pygsk: Generalized Spectral Kurtosis Toolkit

→ View full documentation

Overview

pyGSK is a modular, open-source Python toolkit for computing and visualizing the Generalized Spectral Kurtosis (SK) estimator — a statistical tool for signal detection, RFI excision, and spectral diagnostics.
It provides both programmatic and command-line interfaces for reproducible, open-science workflows.

Developed within the SUNCAST collaboration, pyGSK modernizes the legacy IDL implementation of the SK estimator into a fully transparent and community-maintained Python package.

Key Features

• ⚙️ Compute SK statistics for arbitrary integration parameters (M, N, d)
• 🧮 Derive PFA-based detection thresholds and visualize their evolution
• 📊 Plot SK distributions and detection boundaries
• 💻 Command-line interface (pygsk) with subcommands:
  • sk-test — compute and visualize SK thresholds
  • threshold-sweep — sweep thresholds over PFA ranges
  • renorm-sk-test — use the renormalized SK estimator
• 🔬 Pedagogical and reproducible: designed as a SUNCAST reference implementation

Installation

Install the latest stable version from PyPI:

pip install pygsk

To verify the installation:

python -m pygsk --version

For the latest development version:

pip install git+https://github.com/suncast-org/pygsk.git

Quick Example

from pygsk.thresholds import compute_sk_thresholds

M, N, d, pfa = 128, 64, 1.0, 1e-3
lower, upper = compute_sk_thresholds(M, N, d, pfa=pfa)

print(f"SK thresholds for pfa={pfa}: lower={lower:.3f}, upper={upper:.3f}")

Or equivalently from the command line:

pygsk sk-test --M 128 --N 64 --pfa 1e-3 --plot

Documentation

Full documentation is available in the docs/ directory:

Citation

If you use pyGSK in your research, please cite:

Nita, G. M. (2025). pyGSK: Generalized Spectral Kurtosis Toolkit. Zenodo.
https://doi.org/10.5281/zenodo.17336193

This concept DOI represents all versions and always resolves to the latest release.

The theoretical foundation is described in:

Nita, G. M., & Gary, D. E. (2010). The Generalized Spectral Kurtosis Estimator.
MNRAS Letters, 406(1), L60–L64.
https://doi.org/10.1111/j.1745-3933.2010.00882.x

License

This project is distributed under the MIT License.
© 2025 Gelu M. Nita and the SUNCAST Collaboration.

Acknowledgment

pyGSK was developed within the GEO OSE Track 1: SUNCAST: Software Unified Collaboration for Advancing Solar Tomography project, funded by the U.S. National Science Foundation (Award No. RISE-2324724).
It serves as a pedagogical and technical template for future SUNCAST community contributions supporting open, reproducible, and FAIR solar data analysis.