Skip to content

SCPN Fusion Core v3.9.10

Choose a tag to compare

View all tags
@anulum anulum released this 05 Jun 15:10
· 153 commits to main since this release
v3.9.10
64b5245

SCPN Fusion Core v3.9.10

This release is a documentation, release-readiness, and FUS-C evidence traceability update.

Highlights

  • Bumps package metadata, README badge, release readiness, Rust/PyO3 package metadata, and generated capability snapshot references to 3.9.10.
  • Surfaces the latest FUS-C.6 pulsed-compression trajectory diagnostics, FUS-C.7 Faraday flux and trajectory-quality gates, and FUS-C.4 MRTI acceleration coupling in the public documentation path.
  • Keeps local Python/Rust/benchmark evidence separate from blocked external full-fidelity parity claims.
  • Preserves the fail-closed status for GENE/CGYRO/GS2 nonlinear GK, full electromagnetic Vlasov-Maxwell parity, DREAM kinetic runaway-electron parity, Aurora/STRAHL impurity transport parity, strict FreeGS/free-boundary parity, and distributed GPU/cluster scaling evidence.

Reproducibility entry points

python validation/full_fidelity_end_to_end_campaign.py
python validation/benchmark_full_fidelity_acceptance.py
python validation/benchmark_gk_electromagnetic_fidelity.py
PYTHONPATH=src python benchmarks/bench_pulsed_compression.py
PYTHONPATH=src python benchmarks/bench_faraday_recovery.py

Rust FUS-C checks:

CARGO_INCREMENTAL=0 cargo test --manifest-path scpn-fusion-rs/Cargo.toml -p fusion-physics compression -- --nocapture
CARGO_INCREMENTAL=0 cargo test --manifest-path scpn-fusion-rs/Cargo.toml -p fusion-physics faraday_recovery -- --nocapture
CARGO_INCREMENTAL=0 cargo bench --manifest-path scpn-fusion-rs/Cargo.toml -p fusion-physics --bench pulsed_compression_bench -- --sample-size 10
CARGO_INCREMENTAL=0 cargo bench --manifest-path scpn-fusion-rs/Cargo.toml -p fusion-physics --bench faraday_recovery_bench -- --sample-size 10

Evidence boundary

This release does not claim completed end-to-end full-fidelity parity. It improves how readers find current local contracts, benchmark commands, release metadata, and the external artifacts still required before blocked parity rows can become accepted evidence.

Assets 3
Loading