qsospec 0.1.0

A fast and robust Python package for fitting quasar spectra

qsospec

qsospec is a Python package for fitting UV, optical, and near-infrared quasar spectra. It provides array-based local and global fitting, recipe-driven emission complexes, bundled iron and Balmer templates, optional pPXF host subtraction, and resumable Parquet batch runs.

Installation

python -m pip install qsospec

Install optional pPXF host decomposition with:

python -m pip install "qsospec[host]"

For development:

python -m pip install -e ".[dev,host]"
pytest

Quick start

import qsospec

spectrum = qsospec.Spectrum.from_arrays(
    wavelength,
    flux,
    err=uncertainty,
    z=redshift,
    wave_frame="observed",
)
result = qsospec.fit_global_lines(spectrum)

Lyα/N V is selected automatically when useful rest-frame coverage exists. The default continuum then uses red-side anchors and the fit records whether coverage is full, red-side-only, edge-truncated, or absent:

result = qsospec.fit_global_lines(
    spectrum,
    lya_nv_config=qsospec.LyaNVComplexConfig(
        nv_mode="effective_blend",
    ),
)
lya = result.line_complexes.get("lya_nv")
if lya is not None:
    print(lya.metadata["lya_coverage_status"])
    print(lya.metadata["lya_fit_reliable"])

Pass an explicit complex list without "lya_nv" to disable it. Use nv_mode="equal_doublet" for a shared-kinematics N V doublet, or pass global_config=qsospec.GlobalContinuumConfig.lya_safe() explicitly when customizing the continuum model.

Single spectra and large samples share the same Parquet-backed run format:

run = qsospec.fit_object_to_store("spectrum.fits", "runs/object", redshift=1.2)
batch = qsospec.fit_batch(["spectra-000.parquet"], "runs/sample")

The project is licensed under GPLv3. The initial implementation was extracted from qsofitmore.neofit; its source history is retained in this repository.