Period Finding

LightCurve wraps four period-search backends. They all populate lc.periods (candidates, strongest first) so downstream methods like plot_phased can pick one up automatically.

from varistar import TimeSeries, LightCurve
from _synthetic import make_sinusoidal_lightcurve

df = make_sinusoidal_lightcurve(period=3.2453)
ts = TimeSeries(magnitude="mag I", time_scale="HJD")
ts.load_data_from_df(df, data_id="synthetic_001")

lc = LightCurve(ts)

Lomb-Scargle (default)

lc.run_ls(min_freq=0.05, max_freq=5.0)
lc.plot_periodogram()
lc.periods[:3]

Best-period selection

find_best_period runs Lomb-Scargle if needed, then checks for harmonic aliasing and phase coverage before committing to one period:

result = lc.find_best_period()
result

{'best_period': 3.246397593105865,
 'is_harmonic': False,
 'not_periodic': False,
 'not_dominant': False}

Alternative backends

Useful when a light curve is strongly non-sinusoidal (e.g. eclipsing binaries), where Lomb-Scargle can lock onto a harmonic instead of the true period.

# n_freq / n_bootstrap reduced from their defaults for faster docs builds
lc.run_pdm(min_freq=0.05, max_freq=5.0, n_freq=2000)

3.251728344855632

lc.run_pdm2(min_freq=0.05, max_freq=5.0)

PDM2: 5851 trial frequencies...

6.490998218381418

lc.run_sr(min_freq=0.05, max_freq=5.0, n_bootstrap=200)

Spectrum Resampling: running 200 bootstraps...
SR result: P = 3.24640 ± 0.00000 d

3.246397593105866

Next: Phase Folding & Models.