← Modern Nuclear Data Processing
ENDF/B-VIII.1 covariances · SCALE 56 groups · release v1.0
The NJOY-processed 56-group covariance library can be opened in JANIS, the NEA's nuclear data display program. This page gives an example of loading the files, and summarizes the checks made with JANIS. To look at the same matrices in a web browser instead, use the covariance explorer.
JANIS reads covariances in the NJOY ERRORR and BOXER formats, so use the ERRORR archive of release v1.0 (9.7 MB). JANIS cannot open the SCALE COVERX libraries in the same release.
curl -LO https://gitlab.com/api/v4/projects/86934647/packages/generic/endfb81-cov56/1.0.0/endfb81-cov56-v1.0-njoy-errorr.tar.xz
curl -LO https://gitlab.com/api/v4/projects/86934647/packages/generic/endfb81-cov56/1.0.0/SHA256SUMS
sha256sum --ignore-missing -c SHA256SUMS
tar -xJf endfb81-cov56-v1.0-njoy-errorr.tar.xz
| Folder or file | Contents | JANIS format |
|---|---|---|
errorr33/ | cross-section covariances (MF33, including the resonance-parameter covariances of MF32), one tape per nuclide: 266 | ERRORR |
errorr31/ | ν̄ covariances (MF31): 73 | ERRORR |
boxer35/ | fission-spectrum covariances (MF35) as BOXER copies: 67 | BOXER |
errorr35/ | the same MF35 data as ERRORR tapes (for other codes; JANIS reads only MF33 from ERRORR) | — |
cross_material_family.cov56.errorr33 | all materials linked by cross-material blocks, in one tape | ERRORR |
JANIS 4.1 (build 5.4.652) fails on every ERRORR tape, including NJOY's own test outputs. Use JANIS 4.0, available from the NEA as a zip; it runs on Java 8.
cd janis-4.0/software
java -Duser.language=en -Duser.country=US -Xmx2g -jar Janis.jar
ERRORR; the tapes are NJOY ERRORR output, not ENDF files.errorr33/n-092_U_238.cov56.errorr33 and press Open.
n-092_U_238.cov56.errorr33. Blocks of strong correlation follow the energy ranges of the evaluation's covariance data.errorr31/<nuclide>.cov56.errorr31, format ERRORR.boxer35/<nuclide>.cov56.boxer35, format BOXER.cross_material_family.cov56.errorr33. JANIS draws a cross-material block only when the partner material is in the same file, with the materials in ascending MAT order, which is why the family file exists. Opening n-092_U_235.cov56.errorr33 alone shows only U-235's own matrices.Every file was loaded and displayed by JANIS 4.0's own code (the same import and plotting path as File › Open), run headless. The map and the table were built for every matrix.
| Files | Cases | Result |
|---|---|---|
| MF33 tapes, MF31 tapes and MF35 BOXER files | 406 | 403 load and display completely. U-233, U-235 and Au-197 display their own matrices; their cross-material blocks need the partner files (next row). |
| Cross-material family tapes | 2 | All matrices display (1155 / 1155 and 204 / 204). |
| Total | 408 | 405 OK, 3 OK with their partners loaded; every matrix displays. |
Findings from the JANIS checks. JANIS flags correlations larger than 1 in 56 matrices. That is the same set as the project's own quality report, and comes from the evaluations; those matrices are left unchanged. JANIS also imports nothing from a tape that contains NaN, which is how a problem in the W-182/183/184 tapes was found; those tapes were repaired, and the repair is recorded.
The full listing is janis_check.txt, made with janis_check.py and JanisLoadCheck.java.