GaussianBeam with a waist radius smaller than wavelength shows strange result

[tomflexcompute]

When the beam radius of a gaussian beam is somewhat smaller than its central wavelength, the beam profile deviates from the ideal gaussian beam. One example shown here is with central wavelength of 350 nm and beam radius of 200 nm.

[dmarek-flex]

I worked on reproducing this and testing how discretization affects the results. Discretization has no effect on the field distribution, so I believe the issue is due to the simple fact that the beam radius is subwavelength and therefore the paraxial approximation for the Gaussian beam is inaccurate.

[tomflexcompute]

Didn't realize we actually use the paraxial approximation for the gaussian beam. That definitely makes sense.