Research Article

Adaptive Randomness: A New Population Initialization Method

Table 1

Comparison of convergence speed (NFC) and success rate (SR) for DE with random population initialization (DEr), with opposition-based population initialization (DEo), with generalized opposition-based population initialization (DEgo), and with AR-based population initialization (DEar).

DErDEoDEgoDEarARE
NFCSRNFCSRNFCSRNFCSR

304972514967014951014950511.00441.00331.0001
305565015573515562015553011.00221.00371.0016
2012166611204821121136112031411.01121.00141.0068
3032767013271851324500132653011.00351.00200.9938
102305250.952545450.952371750.952279350.951.01141.11671.0405
30638950.98624900.96621480.966472510.98720.96550.9602
301175111164511169811162111.01121.00211.0066
309762419781619743719741111.00221.00421.0003
2318013150131931313711.01371.00411.0179
41327211347011354411325511.00131.01621.0218
21525011548811538011522011.00201.01761.0105
2432514475144101421511.02611.05741.0463
30816418157181531812911.00431.00341.0030
2268112705127161267811.00111.01011.0142
400015410.01
10330320.24536520.36500600.32497600.360.66381.07821.0060
3025018412508481251408124971211.00191.00451.0068
3010223611012681102068110134411.00880.99931.0071
300000
302248512237012240212233111.00691.00171.0032
300000
40000
40000
40000
40000
279800.9779230.9681360.9880610.980.99000.98291.0093
285618591839180111.06871.07241.0474
3017248711724891177257117154311.00551.00551.0333
2495214781149241484511.02210.98681.0163
50210.0200
5283960.992920412897412890310.98251.01041.0025
2156711558115641154911.01161.00581.0097
2638616305163091623011.02501.01201.0127
2411014083140861406911.01011.00341.0042

Average values684880.997727695040.997727688790.997727680870.9977271.01161.01521.0138

Note. The reported average values (the last row) are calculated only on the functions where all the algorithms have the same success rate (i.e., all SRs = 1 and one SR = 0.95).