Cataloging nuclear data libraries

Build a reference DataFrame cataloging material, MT numbers, and nuclear-data libraries.

Cataloging nuclear data libraries
├── 1. Read `cross_sections.xml` from each library
├── 2. Extract MT numbers from HDF5 files
├── 3. Rearrange dataframe to one MT per row
├── 4. Build materials-MT-libraries catalog
└── 5. Save df and catalog for future reference

1. Read `cross_sections.xml` from each library¶

import h5py
import openmc
from pathlib import Path
import pandas as pd
import xml.etree.ElementTree as ET

from openmc_workbench_config import DATA_ROOT

rows = []

for neutron_dir in sorted(DATA_ROOT.glob('*/neutron')):
    library = neutron_dir.parent.name
    xs_xml = neutron_dir.parent / 'cross_sections.xml'
    root = ET.parse(xs_xml).getroot()

    for element in root.findall('library'):
        if element.get('type') != 'neutron':
            continue

        rows.append({
            'materials': element.attrib['materials'],
            'path': Path(library) / element.attrib['path'],
            'library': library,
        })

df = pd.DataFrame(rows)
df

2. Extract MT numbers from HDF5 files¶

def get_reactions(filename):
    with h5py.File(DATA_ROOT / filename) as ff:
        nuclide_name = list(ff.keys())[0]
        reactions = ff[nuclide_name]['reactions']

        mts = []
        for name in reactions:
            if name.startswith('reaction_'):
                mt = int(name.removeprefix('reaction_'))
                mts.append(mt)

    return sorted(mts)

df = df.assign(MTs=df['path'].map(get_reactions))
df

3. Rearrange dataframe to one MT per row¶

df = df.explode('MTs', ignore_index=True)
df = df.rename(columns={'MTs': 'MT'})
df['MT'] = df['MT'].astype(int)
assert not df.duplicated(['materials', 'MT', 'library']).any()
df

4. Build materials-MT-libraries catalog¶

catalog = (
    df.groupby(['materials', 'MT'], as_index=False)
      .agg(libraries=('library', ', '.join))
      .assign(
          reaction_name=lambda x: x['MT'].map(openmc.data.REACTION_NAME)
      )
)
catalog

5. Save df and catalog for future reference¶

df_to_save = df.copy()
df_to_save['path'] = df_to_save['path'].astype(str)
df_to_save.to_feather('0_2_catalog_nuclear_data_libraries_df.feather')

catalog.to_feather(
    '0_2_catalog_nuclear_data_libraries_catalog.feather',
)

Feather in context.

Format	Size	Speed	Advantages / limitations
Parquet	Small	Fast	Preserves data types, compression, cross-language support
Pickle	Variable	Fast	Convenient Python serialization, but unsafe from untrusted sources
CSV	Larger	Slower	Human-readable, universal, loses type information
Feather	Small	Very fast	Good for quick DataFrame exchange
SQLite	Variable	Query-dependent	Useful for SQL searches without loading the full catalog

Cataloging nuclear data libraries

1. Read cross_sections.xml from each library¶

2. Extract MT numbers from HDF5 files¶

3. Rearrange dataframe to one MT per row¶

4. Build materials-MT-libraries catalog¶

5. Save df and catalog for future reference¶

1. Read `cross_sections.xml` from each library¶