NaN - Running a smaller CAN4 model

NaN - Running a smaller CAN4 model

Postby Hansen » Mon Jul 10, 2017 2:05 pm

Dear FDSIM team,

I have been running the example CAN4. Since the FD model is too big to be run on my PC (exceed memory), so I cut down the size of the FD model and change these corresponding parameters
in CAN4.MD file:

Modified
&BOUNDARIES
XBMIN = -1900.
XBMAX = 1900.
YBMIN = -7600.
YBMAX = 7600.
ZBMIN = -570.
ZHF = -380.

Original
&BOUNDARIES
XBMIN = -7600.
XBMAX = 7600.
YBMIN = -7600.
YBMAX = 7600.
ZBMIN =-11020.
ZHF = -380.


in the CAN45.SR file:

Modified
&SOURCE NPS = 1, KEY_ONE_STF = F, PWINC = 0 /
0. 0. -500. 0. 40.
22.5 90. 0. 1.e+18

Original
&SOURCE NPS = 1, KEY_ONE_STF = F, PWINC = 0 /
0. 0. -3000. 0. 40.
22.5 90. 0. 1.e+18

and of course the number of receivers (CAN4.IN) and their locations (RECEIVER.DAT).

All the other parameters remain intact.

But I could not obtain any results (firstly 0 and then followed by NaN).

Could you please give me some glues as to where the failures originate?

Input files are attached for your reference, and please feel free to contact me if there is any ambiguity in my post.

Thank you with kind regards

Hansen
You do not have the required permissions to view the files attached to this post.
Hansen
 
Posts: 5
Joined: Thu Jun 01, 2017 5:03 pm

Re: NaN - Running a smaller CAN4 model

Postby fdsim » Mon Jul 10, 2017 2:26 pm

Dear Hansen,

because you decreased the size of a model, you have to use smaller number of MPI processes. Try to use PROCX=2, PROCY=1,PROCZ=1, i.e. just 2 processes.
For such a small model this should be enough.

Let me know, if this helps you.

Regards,
Jozef
fdsim
Site Admin
 
Posts: 33
Joined: Thu Mar 13, 2014 7:34 pm

Re: NaN - Running a smaller CAN4 model

Postby Hansen » Mon Jul 10, 2017 2:36 pm

Dear Jozef,

Thank you very much for your quick reply.

Sorry for not mentioning that both model building and simulation are by serial mode, not parallel. The PROCX etc. will be, I think, simply be ignored by the code.

But any way, I change the parameters as you suggested, the results are still the same. Hence, I think, the The PROCX etc. are not of concern.

Thank you with kind regards

Hansen
Hansen
 
Posts: 5
Joined: Thu Jun 01, 2017 5:03 pm

Re: NaN - Running a smaller CAN4 model

Postby fdsim » Mon Jul 10, 2017 2:57 pm

Dear Hansen,

now I see. We will try to run your problem in serial and we will let you know.
We are using the code only in parallel way so there is possibility of error in serial part of the code.

Regards,
Jozef
fdsim
Site Admin
 
Posts: 33
Joined: Thu Mar 13, 2014 7:34 pm

Re: NaN - Running a smaller CAN4 model

Postby Hansen » Wed Jul 12, 2017 12:40 pm

Dear Jozef,

Just let you know that I run the same model by parallel computing. I got reasonable results.

So I reckon and I agree with you on that there may be some errors in serial part of the code. but it is not a big issue since it makes no sense not to use the parallel computation.

Thank you for your time

Kind regards

Hansen
Hansen
 
Posts: 5
Joined: Thu Jun 01, 2017 5:03 pm


Return to Run simulations

cron