Post Process Modes Function Causing Inaccurate Simulation Results

[jakeh13]

I am trying to simulate a MMI splitter designed using the Lumerical EME solver in the MEOW EME solver. I believe the post process modes function is changing the effective index (from like 1.7 to 2.6 for some higher order modes) such that the result of the simulation is inaccurate. If I modify the meow/fde/tidy3d.py such that the modes are normalized instead of the post process orthonormalization, the result is much closer to the Lumerical result.

The transmission (0.1615) and fields when using the post process orthonormalization function:

The transmission (0.4736) and fields when only normalizing the mode fields:

I have attached the modified tidy3d.py file and my simulation script. I can't attach the gds file I am simulating but can email that if needed.

tidy3d.py
sim_1x2_MMI_splitter.ipynb

[flaport]

Hi Jake,

Thanks for reporting. I will look into it,

Best,
Floris