Overview page
Scaling of run "time" with dimension
Expected running time (ERT in number of f−evaluations as log10 value), divided by dimension for target function value 10−8 versus dimension. Slanted grid lines indicate quadratic scaling with the dimension. Different symbols correspond to different algorithms given in the legend of f1 and f24. Light symbols give the maximum number of evaluations from the longest trial divided by dimension. Black stars (if present) indicate a better result compared to all other algorithms with p < 0.01 and Bonferroni correction number of dimensions (six). Legend: ○: MATSUMOTO Brockhoff, ♢: R-DE-10e2 Tanabe, ☆: R-DE-10e5 Tanabe, +: R-SHADE-10e2 Tanabe, ▼: R-SHADE-10e5 Tanabe, x: RL-SHADE-10e2 Tanabe, △: RL-SHADE-10e5 Tanabe, ◇: SOO Derbel, ◁: BrentSTEPif Posik, ▷: BrentSTEPifeg Posik, ⬠: BrentSTEPqi Posik, ♢: BrentSTEPrr Posik, ☆: CMA-CSA Atamna, +: CMA-MSR Atamna, ○: CMA-TPA Atamna, ▼: GP1-CMAES Bajer, x: GP1-CMAES Bajer 2013instances, △: GP5-CMAES Bajer, ◇: GP5-CMAES Bajer 2013instances, ◁: IPOPCMAv3p61 Bajer, ▷: IPOPCMAv3p61 Bajer 2013instances, ⬡: LHD-10xDefault-MATSuMoTo Brockhoff, ○: LHD-2xDefault-MATSuMoTo Brockhoff, ♢: RAND-2xDefault-MATSuMoTo Brockhoff, ☆: RF1-CMAES Bajer, +: RF1-CMAES Bajer 2013instances, ⬠: RF5-CMAES Bajer, ▼: RF5-CMAES Bajer 2013instances, x: STEPif Posik, △: STEPifeg Posik, ◇: STEPrr Posik