External validation artefacts needed for full-fidelity SCPN-CONTROL claims

[anulum]

Purpose

Track every SCPN-CONTROL validation item that requires external material before a stronger public or facility-facing claim is allowed.

The implementation and local regression tests exist, but the repository intentionally keeps these claims bounded until independently reproducible external artefacts are available.

Collaboration request

External collaborators can help by providing redistributable or reviewable artefacts, reference outputs, benchmark cases, executable-backed validation traces, measured discharge data, or hardware synthesis/timing evidence.

Evidence areas

• External gyrokinetic code validation: TGLF, GENE, GS2, CGYRO, QuaLiKiz binaries or public outputs.
• Nonlinear gyrokinetic saturation campaigns and published/cross-code CBC references.
• Miller geometry, species/collision, JAX parity, OOD, and online-learning reference campaign artefacts.
• EFIT/P-EFIT/GEQDSK equilibrium and reconstruction references.
• Real-shot/facility replay data including DIII-D MDSplus or equivalent artefacts.
• Transport, pedestal, ELM, MARFE, NTM, MHD, SOL, orbit, uncertainty, and integrated-scenario benchmarks.
• Neural equilibrium, neural transport, and neural turbulence reference datasets and weight provenance.
• Plasma-control replay or measured data for RZIP, RWM, density, free-boundary, digital twin, disruption, burn, volt-second, and current-drive claims.
• Hardware/runtime evidence for HDL synthesis, CODAC/EPICS timing, interlocks, backpressure, Rust/Python parity, and deployment-target phase/spiking runtimes.

Acceptance policy

Evidence must include provenance, licence or redistribution policy, immutable checksums where applicable, units, input decks or raw inputs, parsed outputs, validation command, and the specific claim it unblocks.

Child issue plan

• External gyrokinetic validation artefacts
• Equilibrium and reconstruction reference artefacts
• Transport, edge, MHD, and scenario benchmark artefacts
• Neural surrogate validation artefacts
• Plasma-control and facility replay artefacts
• Disruption, halo-current, and mitigation benchmark artefacts
• Hardware, HDL, CODAC/EPICS, and runtime deployment evidence

[anulum]

Child collaboration-request issues

• Collaboration request: external gyrokinetic validation artefacts #47 External gyrokinetic validation artefacts
• Collaboration request: equilibrium and reconstruction reference artefacts #48 Equilibrium and reconstruction reference artefacts
• Collaboration request: transport, edge, MHD, and scenario benchmark artefacts #49 Transport, edge, MHD, and scenario benchmark artefacts
• Collaboration request: neural surrogate validation artefacts #50 Neural surrogate validation artefacts
• Collaboration request: plasma-control and facility replay artefacts #51 Plasma-control and facility replay artefacts
• Collaboration request: disruption, halo-current, and mitigation benchmark artefacts #52 Disruption, halo-current, and mitigation benchmark artefacts
• Collaboration request: hardware, HDL, CODAC/EPICS, and runtime deployment evidence #53 Hardware, HDL, CODAC/EPICS, and runtime deployment evidence

Registry coverage source

This tracker was generated from validation/physics_traceability.json by selecting entries whose status or required actions depend on external/reference material, real facility data, published benchmarks, external executables, or hardware/runtime evidence.

Covered traceability statuses include:

• external_dependency_blocked
• validation_gap entries requiring external/reference evidence
• bounded_model entries that need measured, benchmark, facility, or hardware evidence before stronger public claims

The intent is not to label the implementation as broken. The implementation and local tests exist; these issues identify the artefacts required to promote bounded claims into stronger validated claims.