Revisit HDF5 and extract temperatures

Explore the structure of OpenMC neutron HDF5 files, inspect groups and attributes, extract temperatures, and compare temperature availability across nuclear-data libraries.

Revisit HDF5 and extract temperatures
├── 1. Inspect HDF5 structure
├── 2. Inspect attributes
└── 3. Get metadata

import h5py
import pandas as pd

from openmc_workbench_config import DATA_ROOT

df = pd.read_feather('0_2_catalog_nuclear_data_libraries_df.feather')
df

1. Inspect HDF5 structure¶

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

        return {
            (name, type(obj).__name__)
            for name, obj in nuclide.items()
        }

name_types = set()
for filename in df['path'].unique():
    name_types.update(get_name_types(filename))
name_types

{('energy', 'Group'),
 ('fission_energy_release', 'Group'),
 ('kTs', 'Group'),
 ('reactions', 'Group'),
 ('total_nu', 'Group'),
 ('urr', 'Group')}

Group name	Definition
`kTs`	Temperatures expressed as $k_BT$ , in eV
`energy`	Incident neutron energy grids at different temperatures
`reactions`	Nuclear reaction data, organized by MT number
`urr`	Unresolved resonance probability tables, if available
`total_nu`	Total number of neutrons per fission, when present
`fission_energy_release`	Fission energy release components, when present

source:

2. Inspect attributes¶

def get_attrs(filename):
    with h5py.File(DATA_ROOT / filename) as ff:
        nuclide = next(iter(ff.values()))
        return set(nuclide.attrs)

attrs = set()
for filename in df['path'].unique():
    attrs.update(get_attrs(filename))
attrs

{'A', 'Z', 'atomic_weight_ratio', 'metastable'}

3. Get metadata¶

def get_temperatures(filename):
    with h5py.File(DATA_ROOT / filename) as ff:
        nuclide = next(iter(ff.values()))

        temperatures = sorted(
            nuclide['kTs'],
            key=lambda t: int(t.removesuffix('K')),
        )

        return temperatures

# Read each unique HDF5 file only once, then map the results
# back to every row of the catalog.
paths = df['path'].drop_duplicates()
temperatures = paths.map(get_temperatures)

path2temperature = dict(zip(paths, temperatures))
df['temperatures'] = df['path'].map(path2temperature)
df

All nuclear-data libraries in our collection provide cross sections at six temperatures: 250, 294, 600, 900, 1200, and 2500 K.

The exception is FENDL-3.2c, which provides data only at 294 K.

This is because the distributed FENDL-3.2c ACE files were processed at 293.6 K (rounded to 294 K by OpenMC). Additional temperatures would require Doppler broadening.

df['temperatures'].map(tuple).unique()

array([('250K', '294K', '600K', '900K', '1200K', '2500K'), ('294K',)],
      dtype=object)

reference_temperatures = set(df['temperatures'].iloc[0])

different = df['temperatures'].map(
    lambda temperatures: set(temperatures) != reference_temperatures,
)

df.loc[different, 'library'].unique()

<ArrowStringArray>
['fendl-3.2c']
Length: 1, dtype: str