Quickstart

This page shows the minimal steps needed to detect a transit signal. For a worked example on real TESS data see the notebooks in the examples/ directory of the repository.

Loading a light curve

Pass time (days), flux, and flux error arrays to LightCurve. Flux values should be normalised to a baseline of 1.0.

import numpy as np
from cetra import LightCurve, TransitDetector

lc = LightCurve(times, fluxes, flux_errors)

CETRA resamples the input data onto a uniform cadence using a GPU kernel. Any cadence gaps are filled with null points (infinite error) so that array indexing can be used instead of time lookups during the search.

Running the linear search

Wrap the light curve in a TransitDetector and call linear_search():

detector = TransitDetector(lc)
linear_result = detector.linear_search()

This performs a 2-D grid search over transit duration and mid-transit time (t0), returning a LinearResult containing likelihood, depth and depth-variance arrays.

To retrieve the highest-SNR single transit:

transit = detector.get_max_snr_single_transit()
print(transit)

Running the period search

The period search phase-folds the linear search result over a grid of periods. It must follow the linear search:

periodic_result = detector.period_search()
transit = detector.get_max_likelihood_periodic_transit()
print(transit)

Extracting multiple signals

Use the iterative extraction methods to find all signals above a given SNR threshold. Each signal is masked out before the next search, then the original light curve is restored.

# single transits above SNR 7
transits = detector.get_single_transits_above_snr_threshold(snr=7.0)

# periodic transits above SNR 7
transits = detector.get_periodic_transits_above_snr_threshold(snr=7.0)

Using a different transit model

Three built-in limb-darkened models are available at impact parameters b = 0.32, 0.93 and 0.99:

from cetra import TransitModel

model = TransitModel('b93')
detector = TransitDetector(lc, transit_model=model)

A custom 1-D array can also be passed directly to TransitModel.