HIPS / Spearmint

Spearmint Bayesian optimization codebase
Other
1.55k stars 329 forks source link

simplified grad_dist2 fixing #22 #34

Open jan-glx opened 9 years ago

jan-glx commented 9 years ago

Compilation attempts led to crashes of python on my machine. Instead of just out-commenting the trial as in #22 I propose this simpler implementation. Equivalence "proof" and run time comparisons can be found at this gist. Please test again against a working c compilation. I have no Idea about what the function is doing, but the way the scaling by ls was implemented is odd, so please check if the square is correct.

This is my first pull request ever please let me know if I could have done anything better.