ACTINV v1.1.0

ACTINV v1.1.0 — release notes

ACTINV 1.1.0 adds shielded cross sections, schedule-driven feed/removal, linear-regime flux estimation,
damage observables, per-channel uncertainty breakdowns and mesh-scaling controls. The separately
versioned nuclear-data catalog remains data-v1.0.0.

Upgrade the Python package with:

python -m pip install --upgrade actinv

Rust CLI users can install this exact release with:

cargo install --locked --force actinv-cli --version 1.1.0

What users gain

• Finite-dilution self-shielding. self_shielding on actinv-spec-1/actinv-mesh-spec-1 applies
  unresolved-range Bondarenko shielding from hash-pinned actinv-shield-table-1 tables built by
  actinv build-shielding. Infinite dilution remains the explicit default; resolved-region pointwise
  shielding, escape corrections and probability-table transport remain out of scope.
• Schedule-driven feed and removal. Constant feed rates and first-order removal constants on
  actinv-spec-1 steps, with a removed sink state and named-ledger reporting.
• actinv reverse / actinv.reverse. Linear-regime flux estimation — recovers the flux
  normalization or segment multipliers from measured activities by weighted least squares, with
  residuals, chi-square and explicit method limits.
• Damage observables. A damage spec section folds hash-pinned actinv-damage-table-1 tables
  (built by actinv build-damage from TENDL MF=3/MT=444) into damage-energy and NRT-dpa outputs.
• Per-channel uncertainty breakdown. Propagated response bands now carry a per-channel split across
  cross-section (MF=33), decay-constant (MF=8/MT=457) and independent fission-yield (MF=8/MT=454)
  uncertainty plus a named uncovered remainder.
• Mesh scaling. Signature-keyed workload grouping, cell_result_fields selection, the post-hoc
  memory_limit_bytes guard, and output-as-checkpoint resume that is byte-identical to an
  uninterrupted run modulo footer timing.

Known data limitation

The shipped data-v1.0.0 activation artifacts are built from TENDL-2025, which carries an
upstream-confirmed TALYS channelsout.f90 defect: thermal (n,p) cross sections leak into (n,2n)
ground-state records. The defect affects 86 neutron evaluations in the shipped catalog, including
sources for 17 of 49 IRDFF-II benchmark targets and five of six dosimetry-critical isotopes
(Ni-58, Nb-93, Ag-109, In-113, Au-197). The upstream fix is committed for the next TENDL release;
ACTINV will re-qualify and re-publish corrected data when it lands.

This release's library builder fails closed on the relevant construction defect classes and emits a
state_catalog recording product-isomer identity provenance for rebuilt TENDL-format libraries; the
pre-built data-v1.0.0 artifacts predate that machinery. Three hash-pinned alternates (TENDL-2023,
FENDL-3.2c, EAF-2010) were qualified against the frozen P25b machinery and none survived:
TENDL-2023 carries the same defect class, EAF-2010 cannot express isomeric identity under the current
builder, and none passes comparable-case nonregression under the release builder.

The full disclosure, affected-target list and remediation paths are in
docs/DATA_LIMITATIONS.md.

Verification

• Full release preflight passed on the exact tagged commit: export-snapshot integrity, tracked-file
  manifest, cargo fmt, cargo check --workspace --all-targets --all-features --locked,
  cargo clippy -- -D warnings, the complete Rust test suite, a stable-ABI wheel build, disposable-venv
  install and smoke test, and the desktop build with its model smoke suite.
• Release CI verified the tag on all five Python distribution targets (Linux x86_64/aarch64, macOS
  x86_64/aarch64, Windows x86_64) plus the tagged-Rust-release gate before publication.
• The P18/P18b state-identity phases, P25 defect census and P25b alternate-evaluation qualification all
  closed under frozen protocols with independent, mutation-tested checkers; every verdict, including
  the failures, is preserved in the repository.