Exoplanet Hunter – TESS / Lightkurve

This repo is a laptop-only exoplanet transit search pipeline built on public NASA TESS data. It uses the Lightkurve Python package to pull light curves from the MAST archive, searches for repeating box-shaped dips with a Box Least Squares (BLS) periodogram, and logs candidate signals into a CSV catalog.

• Pi Mensae pipeline (Colab notebook)
  notebooks/pi_mensae_pipeline.ipynb runs end-to-end on the known TESS planet Pi Mensae c:

  • Queries TESS sectors via lk.search_lightcurve.
  • Downloads, stitches, and flattens the light curve.
  • Runs a BLS periodogram to find a repeating transit signal.
  • Folds the light curve on the best-fit period to reveal the transit shape.
  • Applies secondary-eclipse and odd/even transit checks.
  • Estimates planet radius from transit depth and stellar radius.

• Command-line hunter with logging
  src/hunt.py is a CLI script that:

  • Takes a star name (e.g. "Pi Mensae", "HD 63433", "AU Mic").
  • Runs the same Lightkurve + BLS pipeline in Python.
  • Prints TIC ID, period, transit depth, and BLS power score.
  • Appends results to results/candidates.csv with columns: star, tic_id, period_days, depth, power, note.

• Batch survey
  src/batch_search.py reads a list of target stars from data/targets.txt, runs hunt() on each one, and grows the results/candidates.csv catalog in a reproducible way.

• Results catalog
  results/candidates.csv currently includes entries such as:

  • Pi Mensae (TIC 261136679) with a strong repeating transit signal.
  • HD 63433 (TIC 130181866) with a shallow but coherent dip.
  • AU Mic (TIC 441420236) with a deeper candidate transit.
  • Targets with note = no_data where TESS does not have suitable light curves.

Below is the Streamlit interface running the hunt on Pi Mensae. The app shows the star name, TIC ID, best-fit period, transit depth, BLS power score, and a folded transit plot built from real TESS data.

Instead of a toy simulation, this pipeline works directly on real space telescope data from NASA’s TESS mission. The code structure makes it easy to:

• Re-discover known planets (sanity check against published values).
• Apply the same search and vetting logic to new target stars.
• Build up a quantitative catalog for further scientific or ML analysis.

Create a virtual environment and install dependencies:

python3 -m venv venv
source venv/bin/activate
pip install lightkurve numpy streamlit

Then run the hunter on any TESS target whose name resolves via Lightkurve:

python src/hunt.py "Pi Mensae"
python src/hunt.py "HD 63433"
python src/hunt.py "AU Mic"

Each run will append a row to results/candidates.csv.

With the same virtual environment activated:

streamlit run app.py

This will open a local web app where you can enter a star name, run the hunt, see the key numbers (period, depth, power), and view the folded transit plot.

• More vetting helpers as standalone functions (secondary eclipse, odd/even).
• A small report in docs/methods.md summarizing a search over a larger set of stars.
• Optional extensions for ML-based classification of candidate signals.