Start here
An ACTINV calculation is a strict JSON problem file: nuclear-data paths, a material, a group flux, an irradiation/cooling schedule, and the outputs you want. Validate the file, run it, then retain the result and its input certificate together.
Source, wheel, standalone command, Python-version, and nuclear-data setup guidance.
Validate and run one problem through the command line or the matching Python API.
Every accepted field, selector, unit, schedule rule, and complete example.
Calculation model
ACTINV evolves a nuclide inventory under reaction and decay rates, then evaluates requested responses at each reported step. The command line and Python interface call the same Rust implementation.
- Library preparation.Supported ENDF-6 evaluations are parsed, resonance ranges reconstructed, broadened to the requested temperature where applicable, and collapsed onto the selected group structure. Each source and generated artifact is hash-identified.
- Network assembly.Material nuclides, activation products, decay chains, fission yields, and explicit leakage states form the calculation network. Unsupported or missing information is recorded rather than guessed.
- Time evolution.CRAM-16 or CRAM-48 advances ordinary irradiation, cooling, and arbitrary pulse histories. Exposure selects trace or coupled treatment, and exact reachability plus bounded rate pruning keeps the solved network focused.
- Response evaluation.Inventories feed activity, heat, photons, pathways, radiological responses, and selected first-order covariance propagation without changing the underlying state solution.
Inputs and nuclear data
ACTINV deliberately does not bundle nuclear data. Users choose the evaluated files appropriate to their analysis and retain the source, terms, and SHA-256 identity. The software can build deterministic activation libraries from supported TENDL and EAF evaluations and a separate MF=33 covariance sidecar.
- Incident particles.Neutron, proton, deuteron, and alpha activation are supported. Triton, helion, and gamma activation are outside v1.0.
- Decay and yields.Evaluated decay records and fission-yield files are explicit inputs. Products absent from configured data remain named leakage; ACTINV does not invent missing modes or yields.
- Flux.Supply group values directly or import documented subsets of OpenMC statepoints, MCNP MESHTAL/MCTAL, or FISPACT flux files. Physical normalization is always explicit.
- Radiological tables.Clearance, waste, ingestion, and inhalation coefficients are user-supplied and hash-pinned. ACTINV does not select a jurisdiction, edition, scenario, chemical form, aerosol class, or margin.
What a result contains
- State.Nuclide populations and activity at each requested time, with explicit step, exposure, mode, and pruning records.
- Energy and sources.Total and alpha/beta/gamma decay heat, evaluated photon lines or group sources, and OpenMC/MCNP photon-source exports.
- Interpretation.Ranked production pathways, tritium and helium, and selected user-defined clearance, waste, ingestion, or inhalation responses with coefficient coverage.
- Uncertainty and omissions.Selected MF=33 cross-section bands, numerical-method comparison, missing coverage, leakage, and the ledger of every disposition the calculation can identify.
- Provenance.A certificate identifies software, specifications, nuclear data, response tables, and generated artifacts so a result can be reconstructed under configuration control.
Validation in brief
The repository contains executable controls and compact evidence for every release phase. Scientific accuracy is reported for the recorded benchmarks; implementation behavior and reproducibility are gated against frozen criteria.
Across 73 materials, ACTINV’s median geometric-mean C/E is 1.024; the distributed FISPACT-II reference is 1.009.
Complete TENDL-2025 n/p/d/alpha and EAF-2010 builds retain 1,849,479 rows with zero recorded target errors or silent fallbacks.
Independent parsing, analytic sensitivity, dense propagation, sampling, interface, and complete-corpus controls.
Radiological responses, primary tables, bounded input reliability, FNG/ITER activation, packaging, and closure.
The FNS and FNG/ITER comparisons apply to their recorded materials, inputs, histories, and responses. They are not general transport, shutdown-dose, safety, or regulatory qualifications.
Qualification boundary
ACTINV can be one version-controlled component in a formal analysis chain, but it does not approve that chain. The analyst and responsible organization still own input qualification, validation applicability, independent review, configuration management, uncertainty additions and margins, and any required regulatory acceptance.
- No transport.ACTINV does not calculate neutron or photon transport, criticality, shielding geometry, or thermal-hydraulic feedback. Its contact gamma dose is a screening proxy.
- Independent mesh cells.Mesh calculations do not feed composition, temperature, or flux changes between cells; spatial source interpretation remains external.
- Scoped uncertainty.MF=33 bands cover retained cross-section covariance only. Decay, yield, flux, composition, response coefficient, geometry, and model uncertainties are outside that band.
- Infinite-dilution unresolved treatment.Finite-dilution self-shielding, probability tables, and Bondarenko treatment are not implemented in v1.0.
Contributing and licence
ACTINV is offered under your choice of the MIT or Apache-2.0 licence. Contributions are licensed the same way and use the Developer Certificate of Origin rather than a contributor licence agreement. Physics or data-handling changes require a regression control and an append-only evidence record.