ENH: load particle definitions from pdg package

[Flomber]

Closes #332

⚙️ Enhancements

• {func}.load_pdg now accepts a source keyword-only argument. With source="pdg" (default remains "particle"), particle definitions are loaded from the official PDG Python API instead of the particle package provided by scikit-hep, giving a larger and independently sourced set of particle definitions. As of writing, particles loaded this way do not have LaTeX names, because the pdg package does not yet support them (particledatagroup/api#42).

[mmikhasenko]

I think for particle names they can produce latex.
Internally, it is not latex, it is some set of customary scripts

https://pdg.lbl.gov/2026/api/index.html

if it is not in sqlite @Flomber checked already

They say

To provide feedback or report bugs, please contact api@pdg.lbl.gov.

there is issue tracker here
https://github.com/particledatagroup/api/issues

Dean, and Jurgen could add it at the next release.

[redeboer]

Looks good! We just need some more cross-checks and documentation.

[redeboer]

Could you test whether there are any mismatches between two collections from particle and from pdg? We can now directly use this new feature to find bugs in the particle databases (which are hardcoded).

Note

This is something for a follow-up issue. So just try it first and if you find some problems, post an issue. We don't need that mismatch search to be implemented in the unit tests.

[Flomber]

I have wrote a small script to compare both sources. This is the output:

Details

from collections.abc import Iterable

from qrules.particle import Particle, load_pdg

def index_by_pid(particles: Iterable[Particle]) -> dict[int, Particle]:
    """Index particles by their Monte Carlo particle ID."""
    return {particle.pid: particle for particle in particles}


def main() -> None:
    """Report source coverage and particles available only from the PDG API."""
    scikit_hep_particles = index_by_pid(load_pdg(source="particle"))
    pdg_particles = index_by_pid(load_pdg(source="pdg"))

    scikit_hep_ids = set(scikit_hep_particles)
    pdg_ids = set(pdg_particles)
    only_in_pdg = pdg_ids - scikit_hep_ids

    print(f"Scikit-HEP source: {len(scikit_hep_ids)} particles")
    print(f"PDG source:        {len(pdg_ids)} particles")
    print(f"All PDG particles are in Scikit-HEP: {pdg_ids <= scikit_hep_ids}")
    print(f"All Scikit-HEP particles are in PDG: {scikit_hep_ids <= pdg_ids}")
    print(f"\nParticles only in the PDG source ({len(only_in_pdg)}):")
    for pid in sorted(only_in_pdg):
        print(f"{pid:>9}  {pdg_particles[pid].name}")


if __name__ == "__main__":
    main()

[redeboer]

Perhaps a variation of that script could be used to update the corresponding database in the particle package? I mean, one that directly gets the data from pdg and then fetches the missing entries for that CSV file.

[redeboer]

Thanks, LGTM! 📖