Every nuclide

INCOGNITA · data explorer

The INCOGNITA engine is a port of TALYS-2 (A. J. Koning, S. Hilaire, S. Goriely) and uses the TALYS nuclear-structure database. What INCOGNITA adds is listed, switch by switch, at the end of the teaching page.

Chart of nuclides

Drag to pan, scroll or pinch to zoom. Click a square to open it: the nuclide's details appear below the chart. Protons up, neutrons across.

The engine's predicted neutron cross sections for every nuclide in the library, beside the curated measurements of each nuclide. Pick a nuclide on the chart, or search for one.

The curves are library v0.2 (October 2026), made with the same engine version that is scored on the Scoreboard (the engine of 22 October 2026). They are not TERRA, the Scoreboard's data product that is fitted to measurements. Every curve is the engine's own prediction: none was fitted to the measurements shown beside it, except that a nuclide marked anchored has its curve scaled by a single factor so that its stellar average at 30 keV equals the measured value. The engine's general settings were chosen once, for all nuclides, using many measurements (these among them), so a match on a measured nuclide is a fair check of the model, though not a fully independent one. Most nuclides have never been measured (extrapolated), and their bands are wide.

Moderators: thermal scattering

Slow neutrons scatter off whole molecules and crystal lattices, not free atoms. INCOGNITA's own scattering laws, S(α,β), are built from our own physics: molecular dynamics for water, DFT phonons for beryllium. No evaluated library goes in. Both tie ENDF/B-VIII.1 on criticality benchmarks; neither is shown to be better.

MaterialPassesLosses
H in H₂O
molecular dynamics
12/12 criticality cases within ±35 pcm of VIII.1 (6 held out, mean +38 ± 14 pcm); total cross section C/E 1.004–1.036 (VIII.1 1.000–1.037)diffusion 8–11 % slow; +3–4 % above measurement below 10 meV (VIII.1 too); table ends at 4.9 eV
Be metal
DFT phonons
14/14 ICSBEP Be cases within 3σ, Δk vs VIII.1 −17 ± 17 pcm; ultracold up-scattering C/E 0.93double-differential spectra R 0.57 (VIII.1 0.58); Bragg-edge positions fail, VIII.1 too; C/E 0.76 below 200 K
Graphite—not released: every own law so far fails a pre-registered gate

Download the ENDF-6 tapes from library/tsl-v0.1, where the README lists every gate.

Run it yourself

Everything on this page comes from the open repository: the engine, the library and the curated measurements. You need Linux or macOS, Python 3.12 and uv.

git clone https://github.com/seth-michael-weaver/incognita.git && cd incognita
uv sync --extra dev
uv run python -m incognita.library summary --version v0.2        # every nuclide by how it is known
uv run python -m incognita.library show Au-197 --version v0.2    # one capture curve with its 68 / 95 % bands

Running the engine itself also needs the TALYS structure database; the README explains how to fetch it.

Data access. Everything shown here is in the repository under library/v0.2: capture curves with 68 / 95 % bands (capture_v02.parquet), Maxwellian averages (macs30_v02.csv, macs_kt_v02.csv), stellar rates (rates_v02.csv, reaclib_v02.dat) and the (n,2n), (n,p), (n,α) channels (channels_v02.parquet, grid_np_v02.parquet), checked by SHA256SUMS. For one nuclide, the CSV and JSON downloads above are enough. The curated measurements are in data/curation; they come from EXFOR (IAEA Nuclear Data Section) and keep their EXFOR ids. Library data are CC-BY-4.0. The earlier version, v0.1, stays in library/v0.1; CHANGELOG.md in library/v0.2 says what changed.

seth-michael-weaver/incognita on GitHub →

Sources: predictions from library v0.2 (library/v0.2); capture measurements from EXFOR as curated in data/curation; (n,2n), (n,p) and (n,α) measurements from the benchmark rows in incognita/bench. Measurements the curation excluded are hidden unless you ask for them. The optional dashed overlay of evaluated libraries (ENDF/B-VIII.1, JEFF-4.0, TENDL-2025) is built by scripts/build_site_overlay.py for display only; INCOGNITA never reads it.