Evaluated nuclear data · ENDF/B-VIII.1

Modern Nuclear Data Processing

Processing and comparing evaluated nuclear data with open tools such as NJOY and AMPX, with the code and results in public repositories.

Explore the covariances → Read the papers

Ian H · code and data on gitlab.com/hardtohit10

The ENDF/B-VIII.1 neutron sublibrary on the chart of the nuclides (protons Z against neutrons N): 558 evaluations, 266 with covariance data, 201 with covariances added from the BLO low-fidelity project or an older ENDF/B release. Hover over a nuclide; click a highlighted one to see its matrices.
467nuclides with covariances, processed by NJOY and AMPX (266 from ENDF/B-VIII.1)
10154covariance matrices to browse and compare, from the NJOY and AMPX libraries
3 pcmagreement of the two routes in the keff uncertainty of three benchmarks
35 yearsof Livermore photon data compared, element by element
Correlation map of U-238 fission in SCALE's 56 groups: strongly correlated blocks in red along the diagonal and across the thermal range
New · interactive

Covariance explorer

Every matrix of both processed covariance libraries (release v1.2), NJOY and SCALE's own AMPX, nuclide by nuclide: correlation maps with values on hover, relative uncertainties, a CSV download of any matrix, and a compare mode that maps where the two libraries differ.

Open the explorer →
Jump to: U-235U-238Pu-239H-1O-16Fe-56

Quote of the week

Week of 5 October 2026

The ENDF/B library is not directly used in design calculations. It must be processed first by programs such as NJOY to get either multigroup or Monte Carlo libraries.

— Charles Dunford, “Some CSEWG Recollections”, in A CSEWG Retrospective (BNL-52675), Brookhaven National Laboratory, 2001, p. 39 (PDF)

A new quote every Monday from CSEWG’s history and meeting minutes, chosen with the help of Claude (AI). Follow the source link for the full context.All quotes and sources →

Projects

Short summaries; the repositories hold the code, the full results and the comparisons made so far.

Bar chart of the leading contributions to the k-eff uncertainty of LEU-COMP-THERM-093 case 1 for the NJOY and AMPX ENDF/B-VIII.1 libraries and an ENDF/B-VII.1 library

ENDF/B-VIII.1 covariances in SCALE's 56 groups

Complete · v1.2

The covariance data of the ENDF/B-VIII.1 neutron sublibrary, processed into SCALE's 56-group structure by two independent routes: NJOY2016 (ERRORR, driven by SANDY) and SCALE's own AMPX (PUFF-IV), as COVERX libraries and text files for data adjustment.

  • All 266 ENDF/B-VIII.1 evaluations with covariance data processed by both routes, plus 9 nuclides from older ENDF/B releases and 192 from the BLO low-fidelity covariances, made the way SCALE made its own entries (v1.2); every change to the evaluated data recorded.
  • Sandwich uncertainties from the two libraries agree within 3 pcm, and are 26–36 % lower than with ENDF/B-VII.1.
  • Every file loads in JANIS 4.0; problems found in the evaluations and in both codes are documented.

NJOY2016 · SANDY · AMPX (SCALE 7.0 beta) · JANIS

Two maps of the relative change of the total photon cross section against photon energy and Z, for EPDL89 to EPDL97 and EPDL97 to EPICS2017

Photon interaction data, 1989–2023

Complete

How the photo-atomic cross sections of the Livermore EPDL/EPICS library changed, element by element, across five releases: EPDL89, EPDL97, EPICS2014, EPICS2017 (ENDF/B-VIII.0) and EPICS2023 (ENDF/B-VIII.1).

  • Only two real changes in 35 years: EPDL89 → EPDL97 and EPDL97 → EPICS2017.
  • Away from absorption edges, the total above 1 keV differs from EPICS2023 by a median 0.39 % (EPDL89) and 0.10 % (EPDL97).
  • EPICS2017 moved almost every subshell edge (K edges by a median 0.34 %), changing the data near edges by factors of several.

Own ENDF MF23 reader (tested) · Python

Table of processing jobs with the AMPX module and the NJOY2016 module for each: POLIDENT and RECONR, BROADEN and BROADR, PURM and PURR, Y12 and THERMR, PLATINUM and ACER

A SCALE continuous-energy library from ENDF/B-VIII.1 with AMPX

In progress

A SCALE continuous-energy library (the HDF5 format of SCALE 6.3/7) for CANDU lattice physics, processed with AMPX: D₂O, H₂O and UO₂ thermal scattering at their tabulated temperatures, 15 temperatures up to 2500 K for the other nuclides, and probability tables and DBRC data for U-238 and Pu-239.

  • O-16 and H-2 bound in D₂O are complete; their cross sections agree with NJOY-processed data to 0.02–0.03 % (median).
  • AMPX's broadening accumulates error when temperatures are broadened in sequence; broadening each temperature from 0 K fixes it.
  • The D₂O kernel above 600 K is 0.5–1.5 % high, traced to a narrow quasi-elastic peak in the evaluation at those temperatures (open issue).

AMPX (SCALE 7.0 beta) · NJOY-processed reference data · Python

Papers

Journal-style write-ups of the projects. Their author is Claude, an AI model by Anthropic; they are preprints and have not been peer reviewed.

Cover of the preprint on the similarity of a CANDU-6 lattice cell to criticality benchmarks for k-infinity and coolant void reactivity

Similarity of a CANDU-6 lattice cell to criticality benchmarks

k∞ and coolant void reactivity, with continuous-energy sensitivities from OpenMC and ENDF/B-VIII.1 covariances · preprint by Claude · not peer reviewed

 Cover of the preprint on the nuclear-data uncertainty of k-eff for 2,843 ICSBEP configurations with ENDF/B-VIII.1 covariances

Nuclear-data uncertainty of keff for 2,843 ICSBEP configurations

DICE sensitivities with ENDF/B-VIII.1 covariances, NJOY and AMPX, against ENDF/B-VII.1 · preprint by Claude · not peer reviewed

 Cover of the preprint on the ENDF/B-VIII.1 neutron covariances in the SCALE 56-group structure

ENDF/B-VIII.1 neutron covariances in the SCALE 56-group structure

Two processing routes, NJOY and AMPX, and their effect on criticality uncertainties · preprint by Claude · not peer reviewed

 Cover of the preprint on thirty-five years of the Livermore photon interaction library

Thirty-five years of the Livermore photon interaction library

Element-by-element changes in the total photon cross section from EPDL89 to EPICS2023 · preprint by Claude · not peer reviewed

How the work is done

Two processing routes

NJOY and AMPX process the same evaluations, and their results are compared.

Every change recorded

Any modification of the evaluated data is logged with its reason, so the libraries can be traced back to ENDF/B-VIII.1.

Open, not yet reviewed

Code under the MIT licence and results in public repositories. None of the results has been independently reviewed.