← Modern Nuclear Data Processing · AMPX–NJOY processing map
ENDF/B-VIII.1 · SCALE continuous energy · AMPX
A SCALE continuous-energy neutron library (the HDF5 format of SCALE 6.3/7) processed from ENDF/B-VIII.1 with AMPX, for CANDU lattice physics. This page follows the processing run and reports what has been compared with NJOY-processed data so far.
Most of this work, including this page, was generated with AI (Claude, Anthropic) under the direction of Ian H; it has not been independently reviewed.
Snapshot of 15 October 2026, 09:19 EDT. The processing run is still going: 223 of 557 evaluations are processed, 10 are running and 324 are waiting. Only the nuclides under Comparisons with NJOY have been compared with other data; the rest have only run without errors. No library files are released yet, and the numbers on this page will change as the run continues.
The run started on 7 October 2026 and has run 10 AMPX jobs at a time since that afternoon, each job one evaluation, starting a new job only while more than 3.5 GB of memory is free. It has finished about 31 evaluations per day while the machine was up. At that pace the remaining 334 would take about 11 more days of running; 194 of them have an unresolved range, whose probability tables take most of the time, so this is only a rough guide.
In the hatched periods the machine was off or asleep. Each time, the running jobs were lost and started again from scratch, because their working directories are temporary. Since 14 October a startup service restarts the run on its own. The processing driver is running.
Time since the job (re)started. Most of a job's time goes into the unresolved-range probability tables (PURM).
The 557 ENDF/B-VIII.1 neutron evaluations of the run, by atomic number. Hover over a nuclide for its MAT number, SCALE ID, finish time and run time. The same list is in ampx_ce_status.csv.
The reference is the official OpenMC ENDF/B-VIII.1 library, processed with NJOY2016, at the temperatures within 1 K of a library temperature (294, 600, 900, 1200 and 2500 K for most nuclides). Each figure is |AMPX/NJOY − 1| in per cent. The scripts and their full outputs are in the repository's results folder.
On the union of both energy grids, both interpolated linearly. Median: range over the temperatures; maximum: the largest at any temperature, and where it is. Thermal: at 0.0253 eV; resonance and fast: lethargy integrals over 0.5 eV–10 keV and 10 keV–20 MeV (largest difference over the temperatures).
| Nuclide | T (K) | Reaction | Points at 294 K | Median % | Maximum % | Thermal % | Resonance % | Fast % |
|---|---|---|---|---|---|---|---|---|
| O-16 | 293.6, 600, 900, 1200, 2500 | total | 3,979 | 0.017–0.025 | 32.75 at 2.89 MeV | 0.067 | 0.040 | 0.017 |
| elastic | 3,979 | 0.018–0.027 | 32.75 at 2.89 MeV | 0.067 | 0.040 | 0.013 | ||
| inelastic | 883 | 0.000 | 0.11 at 15 MeV | – | – | 0.000 | ||
| (n,2n) | 81 | 0.000 | 0.09 at 19.7 MeV | – | – | 0.013 | ||
| capture | 3,979 | 0.001–0.002 | 2.04 at 1.31 MeV | 0.100 | 0.108 | 0.022 | ||
| The largest point differences (total and elastic, up to 33 % at 2500 K) are at the narrow 2.889 MeV resonance: AMPX Doppler-broadens it, while the NJOY-processed reference was not broadened at MeV energies (its peak is the same at every temperature). The integral over that window agrees within 0.007 %. | ||||||||
