E-surge crashes when making result files?

[Alexlyon]

Hi

I have some troubles when using E surge to fit multi-state models.
My models should allow transitions between 7 observable states and 4 non-observable states (in which one dead state).
Actually the models run, and the optimization is achieved successfully, but E-surge seems to crash after running without giving results files (.txt and.xls).

First, I thought that it was because of the number of states (11), and the mark input format used which not allow for more than 10 states. However, as far as my data consist of only 7 observable states, the mark format cannot explain this crash.

Did anyone encounter a similar problem ?
Does the trouble come from the model ? or when building the results files ?

Any help will be welcome,
thanks in advance

Sincerly

Alexis

[CHOQUET]

What was the error messages ?
Rémi

[Alexlyon]

Hi,

Actually, there was no error message.
I see the iterations scrolling in the DOS window until the message "Optimization terminated succesfully: current search direction [...] 2*options.tolfun." plus a list of values near zero with heading "ans".

then i wait for the result files, but no result files are built.. and E-surge breaks down (out of command).

It 's a little bit surprising.
I ran simpler models and the result files appeared, but the time between the end of iteration and the result file output remained long (>10 minutes).
For the more complex models I am running, I have waited for results output more than one night, but no result has appeared even optimization has been reached successfully

Will E-surge/Matlab have some difficulties to compute the model rank?

thanks again

Alexis

[CHOQUET]

Hello,
Perhaps, even your basic model are quite complex regarding the
number of parameters. However, there is a option to deconnect the computation of the rank in E-SURGE.

Select in Advanced Numerical Options:
Analytical Gradient -> No

Sincerely,
Rémi