spectro-kernel-recipes

Curated measurement recipes for spectro-kernel, the catalogue of literature-cited astronomical spectroscopy algorithms.

A recipe is a measurement written down once, reviewed, referenced and versioned: which bricks, in which order, with which conventions. Everything that depends on the instrument or the observer is a declared variable, and a profile supplies the values at run time. Two observers with two spectrographs who run the same recipe produce numbers a researcher can compare, and every result carries the recipe name and version in its history. The idea is explained at docs.spectrokernel.io/recipes.

Install and use

pip install spectro-kernel spectro-kernel-recipes
spectro presets                                   # the collection appears as package:spectro-kernel-recipes
spectro preset show be_halpha_ew                  # conventions, references, variables, steps
spectro pipeline be_halpha_ew --input gamma_cas.fits --profile toulouse.yaml   # latitude_deg, longitude_deg
spectro pipeline binary_rv_ccf --input obs.fits --profile my_observatory.yaml

From Python:

from spectro_kernel import WorkContext
from spectro_kernel.pipeline import PipelineBuilder

pipeline = PipelineBuilder().from_preset("binary_rv_ccf", profile).build()
result = pipeline.execute(WorkContext(spectrum=spec))
print(result.history[-1])        # pipeline:binary_rv_ccf v1.0.0, the bound variables, the kernel version

The kernel discovers the collection through the spectro_kernel.presets entry point; nothing is fetched from the network at run time.

Recipes

status tells you how far the review went: draft (conventions open for discussion), reviewed (conventions documented and discussed), reference (adopted by the community as the way to make that measurement).

Contributing

See CONTRIBUTING.md: copy TEMPLATE.yaml, fill it in, validate it with spectro preset validate, open a pull request. The review discusses the science, not the YAML.

Licence

MIT, like the kernel.