Forum - Early Discussion

Overview > Models > MBPI

Early Discussion [#3]

T

As a tip: The tool is not working if you initiliaze parameters to zero and usually the error message that you will get is not relevant. On the other hand, it has no problem handling zero limits.

As a suggestion: It would be nice if the tool was keeping more than one result when stopped. This is mainly because if you simulate for multiple curves then there exists the posibility that the overall error is the lowest but some tendencies of specific curves make no physical meaning or some curves could exhibit a large error in a small region. It would be helpful maybe if the user could choose the result from a set of different results.

Finally, I think that stopping criteria, defined by the user, could be added later on, cause after all they are depended on the application.

Posted at 08:20, 4 September 2015 by Tsolarion

C

Good tip, that's something we definitely forgot to mention in the videos as it is a problem the tool faces due to the scaling but that we hope to improve in future releases.

Concerning the suggestion it is a good idea that we can take into consideration. In the meantime you can just copy the parameters you like since you have the graphs with all the information. You can also adjust better the limits or add linear constraints that maybe could help you define what "makes physical sense" better.

As for the stopping criteria it is currently hidden to the user but you can still change it trough the code by modifying the optimset options in fmincon or globalsearch in (as for version 1.0) /bin/MainApp.m I think most users will want a simple tool that "just works" but some options could be added for more experienced users.

Thanks for your feedback!

Posted at 14:41, 4 September 2015 by Carlosgs91 (administrator)