Unexpected termination of the VEDA-FE run

[seckg]

Hello,

I have one issue i'm trying to solve. After a successful synchronisation into VEDA FE, i have an "unexpected termination of run". The QA_check is perfect, almost no warning at all (only the *** Demand: DEM commodity with missing COM_PROJ Projection which corresponds to some demand commodities without any demand projection for them, thus harmless though)

However in the .lst file (please find it enclosed here) there are one error message:

---- 167471 Assignment OBJ_DISC      0.000     4.369 SECS    200 Mb      1

---- 167472 IF-ELSE                  0.000     4.369 SECS    200 Mb 
---- 167474 Loop                     0.000     4.369 SECS    200 Mb 
---- 167476 Assignment F             0.000     4.369 SECS    200 Mb      1
---- 167477 Assignment F             0.000     4.369 SECS    200 Mb      1
**** Exec Error at line 167481: division by zero (0)
---- 167479 Loop                     0.000     4.369 SECS    200 Mb 
---- 167483 Assignment G_DRATE       0.000     4.369 SECS    200 Mb     13

Unfortunately, I really do not understand this issue. Do you have any idea please?
Thanks in advance

[Antti-L]

Some time ago there was another user reporting apparently the same problem:

http://forum.kanors-emr.org/showthread.php?tid=756

I tried to find a way to reproduce the problem by myself, but without any luck. However, that other user had also some other problems in his model, and the problem went away after fixing them. But I would very much like to know what may cause such a division by zero error, which you are now also getting.

I would thus be grateful if you could provide a reproducible test case (the *.DD and *.RUN files) for the case where you get the error). If you can zip and upload these files e.g. to Dropbox or Google Drive, and send me the download link, I would be grateful.

[seckg]

Hello Antti
I maybe finally found where is located the problem.
The problem was to have the sheet defaults in syssettings for ~Currencies to define a default currency for the whole model where i did not define any default currency so i assume the model maybe was having a problem with that (considering maybe zero which explains the divided by zero). Well it is just an assumption and maybe you can confirm that.
In any case after removing it, the model started to work smoothly
Thank you for your time

[Antti-L]

Thanks for the follow-up.

Yes, it never occurred to me that the set CUR would be empty!  That is obviously the explanation for the problem.  But what is even more peculiar, GAMS still reads in the G_DRATE parameter, with the undefined currency 'CUR', and TIMES therefore actually thinks that 'CUR' is a member of the set CUR.  This messy situation is possible, because VEDA sets an option to allow for domain violations.

Anyway, thanks for helping to find out the reason for this problem. I can now implement a new QA check to issue a clear error when this situation occurs (set CUR is empty, or does not contain the currencies used).