Build geometry using default_material

[dalarev]

Hi everybody,

I wrote a "tutorial" script demonstrating how to use the default_material function to define geometry. Please find it here.

I needed this capability for my own work, so I built a simple cylinder model to compare results using default_material with GeometricObject.

We use a dft field monitor to compare field profiles resulting from both methods. They seem to match up almost exactly, which is a nice validation.

Lessons learned:

• When defining the NumPy array to feed to default_material, make sure that your grid axes match those of the sim, or else the geometry will be scaled! This wasn't obvious to me - I assumed that meep would match spatial coordinates and insert the geometry appropriately. Perhaps this is spelled out in the documentation somewhere, but I was not aware of it.
• Comparing the get_epsilon data between the two methods yields some differences in the perimeter of the geometry. Perhaps this is due to slight grid and interpolation mismatches. This doesn't seem to impact the results.

Hope this is useful to someone!

[dalarev]

This was in response to this thread in which I ask if one can Boolean AND GeometricObjects. I realized that eventually more complicated geometry would be needed, so the default_material method would be more versatile.

[Lotfi99999]

Thank you