Closed kitchenknif closed 6 years ago
They are inequivalent when you have both σ and Lorentzians, but since both of them have to be fit to experimental data and both have the same asymptotic behaviors it doesn't seem like a problem.
You do have to do a separate fit, though, if you have a fit from the other form.
The documentation for how meep handles dispersive materials includes a formula for Drude-Lorentz dispersion, that is significantly different from models I've seen in FDTD-related literature.
Equation from your documentation:![render2](https://user-images.githubusercontent.com/8235105/41518934-a2320392-72f7-11e8-9210-c1c9ae72554f.png)
Equation from, e.g., doi:10.1016/j.cplett.2007.08.027![render1](https://user-images.githubusercontent.com/8235105/41518974-df134258-72f7-11e8-803a-8f4700414ad3.png)
For the model used and data from dot:10.1364/AO.37.005271, these two models would be equivalent, because conductivity is set to zero. But for fits where the conductivity is not zero, these two models are vastly different.
Am I misunderstanding something here?