Information Technology Reference
In-Depth Information
[0] = f(0.5373683084, 8.549400615) = 8.80259358641701
[1] = f(1.169346791, 8.654789962) = 10.6843782636572
[2] = f(2.56474401, 8.549400615) = 11.1523261017334
[3] = f(2.798309803, 8.654789962) = 12.2595682909331
[5] = f(5.673876702, 8.654789962) = 13.1898377299992
[15] = f(9.105922305, 8.654789962) = 18.2273236993991
[40] = f(9.073935287, 8.654789962) = 18.4628024958807
[43] = f(9.070846391, 8.654789962) = 18.4777504127046
[94] = f(9.070846391, 8.66669448) = 18.4811446480484
[247] = f(9.058678611, 8.651249962) = 18.5211448849144
[406] = f(9.058678611, 8.664493) = 18.5263745426905
[1182] = f(9.058678611, 8.666354) = 18.5265717435938
[1520] = f(9.023396997, 8.666354) = 18.5370766419725
[1750] = f(9.048491662, 8.666354) = 18.5481190571652
[1905] = f(9.043386688, 8.666354) = 18.5532577056288
[2147] = f(9.042537043, 8.666354) = 18.5537463206678
[2576] = f(9.042537043, 8.668468023) = 18.5538093494319
[6747] = f(9.035891, 8.668468023) = 18.5540237663948
[6788] = f(9.036623, 8.668468023) = 18.5543135025517
[7108] = f(9.036837229, 8.668468023) = 18.5543836313701
[7865] = f(9.037875229, 8.668468023) = 18.5546293670704
[42316] = f(9.039186229, 8.668468023) = 18.5547168426169
[43243] = f(9.039186229, 8.668372756) = 18.5547176877183
[110851] = f(9.039186229, 8.66822349) = 18.5547183122798
[135051] = f(9.038857816, 8.66822349) = 18.5547197587122
[148895] = f(9.038901166, 8.66822349) = 18.5547204624021
[176618] = f(9.038940108, 8.66822349) = 18.5547208625485
[198737] = f(9.038978087, 8.66822349) = 18.554721041306
[465844] = f(9.039004378, 8.66822349) = 18.5547210427225
[569578] = f(9.039004378, 8.668181223) = 18.5547210644227
[819512] = f(9.039004378, 8.668181699) = 18.5547210645594
As you can see, the GEP-PO algorithm can approximate the global maxi-
mum of function (8.7) even better than the HZero algorithm. Indeed, the four
best solutions designed with the GEP-PO algorithm are better than the best
discovered with the HZero algorithm,
f
(9.039001464, 8.668243408) =
18.5547210135332, emphasizing again the fine-tuning capabilities of the
GEP-PO algorithm.
Let's now see how both algorithms perform at a much more difficult task,
the optimization of the following five-parameter function:
2
1
2
2
sin
p
p
1
.
25
cos(
p
p
p
)
3
4
5
f
(
p
,
p
,
p
,
p
,
p
)
0
.
(8.8a)
1
2
3
4
5
2
3
2
4
1
0
.
001
(
p
p
)
