4, upper part, shows the outcome of the
Search WWH ::
Custom Search