PyEPICS

Python library for reading and converting EPICS (Electron Photon Interaction Cross Sections) nuclear data.

PyEPICS parses EEDL, EPDL, and EADL files from the LLNL EPICS 2025 database (in ENDF-6 format) and converts them into structured HDF5 files suitable for Monte Carlo transport codes such as MC/DC.

---

Package Structure

PyEPICS/
├── pyepics/                         # Source code
│   ├── __init__.py              # Public API
│   ├── client.py                # High-level element query API (EPICSClient)
│   ├── plotting.py              # Optional plotting helpers (matplotlib)
│   ├── cli.py                   # Batch processing CLI
│   ├── exceptions.py            # Custom exception hierarchy
│   ├── pyeedl_compat.py         # Backward-compatibility shim for legacy PyEEDL code
│   ├── readers/
│   │   ├── base.py              # Abstract BaseReader
│   │   ├── eedl.py              # EEDLReader — electron data
│   │   ├── epdl.py              # EPDLReader — photon data
│   │   └── eadl.py              # EADLReader — atomic relaxation data
│   ├── models/
│   │   └── records.py           # Typed dataclass models (EEDLDataset, EPDLDataset, EADLDataset)
│   ├── converters/
│   │   ├── hdf5.py              # High-level API (create_raw_hdf5, create_mcdc_hdf5)
│   │   ├── raw_hdf5.py          # Raw HDF5 writer (original grids, breakpoints)
│   │   └── mcdc_hdf5.py         # MCDC HDF5 writer (common grid, PDFs, interpolated)
│   ├── utils/
│   │   ├── constants.py         # Physical constants, periodic table, MF/MT tables
│   │   ├── parsing.py           # ENDF format parsing helpers
│   │   └── validation.py        # Post-parse validation routines
│   └── io/
│       └── download.py          # Dataset downloader from LLNL
├── data/                            # All data (gitignored)
│   ├── endf/                    # Downloaded ENDF source files
│   │   ├── eedl/                # EEDL electron data
│   │   ├── epdl/                # EPDL photon data
│   │   └── eadl/                # EADL atomic relaxation data
│   ├── raw/                     # Generated raw HDF5 files
│   │   ├── electron/
│   │   ├── photon/
│   │   └── atomic/
│   └── mcdc/                    # Generated MCDC HDF5 files
│       ├── electron/
│       ├── photon/
│       └── atomic/
└── tests/
    ├── conftest.py              # Shared pytest fixtures
    ├── test_eedl.py             # EEDL reader + parsing + validation tests
    ├── test_epdl.py             # EPDL reader tests
    ├── test_eadl.py             # EADL reader tests
    ├── test_hdf5.py             # Legacy HDF5 converter tests
    ├── test_pipeline.py         # Raw + MCDC pipeline tests
    ├── generate_report.py       # PDF regression-test report generator
    ├── fixtures/                # Reference validation data
    └── reports/                 # Generated regression reports

Architecture

The package follows a strict layered dependency graph:

utils ← models ← readers ← converters (raw_hdf5 / mcdc_hdf5)

Layer	Responsibility
utils	ENDF parsing helpers (float_endf, parse_mf26_mt525, …), validation routines, physical constants, MF/MT mapping tables
models	Typed dataclass records (EEDLDataset, EPDLDataset, EADLDataset) — the sole output of readers and sole input to converters
readers	EEDLReader, EPDLReader, EADLReader — parse ENDF files via the endf library and return model instances
converters	Two-step conversion: raw_hdf5 (full-fidelity) and mcdc_hdf5 (transport-optimised)
client	High-level EPICSClient for querying/comparing element properties
plotting	Optional visualisation helpers (requires matplotlib)
io	Dataset download from LLNL
cli	Batch processing for the full pipeline

Installation

# From PyPI (when published)
pip install pyepics-data

# With all optional dependencies
pip install "pyepics-data[all]"

# From source (editable, for development)
git clone https://github.com/melekderman/PyEPICS.git
cd PyEPICS
pip install -e ".[dev]"

See INSTALL.md for full details on optional extras, developer setup, and platform notes.

---

Data Pipeline

PyEPICS follows a three-step pipeline:

LLNL website                    download
    │                           ─────────────────────►
    ▼
data/endf/{eedl,epdl,eadl}/    raw ENDF files (.endf)
    │                           ─────────────────────►
    ▼
data/raw/{electron,photon,     raw HDF5 (original grids, breakpoints)
          atomic}/              for external users
    │                           ─────────────────────►
    ▼
data/mcdc/{electron,photon,    MCDC HDF5 (common grid, PDFs)
           atomic}/             for transport codes

Step 1: Download ENDF Data from LLNL

Download all three EPICS libraries (EEDL, EPDL, EADL) from LLNL Nuclear Data:

# Download all libraries
python -m pyepics.cli download

# Download only electron data (EEDL)
python -m pyepics.cli download --libraries electron

# Download to a custom directory
python -m pyepics.cli download --data-dir /path/to/data

This creates three directories with .endf files:

data/endf/eedl/   ← EEDL.ZA001000.endf, EEDL.ZA002000.endf, ... (Z=1–100)
data/endf/epdl/   ← EPDL.ZA001000.endf, ... 
data/endf/eadl/   ← EADL.ZA001000.endf, ...

Step 2: Create Raw HDF5 Files

Raw files preserve every piece of information from the ENDF source: original energy grids, breakpoints, and interpolation law codes. These are intended for external users who need full-fidelity data.

