esa / NIDN

Neural Inverse Design of Nanostructures
GNU General Public License v3.0
36 stars 6 forks source link

Use largest gridsize possible for FDTD (if frequencies differ) #69

Open gomezzz opened 2 years ago

gomezzz commented 2 years ago

Then, resample time domain signal

gomezzz commented 2 years ago

relates to #58

torbjornstoro commented 2 years ago
# Test to see if scaling with each wavelength makes a difference in terms of spectrum
"""scaling = max(
    UNIT_MAGNITUDE / (wavelength * FDTD_GRID_SCALE),
    cfg.FDTD_min_gridpoints_per_unit_magnitude,
)"""

Tried this, but the results was very different from before, i.e. not good

gomezzz commented 2 years ago

@torbjornstoro write down thoughts here pls :pray: