Note: This discussion is about an older version of the COMSOL Multiphysics® software. The information provided may be out of date.

Discussion Closed This discussion was created more than 6 months ago and has been closed. To start a new discussion with a link back to this one, click here.

Cluster Computing

Please login with a confirmed email address before reporting spam

Hello,

I am trying to run a mph file with a parametric sweep on a cluster with PBS scheduler.
From the log generated, it seems the file ran successfully for 3 of the 5 parameters.
Since it did it run successfully for all parameters, it seems that the output file was not saved.
When I run the file on my laptop using the GUI, the output is saved for the successful parameters.
Why can't I or how can I do this using cluster computing?

Melanie

8 Replies Last Post 2011年10月26日 GMT-4 19:48

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2011年10月24日 GMT-4 05:03
Hi Melanie,

Humm, interesting! I am facing similar problem!
Could you please provide more details about your calculation! i.e., OS, memory, the size of the generated file, DOFs, if it's a time dependent study etc!

Take a look @ my post here: www.comsol.com/community/forums/general/thread/22498/

Cheers!
Hi Melanie, Humm, interesting! I am facing similar problem! Could you please provide more details about your calculation! i.e., OS, memory, the size of the generated file, DOFs, if it's a time dependent study etc! Take a look @ my post here: http://www.comsol.com/community/forums/general/thread/22498/ Cheers!

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2011年10月24日 GMT-4 08:43
It is successfully generating output files but only if the entire model runs successfully. Therefore, if I feed it a bunch of parameters, it needs to be successful on all counts for the output file to save. Otherwise, I need to look at where it crashed (from the log generated), update the parameters and then run it again to see those results.....this seems a little stupid. I want the output to save regardless of successful convergence.

Do you have this problem too?
It is successfully generating output files but only if the entire model runs successfully. Therefore, if I feed it a bunch of parameters, it needs to be successful on all counts for the output file to save. Otherwise, I need to look at where it crashed (from the log generated), update the parameters and then run it again to see those results.....this seems a little stupid. I want the output to save regardless of successful convergence. Do you have this problem too?

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2011年10月24日 GMT-4 10:18
Ha OK!
It's different from what I have understood!

I think, may be you should use "stop condition" in order to get the model saved for the successful convergence calculation (parameters)! Just to not to re-run them again!

Hope it helps!
Ha OK! It's different from what I have understood! I think, may be you should use "stop condition" in order to get the model saved for the successful convergence calculation (parameters)! Just to not to re-run them again! Hope it helps!

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2011年10月24日 GMT-4 10:27
Where can I input this stop condition (in the mph file itself or this is a cluster command)?
What would be deemed an appropriate stop command?
Where can I input this stop condition (in the mph file itself or this is a cluster command)? What would be deemed an appropriate stop command?

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2011年10月25日 GMT-4 05:45
If you use the time-dependent solver for example, do this by right-clicking: on "Time Dependent Solver" and selecting Stop Condition from the context menu.

I don't know exactly the nature of the problem caused by this bunch of bad parameters, and if it can have a stop condition! But I am pretty sure that you can do something with this option! Just play with it!

Hope it helps!
Cheers
If you use the time-dependent solver for example, do this by right-clicking: on "Time Dependent Solver" and selecting Stop Condition from the context menu. I don't know exactly the nature of the problem caused by this bunch of bad parameters, and if it can have a stop condition! But I am pretty sure that you can do something with this option! Just play with it! Hope it helps! Cheers

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2011年10月25日 GMT-4 18:13
Melanie,
I am also looking for an answer to this problem. I emailed the tech support and am waiting for their response. If you found out anything then please update it on this forum!

Thanks,
Sirisha

PS: You cannot use stop condition for a stationary problem.
Melanie, I am also looking for an answer to this problem. I emailed the tech support and am waiting for their response. If you found out anything then please update it on this forum! Thanks, Sirisha PS: You cannot use stop condition for a stationary problem.

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2011年10月26日 GMT-4 08:58
Quoting answer given to me by COMSOL Support staff:

Regarding the output file, the
default behavior in COMSOL 4.2 is to simply stop the batch run without
saving when an error is encountered. This behavior can be disabled, in the
model file go to

Study > Job Configurations > Parametric

and disable the "Stop if error" check box.

In COMSOL 4.2a (which is being released in the coming weeks) the default
behavior has been changed; if an error occur during the batch job the
current model is saved.

Hope this helps!
Quoting answer given to me by COMSOL Support staff: Regarding the output file, the default behavior in COMSOL 4.2 is to simply stop the batch run without saving when an error is encountered. This behavior can be disabled, in the model file go to Study > Job Configurations > Parametric and disable the "Stop if error" check box. In COMSOL 4.2a (which is being released in the coming weeks) the default behavior has been changed; if an error occur during the batch job the current model is saved. Hope this helps!

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2011年10月26日 GMT-4 19:48
Thanks Melanie but I am using a scheduler script. Basically I use the batch mode from the command line. Looks like in this case, the job settings are not being used.

Sirisha
Thanks Melanie but I am using a scheduler script. Basically I use the batch mode from the command line. Looks like in this case, the job settings are not being used. Sirisha

Note that while COMSOL employees may participate in the discussion forum, COMSOL® software users who are on-subscription should submit their questions via the Support Center for a more comprehensive response from the Technical Support team.