[Scilab-users] MMA / GCMMA (optimization method)

[paul.carrico at free.fr]

Dear All

I take here the initiative to ask advices on MMA/GCMMA method ... both in understanding it and after how to code it in Scilab.

So far I used optimizer(s) while I'm knowing the cost function (typically Sum Of the Square Errors SSE in order to fit measurements); For design optimization purpose using FEM solvers, It's quite often difficult to determine this cost function. 

An interesting French Book from JC Craveur et al (see EAN13 9782100600229) introduces this kind of issue speaking about the family of MMA methods (MMA / GCMMA / GBMMA) in order to create  and solve a set of sub-problems and then to master this kind of optimization topic ...

I do not measure the difficulties behind it and that's why I'm asking for feedbacks, so thanks in advance for it

NB:
- I found the Krister Svanberg articles for the GCMMA method in currently studying
- I know that pyopt librairies include MMA and GCMMA method


Thanks

Paul