Skip to content

Surrogate Assisted Distributed Evolutionary Algorithm for Computationally Expensive Problems

Notifications You must be signed in to change notification settings

Yash-Vardhan-Sharma/surrogate-assisted-distributed-evo-alg

 
 

Repository files navigation

surrogate-assisted-distributed-evo-alg Surrogate Assisted Distributed Evolutionary Algorithm for Computationally Expensive Problems

Problem 1 = Rosenbrock Problem 2 = Spherical Problem 3 = Rastrigin Problem 4 = Stablinski-tang

  • PSO (30 runs):
problem dim Avg Score Std Best Score Avg time (mins)
1 30 182.5 119 46.4 0.1234
1 50 1136.9 559.6 566.7 0.4322
2 30 9.9 12.8 0.78 0.0341
2 50 237.6 134.4 67.25 0.0612
3 30 68.56 16.16 37.94 0.0544
3 50 180.8 34.1 118.5 0.0977
  • Island- PSO : (num_islands = 8 , 30runs)
problem dim Avg score Std Best Score Avg time (mins)
1 30 119.25 34.39 49.72 0.2035
1 50 727.22 186.15 473.69 0.3068
2 30 3.59 2.20 0.44 0.1707
2 50 139.76 38.82 67.31 0.2039
3 30 57.82 9.09 41.65 0.1811
3 50 146.64 21.97 97.78 0.2349
  • Surrogate assisted distributed PSO : (num_islands=3 , 3runs)
problem dim Avg score Std Best Score Avg time (mins)
1 30 66.26 27.178 28.53 2.155
1 50 214.31 34.424 168.328 14.215
2 30 0.00028 9.849589159796698e-05 0.0001459 20.167
2 50 6.427044916618902e-05 2.1919447214603293e-05 4.418531105089173e-05 17.009
3 30 55.5 5.1052500149849145 49.097 7.168
3 50 154.5 9.569783116021142 144.88 9.613
  • Differential Evolution :(crossp = 0.7 , mut = 0.8)
problem dim pop_size MaxEvals Fitness
2 20 20 100k -(0)
2 30 20 100k -(0)
2 50 20 500k -(0)
3 20 20 100k 56.4(0)
3 20 10 100k 12.95(0)
3 30 10 100k 24(0)
3 50 10 500k 80(0)
4 20 10 100k -740(-780)
4 30 10 100k -1120.3(-1170)
4 50 10 500k -1771.4(-1950)
  • Island-Differential Evolution :(crossp = 0.7 , mut = 0.8 , island=3)
problem dim pop_size per island MaxEvals Fitness
2 20 10 100k -(0)
2 30 10 100k -(0)
2 50 10 500k -(0)
3 20 10 100k 13(0)
3 30 10 100k 30(0)
3 50 10 500k 80(0)
4 20 10 100k -755(-780)
4 30 10 100k -1100(-1170)
4 50 10 500k -1765(-1950)
  • G3-PCX
problem dim pop_size MaxEvals Fitness
1 20 100 100k 0 (0)
1 30 100 100k 5.1 (0)
1 50 100 500k 18.7(0)
2 20 100 100k 0(0)
2 30 100 100k 0(0)
2 50 100 500k 0(0)
3 20 100 100k 114.8(0)
3 30 100 100k 134.25(0)
3 50 100 500k 303(0)
4 20 100 100k -680.3(-780)
4 30 100 100k -968(-1170)
4 50 100 500k -1600(-1950)
  • Island-G3-PCX (Island=3)
problem dim pop_size per island MaxEvals Fitness
1 20 50 100k 19 (0)
1 30 50 100k 47(0)
1 50 50 500k 48(0)
2 20 50 100k 0.1(0)
2 30 50 100k 1.33(0)
2 50 50 500k 0.1(0)
3 20 50 100k 95.1(0)
3 30 50 100k 161(0)
3 50 50 500k 288(0)
4 20 50 100k -680(-780)
4 30 50 100k -1003(-1170)
4 50 50 500k -1621(-1950)

About

Surrogate Assisted Distributed Evolutionary Algorithm for Computationally Expensive Problems

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Python 100.0%