| H-2 (D in D₂O file) | 293.6, 600 | total | 907 | 0.019–0.022 | 0.32 at 1.5 MeV | 0.002 | 0.033 | 0.030 |
| elastic | 907 | 0.019–0.022 | 0.32 at 1.5 MeV | 0.002 | 0.030 | 0.030 | ||
| (n,2n) | 47 | 0.000 | 0.14 at 7 MeV | – | – | 0.017 | ||
| capture | 907 | 0.049–0.051 | 0.95 at 23.2 eV | 0.002 | 0.168 | 0.010 | ||
| The point data above the thermal range of the bound D-in-D₂O file, at the two reference temperatures within 1 K of a library temperature. | ||||||||
| U-235 | 293.6, 600, 900, 1200, 2500 | total | 141,428 | 0.021 | 17.54 at 2.25 keV | 0.040 | 0.006 | 0.004 |
| elastic | 141,428 | 0.021–0.022 | 4.70 at 2.25 keV | 0.028 | 0.009 | 0.004 | ||
| inelastic | 66,884 | 0.000 | 69.28 at 77.4 eV | – | 0.000 | 0.003 | ||
| (n,2n) | 42 | 0.000 | 0.00 at 6.2 MeV | – | – | 0.000 | ||
| fission | 141,428 | 0.033–0.035 | 38.94 at 2.25 keV | 0.039 | 0.011 | 0.000 | ||
| capture | 141,428 | 0.035–0.036 | 39.87 at 2.25 keV | 0.050 | 0.007 | 0.000 | ||
| The largest differences are at 2.25 keV, the boundary of the resolved and unresolved ranges, where the two codes take different sides of the step, and for inelastic scattering next to its 77 eV threshold. In the unresolved range the point data are reconstructed on different grids. | ||||||||
| U-238 | 293.6, 600, 900, 1200, 2500 | total | 352,914 | 0.026–0.028 | 33.07 at 20 keV | 0.042 | 0.007 | 0.003 |
| elastic | 352,914 | 0.026–0.027 | 32.53 at 20 keV | 0.032 | 0.003 | 0.005 | ||
| inelastic | 370 | 0.001 | 0.11 at 860 keV | – | – | 0.004 | ||
| (n,2n) | 31 | 0.000 | 0.08 at 11 MeV | – | – | 0.000 | ||
| fission | 352,914 | 0.015–0.024 | 19.62 at 20 keV | 0.070 | 0.033 | 0.000 | ||
| capture | 352,914 | 0.091–0.122 | 47.68 at 20 keV | 0.071 | 0.015 | 0.045 | ||
| The largest differences are at 20 keV, the boundary of the resolved and unresolved ranges, where the two codes take different sides of the step. | ||||||||
| Am-242m | 293.6, 600, 900, 1200, 2500 | total | 5,527 | 0.027–0.029 | 94.13 at 20 MeV | 0.062 | 0.005 | 0.015 |
| elastic | 5,527 | 0.024–0.025 | 13.72 at 43 eV | 0.028 | 0.019 | 0.008 | ||
| (n,2n) | 57 | 0.000 | 0.09 at 18 MeV | – | – | 0.003 | ||
| fission | 5,527 | 0.029–0.032 | 11.41 at 43 eV | 0.066 | 0.007 | 0.036 | ||
| capture | 5,527 | 0.035–0.037 | 28.85 at 43 eV | 0.041 | 0.013 | 0.044 | ||
| The largest differences are at 43 eV, the boundary of the resolved and unresolved ranges, except for the total, whose 94 % is the zero at 20 MeV described under open issues. | ||||||||
At the incident energies both tables list, without the first (the end of the resolved range). Band average: the infinitely dilute cross section; self-shielded: the Bondarenko cross section at a background of 1 b, 10 b and infinity. Each cell: median (range over the temperatures) / maximum. Both codes sample resonance ladders at random (AMPX PURM, NJOY PURR), so differences of a few per cent at low backgrounds are of the size of the sampling noise. AMPX's table total holds only elastic, fission and capture, while NJOY's includes inelastic scattering, so totals are not compared here.
