gds_fdtd.extraction

gds_fdtd simulation toolbox.

Mode-overlap extraction — the last Tier-B enabler. Kernel engines (fdtdz, fdtdx) record raw DFT fields at port planes; this module turns them into the complex modal amplitudes that become SMatrix entries. Everything is solver-independent numpy: fields in, complex out.

Convention (the standard bidirectional mode-orthogonality projection):

a_m = [ integral(E x H_m*).n dA + integral(E_m* x H).n dA ] / (4 N_m)
N_m = 0.5 Re integral (E_m x H_m*).n dA

so a field equal to the mode gives a_m = 1 exactly, the backward-traveling mode gives 0, and |a_m|**2 is the fraction of the mode’s normalized power carried forward. direction="-" projects onto the backward mode instead (transverse H flips sign).

Functions

decompose(fields, modes, du, dv[, normal, ...])	Amplitudes of every mode in a recorded field; shape (len(modes),).
mode_amplitude(fields, mode, du, dv[, ...])	Complex amplitude of mode in a recorded field on the same grid.
mode_power(mode, du, dv[, normal])	N_m = half the real Poynting flux of the mode through its plane.