Learning Forum

[fdtdisgreat]

Hi all,

I would like to use inverse design to optimize the shape of a resonator for maximum Q Factor (measured via transmission spectrum in FDTD). Am I right in believing that I won't be able to use LumOpt to optimize Q Factor? If this is the case, does Lumerical have a standard or suggested way of using the python API for inverse design with FOM such as Q Factor?

Thank you in advance.

[taige.li]

 

 

I think define the Q factor as an FoM and using the built-in optimization and PSO algorithm should work, yet this could be less efficient than LumOpt. 

I also struggle to utilize adjoint method by LumOpt and python. 

 

 

[Taylor Robertson]

Hello All,

 

Sorry I am late to this conversation; I think a previous response was lost due to technical difficulties. So the Q factor is not an FOM that can be directly optimized for with lumopt. Lumopt has to be used with guided mode power. If you could tie the output mode power to a particular spectra then it may be possible, but measuring the Q factor may require a different simulation strategie.

Q-factor calculation

That page describes how you can calculate the Q factor from the decay of the time domain fields. For high Q cavities, it will be hard to run for long enough for the impulse response to decay, so I would suggest using the high-Q analysis group, with Lumerical built in optimization tool.

Optimization utility