| Nuclide | Range | Reaction | Band average | s₀ = 1 b | s₀ = 10 b | s₀ = ∞ |
|---|---|---|---|---|---|---|
| U-235 | 2.25 keV – 25 keV 5 temperatures; NJOY 19 energies | elastic | 0.001 / 0.33 | 0.04–0.13 / 0.54 | 0.03–0.10 / 0.50 | 0.00 / 0.33 |
| fission | 0.030–0.084 / 1.28 | 0.22–0.65 / 2.13 | 0.15–0.48 / 1.65 | 0.03–0.08 / 1.28 | ||
| capture | 0.006–0.009 / 0.06 | 0.14–0.46 / 2.37 | 0.10–0.34 / 1.80 | 0.01 / 0.06 | ||
| U-238 | 20 keV – 150 keV 5 temperatures; NJOY 31 energies | elastic | 0.005–0.007 / 0.13 | 1.15–6.56 / 9.16 | 0.57–1.46 / 2.08 | 0.00–0.01 / 0.13 |
| capture | 0.000–0.001 / 0.00 | 0.28–2.04 / 5.08 | 0.12–0.38 / 0.63 | 0.00 / 0.00 | ||
| Am-242m | 43 eV – 27.3 keV 5 temperatures; NJOY 44 energies | elastic | 0.013–0.018 / 0.17 | 0.14–0.24 / 1.38 | 0.11–0.21 / 1.35 | 0.01–0.02 / 0.17 |
| fission | 0.033–0.055 / 0.90 | 0.60–0.82 / 1.93 | 0.49–0.75 / 1.86 | 0.03–0.05 / 0.90 | ||
| capture | 0.035–0.065 / 0.89 | 0.72–1.43 / 7.81 | 0.59–1.23 / 7.59 | 0.03–0.06 / 0.89 |
DBRC data (0 K elastic scattering, for the Doppler-broadened rejection correction) of U-238 against OpenMC's 0 K elastic cross section, median difference 0.0125 %, 99th percentile 0.688 %, lethargy integral over 1 eV–20 keV +0.001 %. The maximum is at 20 keV, the end of the resolved range.
D bound in D₂O, the thermal cross section against NJOY's processing of the same S(α,β) data, at the 17 temperatures of the evaluation; the last three columns are AMPX/NJOY − 1 in per cent at three energies. The maxima (2.5–2.9 % at every temperature) are near 1.4e-5 eV, from interpolating NJOY's coarse grid there. The highlighted rows (600–650 K) are discussed under open issues. O-16's free-gas thermal cross section agrees with the exact free-gas formula at all 15 temperatures: median 0.014–0.017 %, maximum 0.14 %.
| T (K) | Median % | Maximum % | 0.0253 eV | 0.1 eV | 1 eV |
|---|---|---|---|---|---|
| 283.6 | 0.272 | 2.50 | +0.23 | +0.19 | +0.09 |
| 293.6 | 0.261 | 2.51 | +0.19 | +0.14 | +0.03 |
| 300 | 0.236 | 2.48 | +0.15 | +0.14 | +0.03 |
| 323.6 | 0.186 | 2.50 | +0.09 | +0.12 | +0.03 |
| 350 | 0.153 | 2.53 | +0.11 | +0.10 | +0.00 |
| 373.6 | 0.108 | 2.59 | +0.02 | +0.07 | -0.03 |
| 400 | 0.080 | 2.65 | +0.02 | +0.07 | +0.00 |
| 423.6 | 0.061 | 2.69 | +0.06 | +0.05 | +0.00 |
| 450 | 0.053 | 2.76 | -0.02 | +0.07 | +0.00 |
| 473.6 | 0.053 | 2.83 | -0.02 | +0.05 | -0.03 |
| 500 | 0.063 | 2.87 | -0.02 | +0.00 | -0.03 |
| 523.6 | 0.062 | 2.90 | -0.02 | +0.00 | +0.00 |
| 550 | 0.071 | 2.95 | -0.02 | +0.07 | +0.00 |
| 573.6 | 0.083 | 2.88 | +0.00 | +0.12 | +0.06 |
| 600 | 0.289 | 2.81 | +0.26 | +0.35 | +0.43 |
| 623.6 | 0.577 | 2.75 | +0.56 | +0.83 | +1.07 |
| 650 | 0.821 | 2.68 | +1.07 | +1.48 | +1.90 |
He-4 was compared the same way at the start of the run (within 0.18 %); its output is recorded in the work log only.
