Learning Forum

[amnonjw]

Hello,

I reported in the past an issue with Lumerical and model parameter of type matrix: after simulation run the matrix value was nulled - it became became of dimension 0 (not the number 0); the parameter value became null. If I took such a ftp file after simulation, and then "return to layout mode" and re-run the simulation - then the IDE would crash.

This issue was reported in the old Lumerical Knowledge Exchange forum, and the post didn't pass on to this forum for reference. Kyle Johnson was PoC. In any case, you resolved the issue about 5 months ago.

I now report that this issue persists with a matrix parameter of a structure group. I attached simulation files (before and after running). The image below shows that the parameters of type matrix are nulled after the simulation run.

In short:

• The fsp file has matrix parameters in both the main "model" object and an inner structure group named "Holes".
• The model setup script takes values from its matrix parameters and passes them to matrix parameters in the structure group "Holes"
• After simulation run, the values of the matrix parameters of the model are persistent, but those of the structure group are nulled.
• If I have a sub-structure group inside a structure group - then its matrix parameters are also null after simulation. The issue aggregates.

[Taylor Robertson]

This should have been fixed in 2021 R2. I tested this on 2021 R2, and could not replicate the bug. Could you try updating the software to see if that fixes the issue?

Thanks

[amnonjw]

Hi This happens when running jobs on our parallel computing server. I think it runs using the command
lumerical-2021R1/mpich2/nemesis/bin/mpiexec
In the log file I see the version:
Ansys Lumerical 2021 R1.1 FDTD Solver Version 8.25.2599 (Linux 64bit)
Is it possible this build was also patched for this issue?
What are the latest builds for mpiexec and fdtd-engine-mpich2nem on linux?

[amnonjw]

I checked now on version 2021-R2.5 and the values are not nulled.
Good job