# Create raw HDF5 for all libraries (Z=1–100)
python -m pyepics.cli raw

# Only electron data
python -m pyepics.cli raw --libraries electron

# Process a specific Z range
python -m pyepics.cli raw --z-min 1 --z-max 30

# Overwrite existing files
python -m pyepics.cli raw --overwrite

Output directories:

data/raw/electron/   ← H.h5, He.h5, ..., Fe.h5, ... (electron)
data/raw/photon/     ← H.h5, He.h5, ...              (photon)
data/raw/atomic/     ← H.h5, He.h5, ...              (atomic relaxation)

Step 3: Create MCDC HDF5 Files

MCDC files are optimised for transport codes. All cross sections are interpolated onto a common energy grid, angular distributions are compressed into (grid, offset, value, PDF) tables, and small-angle elastic scattering cosine PDFs are analytically computed.

# Create MCDC HDF5 for all libraries
python -m pyepics.cli mcdc

# Only electron data
python -m pyepics.cli mcdc --libraries electron

# Specific Z range
python -m pyepics.cli mcdc --z-min 26 --z-max 26   # Fe only

Output directories:

data/mcdc/electron/   ← H.h5, He.h5, ..., Fe.h5, ... (electron)
data/mcdc/photon/     ← H.h5, He.h5, ...              (photon)
data/mcdc/atomic/     ← H.h5, He.h5, ...              (atomic relaxation)

Full Pipeline (Raw + MCDC in One Step)

# Run raw + MCDC for all libraries, all elements
python -m pyepics.cli all

# Run everything but continue if an element fails
python -m pyepics.cli all --continue-on-error

# Full pipeline for first 30 elements only
python -m pyepics.cli all --z-min 1 --z-max 30 --overwrite

Python API

You can also use the pipeline functions directly from Python:

from pyepics import create_raw_hdf5, create_mcdc_hdf5

# Step 2: Raw HDF5
create_raw_hdf5("EEDL", "data/endf/eedl/EEDL.ZA026000.endf", "data/raw/electron/Fe.h5", overwrite=True)

# Step 3: MCDC HDF5
create_mcdc_hdf5("EEDL", "data/endf/eedl/EEDL.ZA026000.endf", "data/mcdc/electron/Fe.h5", overwrite=True)

# Download programmatically
from pyepics.io.download import download_library, download_all
download_library("eedl")      # downloads to ./data/endf/eedl/
download_all()                 # downloads all three

---

Quick Start

High-Level Client API

from pyepics import EPICSClient

client = EPICSClient("data/endf")

# Query a single element (by symbol, name, or Z)
fe = client.get_element("Fe")
print(fe.Z, fe.symbol)            # 26, "Fe"
print(fe.binding_energies)         # {'K': 7112.0, 'L1': 844.6, ...}
print(fe.electron_cross_section_labels)  # ['xs_tot', 'xs_el', ...]

# Compare multiple elements
rows = client.compare(["Fe", "Cu", "Au"])

# DataFrame output (requires pandas)
df = client.compare_df(["Fe", "Cu", "Au"])

# Get raw cross-section arrays
energy, xs = client.get_cross_section("Fe", "xs_tot")

Plotting (requires matplotlib)

from pyepics.plotting import plot_cross_sections, compare_cross_sections

plot_cross_sections(client, "Fe")
compare_cross_sections(client, ["C", "Fe", "Au"], "xs_tot")

Low-Level Reader API

from pyepics import EEDLReader

# Parse an EEDL file directly
reader = EEDLReader()
dataset = reader.read("data/endf/eedl/EEDL.ZA026000.endf")
print(dataset.Z, dataset.symbol)  # 26, "Fe"
print(list(dataset.cross_sections.keys()))  # ['xs_tot', 'xs_el', 'xs_lge', ...]

Exception Hierarchy

All library exceptions inherit from PyEPICSError:

PyEPICSError
├── ParseError          # Malformed ENDF content
├── ValidationError     # Failed physics checks (e.g. negative cross sections)
├── FileFormatError     # Wrong file type or unrecognised filename
├── ConversionError     # HDF5 write failures
└── DownloadError       # Network errors (future)

Running Tests

pip install -e ".[dev]"
python -m pytest tests/ -v

Backward Compatibility

A pyeedl_compat shim re-exports legacy API symbols for backward compatibility:

from pyepics.pyeedl_compat import PERIODIC_TABLE, float_endf, SUBSHELL_LABELS

Data Sources

PyEPICS uses the following authoritative data sources:

Data Type	Source	Reference
Electron cross sections	EEDL (Evaluated Electron Data Library)	LLNL EPICS 2025
Photon cross sections	EPDL (Evaluated Photon Data Library)	LLNL EPICS 2025
Atomic relaxation	EADL (Evaluated Atomic Data Library)	LLNL EPICS 2025
Binding energies	EADL via ENDF-6 format (not NIST)	Parsed from EADL .endf files
Physical constants	NIST CODATA 2018	NIST CODATA

Note: Binding energies are sourced from EADL (parsed from ENDF files), not from the NIST X-Ray Transition Energies database. The reference validation data in tests/fixtures/reference_binding_energies.csv is extracted from EEDL ENDF files and compared against the PyEPICS-parsed values to ensure round-trip consistency.

Acknowledgements

This work was supported by the Center for Advancing the Radiation Resilience of Electronics (CARRE), a PSAAP-IV project funded by the Department of Energy, grant number: DE-NA0004268.

License

Please see LICENSE for details.