OpenA zero at 20 MeV. In 63 of the 224 nuclide files finished by 14 October (among them Am-242m, Pu-239, Pu-240, Fe-56 and U-234), elastic scattering, and for Am-242m also fission and capture, falls to zero over the last 0.002 eV below 20 MeV, where the library repeats the 20 MeV point. The ENDF/B-VIII.1 data run on smoothly to 20 MeV. The interval is too narrow to affect transport results, but the module that introduces it has not been found yet.
AcceptedD in D₂O at 600–650 K. At these three temperatures the thermal cross section is 0.4–1.9 % above NJOY at 1 eV. The cause is the integration tolerance of AMPX's Y12 module: an independent integration of the same S(α,β) data matches NJOY within 0.03 %. Tightening the tolerance to 5e-4 halved the excess; tighter runs did not fit in the machine's memory. The file was accepted as it is. CANDU moderator (about 340 K) and coolant (up to 573.6 K) temperatures are not affected.
AcceptedNa-22 without probability tables. AMPX's PURM cannot sample Na-22's unresolved range, because the evaluation's File 3 background there is negative (about −0.04 b at 15 keV). Na-22, an activation product with a 2.6-year half-life, is processed with smooth cross sections in that range (15–100 keV).
FixedMetastable nuclides. PURM labels its probability tables with the ENDF ZA, which carries no metastable flag, so the first Am-242m run found no tables and failed after 35 hours. The processing chain now gives PURM_UP and PLATINUM the ENDF ZA; Am-242m was rerun on 9 October, and the 11 other metastable nuclides with an unresolved range use the corrected inputs.
FixedBroadening in sequence. AMPX's BROADEN accumulates error when one call broadens several temperatures in turn; each temperature is now broadened directly from 0 K.
KnownDifferences between the codes, not errors. The largest point differences sit at the end of the resolved range, where the codes take different sides of the step, and at O-16's 2.889 MeV resonance, which AMPX broadens and the NJOY reference does not. In U-235's unresolved range the two codes reconstruct the point data on different grids: locally up to 24 % apart, with integrals within 0.11 %.
| AMPX | SCALE 7.0 beta 3 public source; chain per nuclide: POLIDENT (0 K data), BROADEN (each temperature from 0 K), PURM and PURM_UP (probability tables), Y12 (secondary distributions and thermal scattering), JAMAICAN, PLATINUM (nuclide file). See the AMPX–NJOY processing map. |
| Tolerance | 1e-3 for the cross sections (POLIDENT, BROADEN, the Y12 passes for thermal cross sections; 5e-4 for D in D₂O at 600–650 K); 1e-2 for Y12's secondary distributions, 1e-3 for JAMAICAN's thinning |
| Temperatures | 293.6, 350, 400, 450, 500, 550, 600, 700, 800, 900, 1000, 1200, 1500, 1800 and 2500 K; bound moderators at their evaluations' own temperatures (D and O in D₂O: 17 from 283.6 to 650 K) |
| Thermal scattering | free gas for every nuclide; S(α,β) for D and O in D₂O, H in H₂O, and U and O in UO₂ (D in D₂O done; the others follow the free-gas run) |
| Unresolved range | probability tables, 20 bands, for every evaluation with an unresolved range (350 of them; Na-22 excepted) |
| DBRC | 0 K elastic data for U-238 and Pu-239 |
| Files | each nuclide compressed without loss (gzip inside HDF5, to about 40 % of the size) and kept only if h5diff finds the contents identical; SCALE's data loader read the compressed O-16 file with the same results as the uncompressed one |