Gaussian source larger than simulation region

What happens if I set up a Gaussian source with the waist larger than the simulation region? Will the monitor results be normalized with respect to the total power of the sourse, or only with rrespect to the part of the power that actually enters the simulation region?

When CW normalization is used, the transmission through a monitor will be normalized to the power injected by the source which is calculated using the formula here:

The formula does integrate the power injected over the simulated injection plane, so if the fields are truncated the calculated injected power will be smaller. However, if you have a truncated Gaussian beam profile, this will lead to diffraction effects similar to what happens when a plane wave is injected in a simulation region with PML boundaries at the sides:

More information about the diffraction edge effects here:

I would definitely recommend increasing the simulation span to include the full Gaussian beam profile.

1 Like

Thank you, however I could not understand how TFSF sourse can help with eliminating diffraction effect. Could you please explain?


The TFSF source injects a plane wave with a finite span without diffraction effects at the edges. You can see a video showing the propagation of the finite plane wave injected by a TFSF source on this page:

The TFSF source is useful in the case where you want to simulate the absorption or scattering or mode profile of a standalone structure due to an incident plane wave source such as in this Mie scattering simulation:

This type of source would be appropriate if the spot size of the beam which is being used in your experiment is much larger than the scattering object so that it can be approximated as a plane wave in the simulation.

Otherwise, if the beam cannot be approximated as a plane wave source, you should use the beam source with a large enough span to include the full beam profile.

Hopefully this helps!

Thank you!

Hmm I think I’ll have a similar problem. I am going to project a Gaussian beam (30um spot size) onto a unit cell (of 3um on a side). This unit cell will have Bloch Period BC for X and Y and PML for Z. The beam can come in from any angle. I would rather not simulate the entire structure because of it’s size, but are you suggesting that this is my only option?

When could a Gaussian beam be approximated as a plane wave?