The discussion centers on the concept of mesh density as it relates to steps per wavelength in the context of a frequency domain electromagnetic (EM) solver, specifically in simulating a photonic crystal using CST Microwave Studio. Participants explore the implications of mesh density on the accuracy of simulations and the constraints imposed by hardware limitations.
Participants generally agree on the relationship between mesh density and the ability to resolve wavelengths, but there is no consensus on the specific definitions and appropriate values for steps per wavelength and minimum number of steps.
The discussion does not resolve the assumptions regarding the definitions of "step" and the implications of altering mesh density due to hardware constraints.
boneh3ad said:It's related to the Nyquist rate for time-varying signals, only in this case it is spatial. If you were to take the Fourier transform of your solution, you would only be able to resolve wavelengths of twice your step size or greater. If you are trying to actually build an accurate time or spatial history of the solution including those wavelengths, you would need to have an even more dense mesh.