> De: "Katerina Rippi" <[email protected]> > À: "users" <[email protected]> > Envoyé: Lundi 6 Juin 2016 17:48:29 > Objet: [ot-users] convergence criteria and optimization algorithms of FORM
> Dear all, > I was recently dealing with FORM analysis by using different optimization > algorithms (Cobyla, AbdoRackwitsz and SQP). In the manual, “ Cases Guide for > the Textual User Interface” there is an explanation of the convergence > criteria. However, I found it difficult to understand their exact meaning and > what each error might represent in FORM iterations. It would be helpful if > there was a figure and/or explanation in the manual with illustrating the > errors during FORM iterations. Do you agree? > Moreover, I could not find any explanation about the different optimization > algorithms. I found some external documentation describing the theory behind > these algorithms but it would be also helpful that an explanation of how they > were implemented in OpenTURNS (OT) is available in the manual. In my case, for > example, I experienced convergence problems with Cobyla and then when I > steered > a parameter (i.e. ρ beg), it seemed that it worked. However, I think that it > would be much faster and efficient if the explanation of these parameters and > their exact implementation in OT are also available in the manuals. In my > case, > I only fully understood it after try and error. > I would really like to know your experience on this matter. Is there already a > more detailed elaboration of the aforementioned issues ? and if yes, where can > I find it? If not, I suppose that including the things above in the manual > would also help future users of OT and additionally help them understand > better > the usage of the different criteria and parameters. > Thank you in advance for your consideration and let me know your opinion. > Cheers, > Katerina Rippi > DISCLAIMER: This message is intended exclusively for the addressee(s) and may > contain confidential and privileged information. If you are not the intended > recipient please notify the sender immediately and destroy this message. > Unauthorized use, disclosure or copying of this message is strictly > prohibited. > The foundation 'Stichting Deltares', which has its seat at Delft, The > Netherlands, Commercial Registration Number 41146461, is not liable in any way > whatsoever for consequences and/or damages resulting from the improper, > incomplete and untimely dispatch, receipt and/or content of this e-mail. > _______________________________________________ > OpenTURNS users mailing list > [email protected] > http://openturns.org/mailman/listinfo/users Hello, I agree, we definitely could use some more details in the doc. Progress to be tracked here: http://trac.openturns.org/ticket/839 J.
_______________________________________________ OpenTURNS users mailing list [email protected] http://openturns.org/mailman/listinfo/users
