Thermal Switch example


I am trying to run the Thermal Switch example you have under Applications/Photonic Integrated Circuits/Circuits and systems/ using device specifically this file (thermal_switch_DEVICE.ldev), but it give me an error when it start meshing the structure (Tessellating).
"/opt/lumerical/mpich2/bin/mpiexec" -n 1 “/Applications/Lumerical/DEVICE/” “/Users/abdelsalamaboketaf/Desktop/thermal_switch_DEVICE.ldev”


I just downloaded the file and I made no changes to it. Can you please try it and tell me how can make it work. I tried different ways but no way.

Thank you

Unknown parallel error, the error code is 108 (DEVICE)

Hi @axa8863,

It seems this file requires a lot of memory to work! It might be the reason it crashes. I will check with my colleagues and I’ll keep you updated.


Hi @axa8863, apologies for the long wait. It looks like the HEAT solver is having trouble creating the geometry and is using up a lot of memory forcing the system to crash or become unresponsive in the process. For a structure as large as this one (160 um x 160 um) the solver sometimes have trouble creating and meshing long segments trying to make the mesh very coarse. In such cases it helps to force to solver to split these long segments into smaller ones so that they are more easily manageable. When I enabled the “split plc edges” option in the advanced options of the “Mesh” tab and set the “max plc edge length” to 5 um, the solver was able to mesh using very small amount of memory (5~6 GB).

The simulation however still failed to converge on the second bias point. I’ve found that turning the nonlinear (temperature dependent) conductivity model for the “NiChrome” metal OFF helps with the convergence. I will do some more testing to make this work with the nonlinear model. For now you can disable the temperature dependency of the conductivity model (of NiChrome) to make the simulation run for all bias points.

Here is the modified file: thermal_switch_DEVICE_mod.ldev (6.4 MB)