Learning Forum

[fdtdisgreat]

Hi all,

Recently, I've noticed strange errors in my simulations. I will run a parameter sweep, and Lumerical will successfully execute the first 10 or 15 simulations, but then one simulation will complete but there won't be any data in any monitors, and so when my script tries to retrieve the data I get an error "Can not find result 'x' in the result provider 'Monitor_Name' ". When I re-run the specific simulation that this error occurred on, it will almost always complete without any error and the monitors will now have data. I am using lsf script to setup, run and post-process the individual simulations, as well as the sweep itself. Other people in my group who use different servers but the same license and Lumerical install have also reported this issue recently. This error appears with a frequency of roughly once every 8 or 10 simulations. Is this a known issue? Does anyone know what we can do to prevent these errors from happening, as it is essentially making parameter sweeps impossible to run at the moment? Thank you in advance.

[Guilin Sun]

I know occasionally one file could not run due to license issue, or could diverge, or could not be allocated to computing resources. But your case seems that specific file fools the software as if it is simulated, which I have not known of before. Please provide more information:
1: what computer you are using: desktop, or a cluster?
2: What the operation system is this?
3: what the job scheduler ?
4: what the mpich you are using?
5: is this a remote access or local? what remote tool you are using?
if the installation and license as well as manipulation are in different computers and operation systems, please describe each of them.

[fdtdisgreat]

Thanks for your reply. I actually managed to observe the error this morning in real time - the simulation will start and the Lumerical Job Manager window will pop up, but after a few seconds it closes again. Lumerical then moves onto the next part of my lsf script, which is retrieving the data from the monitors, but since there is no data in the monitors, Lumerical then throws an error. To be clear - it is not a divergence issue. To answer your questions:
1) A single server computer
2) Red Hat Linux
3) The standard Lumerical Job Manager (I run my simulations via a lsf script in the Lumerical GUI)
4) mpich2nem
5) Remote access - I'm using VNC Viewer from my personal computer to access the server, and running the simulations directly on the server i.e. I open up a Lumerical GUI via VNC Viewer directly on the server.
Yes, I believe we have a separate machine that handles the licenses, and our server communicates with that machine. I have never had a licensing issue though, or had an error message saying that the License couldn't be found. The Lumerical GUI doesn't close; the error just comes from the lsf script not being able to find any data.

[Lito]

Can you assist us with the following?
Which version of RHEL are you using?
Which version of Lumerical/FDTD are you running?
Can you send a screenshot of your resource configuration?
When this happens, check each of the "sweep's_p#.log files" for any errors.
Is there any error message from the job manager when running the sweeps?
Try to run your parameter sweeps outside of your script. -- i.e. using the "Run Sweep" button on the CAD/GUI. Then check if you have all the data/results on the sweeps. This should confirm if the issue is when running the sweeps on something on the script/scripting environment.

[fdtdisgreat]

Hi, thanks for your message. Here is the information
1) RHEL 8.5
2) Ansys Lumerical 2021 R2.4 FDTD Solver Version 8.26.2865 (linux 64bit)
3)

4) I checked the plog files. They have no information; they only say the Lumerical version, expiration date, config file and license host i.e. the first 4 lines of every plog file. Then nothing.
5) There is no error message. It just closes
I just noticed that before the fatal error, there is a warning in the Script Prompt stating that the simulation failed to run
Thank you.

[Lito]

Did you try to run your sweep from the CAD/GUI as per my previous message/image?
Can you try to run this example sweep from our Application Gallery? https://support.lumerical.com/hc/en-us/articles/360034922873-Parameter-sweep-utility . Run the sweep from the CAD/GUI as shown in the image.
If the issue only happens to your simulation/sweeps, please check if the simulation/sweep files are not corrupted/damaged. Try to remove the previous sweep's files from the sweep's folder and rerun your sweeps. Otherwise, if the sweeps run without any issues when run directly from CAD/GUI, there could be something wrong with your script.