Performance summary of Unfit

The table below shows a summary of the time (in ms) that Matlab and Unfit 1.0 (Nelder-Mead algorithm) took to find the minimum of some common functions. The tests were run on the same machine (i.e., both Matlab and Unfit).

Function Matlab R2008a Unfit v1.0 Speedup
Beale 62081.0 10.2 6086
Easoms 9255.8 22.4 413
FreudensteinRoth 10527.8 10.0 1053
FreudensteinRoth2 8672.4 11.4 761
Griewangk 90148.4 1039.0 87
HelicalValley 12555.6 69.6 180
Madsen 8812.4 20.2 436
PowellSingular 28797.2 54.0 533
Rastrigin 7822.0 17.6 444
Rosenbrock 10013.8 14.6 686
SixHumpCamelBack 7225.4 10.4 695
SixHumpCamelBack2 6270.8 8.6 729
Watson 55743.4 1019.4 55
Woods 24851.6 68.4 363
Average 894