simple_ell1_deorbit_numba

ell1fit.phase_utils.simple_ell1_deorbit_numba(times, PB, A1, TASC, EPS1, EPS2, tolerance=1e-08, max_iter=1000)

Deorbit event times, threading only when there are enough of them.

See _ell1_deorbit() for the model and the iteration, and _DEORBIT_PARALLEL_MIN for where the two variants cross. The single reference time that _calculate_phases() deorbits alongside each file’s events is the case this exists for: one element through the parallel kernel cost 111 us, a hundred times what the arithmetic needs.