open TELEMAC-MASCARET The mathematically superior suite of solvers

Dear all,

I am coupling Telemac with a home-made vegetation model. For this coupling, I need to compute a lot of intermediate variables in Telemac, which I then export in the result file and is finally read by my vegetation model.

The number of these intermediate variables is big enough so that I have to increase MAXVAR in declarations_telemac2d.f. I changed the value from 100 to 1000.

Everything works well in Telemac, except that only the first 100 variables in telemac2d seem eligible for export in the output file.

After many tests, I noticed that, even though MAXVAR = 1000 in declarations_telemac2d.f, it is not the case in bief_desimp.f (it is still equal to the initial default value of 100).

Does anyone know why that is? Can that be related to my problem that variables after the 100th cannot be exported?

The issue of MAXVAR = 1000 in declarations_telemac2d.f and MAXVAR = 100 in bief_desimp.f can be reproduced by adding the two modified fortran files into the user fortran folder of the malpasset example.

Hi Olivier

In fact, declarations_xxx files shouldn't be used in user fortran as they are used by many subroutines and then compiled first.
If you need to change a value in declarations_xxx file you need to recompile all the system
Hope this will works

Kind regards

I was afraid of that But that solved my problem, indeed.

Thank you Christophe for your quick and very useful answer!

Dear all,

As I said, I need to increase MAXVAR in declarations_telemac2d.f, for example to 1000 (to be sure, but 200 should be enough for most of my applications).

The ICT department in charge of one of the supercomputers I am using is a bot reluctant to do that change.

Does someone see a strong argument against that change on our local installations? Is there a strong reason why the limit is at 100 and not higher?

Thank you for your help!

Hi Olivier
There is no particular reason behind the number 100 which had been chosen long time ago...
This was a value which seems reasonable at this moment and without reserving too much memory for nothing.
During the last years, some of these harcoded limit was deleted, each time we saw that this become problematic.
So this will probably also the case in the future for MAXVAR. The task is not really complex but takes a little bit time...

Regards

Thank you again Christophe!