Skip to content

CLAUDE.md

Latest commit

 

History

History
99 lines (67 loc) · 4.04 KB

CLAUDE.md

File metadata and controls

99 lines (67 loc) · 4.04 KB

PypeIt

PypeIt is a pure Python package for processing raw spectroscopic data from astronomical telescopes into calibrated spectra for scientific analysis.

Development Setup

  • Create and activate a fresh python environment

  • Clone your fork of the main pypeit repo, hosted at GitHub, and add a remote git connection to the main repo called upstream.

  • Install pypeit using the zshell (or bash equivalent) command pip install -e ".[dev]".

  • Create a new git branch, where simple bug fixes branch from release and all other development branches from develop.

  • Frequently fetch changes to the upstream base branch (release or develop) and rebase, as necessary.

  • Additional development guidelines is provided in doc/dev/development.rst.

Key Architectural Components

  • PypeIt implements three primary processing paths, depending on the format of the spectrograph used to collect the raw data:

    • MultiSlit: Standard long-slit and multi-slit spectrographs
    • Echelle: Cross-dispersed echelle spectrographs
    • SlicerIFU: Slicer-based integral-field units

  • Instrument specifications and data-processing considerations that are specific to each spectrograph (including the processing path that should be used) are isolated in their respective spectrograph classes, all of which inherit from pypeit.spectrographs.spectrograph.Spectrograph.

  • Users primarily modify the code performance via user-level parameters held by the pypeit.par.pypeitpar.PypeItPar class, which packages the hierarchy of parameter sets used throughout the code.

  • Core processing modules, particularly for calibrations, produce FITS files that are saved to disk and reused as necessary. These modules commonly use pypeit.datamodel.DataContainer as a base class to enforce strict adherence to a well-defined datamodel and to provide a common IO interface.

  • Significant portions of the pypeit/data directory are not included in the package distribution of the code, but rely on a cache system that downloads files as needed for processing.

Testing

  • All tests are collected in the pypeit/tests directory.

  • Tests in pypeit/tests should be limited to unit tests that do not require the use of large data files.

  • Tests should be deterministic; i.e., all random-number generators should use a fixed seed.

  • Test coverage is supplemented by the PypeIt development suite, hosted at https://github.com/pypeit/PypeIt-development-suite, which requires data files hosted on Google Drive; see README.rst for the link.

Documentation

  • All functions and classes, except for tests in pypeit/tests should include docstrings that explain their purpose, input arguments, and output objects.

  • The docstring style is currently not consistent within the repository, but Numpy style docstrings are preferred; see https://www.sphinx-doc.org/en/master/usage/extensions/example_numpy.html

  • Package documentation for users is held in the doc directory, which is built using Sphinx and hosted on ReadTheDocs at https://pypeit.readthedocs.io/.

  • A complete rebuild of the documentation is performed by executing the bash command cd doc; make clean ; make html. This requires access to the internet and the PYPEIT_DEV environmental variable, which points to the directory containing containing the RAW_DATA directory copied from the PypeIt development suite Google Drive (see "Testing" above). If these requirements are not met, a limited rebuild of the documentation can be achieved by executing cd doc ; make htmlonly.

  • Documentation should be updated with each GitHub pull request.

Usage

  • Users interact with the code base via execution of the command-line scripts found in pypeit/scripts.

  • The primary data-processing script is run_pypeit.py, which primarily instantiates the pypeit.pypeit.PypeIt class and runs its methods.

External Resources