Skip to main content

Installation

phspectra logo

phspectra is a Python library for decomposing 1-D astronomical spectra into Gaussian components using 0-dimensional persistent homology for peak detection. Instead of derivative-based methods (GaussPy) or brute-force parameter sweeps, it ranks peaks by their topological persistence and uses a single parameter (β\beta) to threshold noise from real structure.

Requirements​

  • Python >= 3.11
  • NumPy >= 1.26
  • SciPy >= 1.12

Install from PyPI​

pip install phspectra

An optional C extension (_gaussfit) is compiled automatically from source when building from a checkout that includes src/phspectra/_gaussfit.c. If compilation fails, the library falls back to SciPy's curve_fit transparently.

Publishing​

The package is published to PyPI via python-semantic-release. On every merge to main, the release pipeline:

  1. Parses conventional commits scoped to packages/phspectra/.
  2. Bumps the version in pyproject.toml and __init__.py.
  3. Tags the release as phspectra-v{version}.
  4. Builds an sdist + wheel with python -m build.
  5. Publishes to PyPI via trusted publishing (OIDC, no API tokens).