Hi Hélène, I am not an OpenMOLE developer, but I think I can help with this question:
I'm also not sure about what's in the result files generated by the > SavePopulationHook function (best results? but then why is my entire > population in it for the last few generations?). As you know, NSGA-II is trying to find as many points as possible on the Pareto front, the set of all non-dominated compromises. As the generations go on, more and more points will be found, and at a certain point the whole population might end up on the Pareto front. You should take a look at the final results, and if you think that the front has not been explored enough, you might want to restart the algorithm increasing the size of the population (sometimes this parameter is called 'mu'). Best regards, On Wed, May 31, 2017 at 2:24 PM, Hélène Arduin <[email protected]> wrote: > Hi, > > I need to calibrate a model, and I've been trying the NSGA2 algo along > with the SteadyStateEvolution function. It seems to be working fine, > however genetic algorithms are completely new to me and I'm not sure I > fully understand all that's happening. > In particular, the "parallelism" parameter in SteadyStateEvolution is very > obscure to me, and I get weird results when I try different values. I'm > also not sure about what's in the result files generated by the > SavePopulationHook function (best results? but then why is my entire > population in it for the last few generations?). > > I haven't found detailed documentation on these functions on the openmole > website (besides the "GA with NetLogo Tutorial") or on the internet, so if > anyone could point me to it if it exists, or is willing to explain this to > me it would be great! > > Thanks in advance, > > Hélène > > -- > Hélène Arduin > PhD student > Biostatistics, Biomathematics, Pharmaco-Epidemiology and Infectious > Diseases > Inserm UMR1181 / Université Versailles Saint-Quentin en Yvelines / > Institut Pasteur > > 25 rue du Docteur Roux > 75724 Paris Cedex 15, France > Tél : +33 (0)1 44 38 91 38 <+33%201%2044%2038%2091%2038> > > http://b2phi.inserm.fr > https://research.pasteur.fr/en/team/pharmacoepidemiology- > and-infectious-diseases > <https://research.pasteur.fr/en/team/pharmacopepidemiology-and-infectious-diseases> > > _______________________________________________ > OpenMOLE-users mailing list > [email protected] > http://fedex.iscpif.fr/mailman/listinfo/openmole-users > > -- Alberto Tonda
_______________________________________________ OpenMOLE-users mailing list [email protected] http://fedex.iscpif.fr/mailman/listinfo/openmole-users
