Lumerical FDTD

UCSB has a site license for Lumerical's FDTD software (as well as DEVICE and MODE - see this page for how to run them).

To run on the head node - just for data analysis, or short jobs - see the 'batch' section for longer job -, login with X forwarding (i.e. ssh -X) or use NX,

First, load the module (do "module avail" to see the different versions available)

module load lumerical/2023a

then you can run with

fdtd-solutions

When you're finished, please quit the program so other people can use it, we have a lot of licenses, but not unlimited, and it's a popular program.

Batch jobs for FSP files - see this page for how to run a script (LSF) file.

To run a 'fsp' file in the queue, you set your project up with the gui, then submit with the command fdtd-run-pbs.sh (on Pod, it's fdtd-run-slurm.sh).

[pcw@knot ~/lumerical]$ fdtd-solutions 2d.fsp

[pcw@knot ~/lumerical]$ fdtd-run-slurm.sh -n 8 2d.fsp

Submitting: 2d.sh

580512.node96

This will have submitted the job.  You can check on them with the 'qstat -u username' command, where you substitute your username in there so you only see your jobs.  

[pcw@knot ~/lumerical]$ showq $USER

   JOBID         USER PARTITION            NAME     CPUS    TIME_LIMIT         TIME    STATE     NODELIST(REASON)
 3274039          pcw     batch       2d.fsp.sh        4         10:00         0:02  RUNNING                node4

In this case it was run with 8 cores - that's the -n flag  (you can run it with up to 12 cores on knot, and 40 cores on pod.)

The last command 'showq' shows you what the job is doing.  It shows 'running' (in this example) or 'pending.  Once it doesn't show anymore, it means it's done and you can open the fsp file with the GUI and do the analysis.