{ "schema": "scpn-control.competitive-evidence.v1", "as_of": "2026-08-28", "project_version": "0.23.0", "comparison_policy": { "absence_state": "not assessed", "numeric_admission": "Identical problem, inputs, precision, tolerances, convergence criteria, warm-up and compilation treatment, samples, isolated hardware, and failure accounting are required.", "scope": "Evidence scope and documented capabilities; not a product ranking or market-wide survey." }, "systems": [ { "id": "scpn-control", "name": "SCPN Control", "artifact_kind": "repository-snapshot", "assessed_version": "v0.23.0 snapshot f1f6e36d", "artifact_date": "2026-08-28", "commit_sha": "f1f6e36d05f259f81c7b6c33d0d1d0089c921635", "primary_role": "Neuro-symbolic plasma-control research toolkit and evidence-admission layer.", "documented_strengths": [ "Runtime-selectable classical, robust, predictive, and neuro-symbolic control surfaces.", "Executable safety and evidence-admission contracts with Python, Rust, TypeScript, Lean, and C++ surfaces.", "Offline simulation, replay, and target-hardware evidence interfaces with facility claims blocked by default." ], "evidence_boundary": "Not commissioned on a plasma-control system; external-code parity, target-hardware timing, and facility validation remain separately admitted claims.", "source_urls": [ "https://github.com/anulum/scpn-control/blob/f1f6e36d05f259f81c7b6c33d0d1d0089c921635/docs/production_readiness.md", "https://github.com/anulum/scpn-control/blob/f1f6e36d05f259f81c7b6c33d0d1d0089c921635/docs/_generated/capability_manifest.json" ] }, { "id": "torax", "name": "TORAX", "artifact_kind": "release", "assessed_version": "v1.4.3", "artifact_date": "2026-07-03", "commit_sha": "4aea2377385ba4dfe37b0ef4396374162af1314b", "primary_role": "Differentiable JAX tokamak core-transport simulator.", "documented_strengths": [ "Coupled ion/electron heat, particle, and current-diffusion PDEs.", "Differentiable nonlinear solves and optimisation-oriented JAX execution.", "QLKNN and direct TGLF coupling plus IMAS core-source integration." ], "evidence_boundary": "SCPN Control has reduced transport and TORAX-shaped integration surfaces, but no admitted identical-input TORAX comparison.", "source_urls": [ "https://github.com/google-deepmind/torax/releases/tag/v1.4.3", "https://github.com/google-deepmind/torax/blob/v1.4.3/README.md" ] }, { "id": "fuse", "name": "FUSE", "artifact_kind": "release", "assessed_version": "v1.2.0", "artifact_date": "2026-08-10", "commit_sha": "9ef2f99af73497706a097d99a2aaac2f08405370", "primary_role": "Integrated fusion-device simulation and design framework.", "documented_strengths": [ "Coupled plasma physics, engineering, control, balance-of-plant, and costing actors.", "Multi-objective constrained design optimisation.", "ITER IMAS ontology with OMAS and OMFIT interoperability." ], "evidence_boundary": "SCPN Control provides control and admission components, not an equivalent whole-device design and costing environment.", "source_urls": [ "https://github.com/ProjectTorreyPines/FUSE.jl/releases/tag/v1.2.0", "https://github.com/ProjectTorreyPines/FUSE.jl/blob/v1.2.0/README.md" ] }, { "id": "freegs", "name": "FreeGS", "artifact_kind": "release", "assessed_version": "v0.8.2", "artifact_date": "2024-09-25", "commit_sha": "8b838c1df162ca770a6937ac79d8c73e8f10a53b", "primary_role": "Static free-boundary Grad-Shafranov equilibrium solver.", "documented_strengths": [ "Free-boundary equilibrium with plasma profiles, coils, machine circuits, and inverse constraints.", "G-EQDSK input/output and established transparent Python reference implementation." ], "evidence_boundary": "FreeGS is an external code-to-code reference; agreement with it is not experimental truth.", "source_urls": [ "https://github.com/freegs-plasma/freegs/releases/tag/v0.8.2", "https://github.com/freegs-plasma/freegs/blob/v0.8.2/README.md" ] }, { "id": "freegsnke", "name": "FreeGSNKE", "artifact_kind": "release", "assessed_version": "v3.0.1", "artifact_date": "2026-04-12", "commit_sha": "f776e908c8c333411f9824cbcfed674fafff8dfd", "primary_role": "Static and evolutive free-boundary equilibrium and circuit simulator.", "documented_strengths": [ "Fourth-order Newton-Krylov static and implicit evolutive solves.", "Self-consistent active-coil, passive-structure, plasma-current, and magnetic-diagnostic modeling.", "Peer-reviewed MAST-U/EFIT++ validation and pulse/control-design literature." ], "evidence_boundary": "SCPN Control does not currently have equivalent admitted evolutive free-boundary or MAST-U/EFIT++ validation evidence.", "source_urls": [ "https://github.com/FusionComputingLab/freegsnke/releases/tag/v3.0.1", "https://github.com/FusionComputingLab/freegsnke/blob/v3.0.1/README.md", "https://doi.org/10.1088/1402-4896/ada192" ] }, { "id": "dream", "name": "DREAM", "artifact_kind": "release", "assessed_version": "v26.5", "artifact_date": "2026-05-26", "commit_sha": "ecdd5e146537c77602c9d7cc76b36100200e4b9a", "primary_role": "Fluid-kinetic tokamak disruption and runaway-electron simulator.", "documented_strengths": [ "Nonlinear fluid equations coupled to kinetic electron equations.", "Specialised runaway generation, dynamics, losses, and shattered-pellet-injection models.", "Dedicated peer-reviewed physics and software lineage." ], "evidence_boundary": "SCPN Control contains control-grade disruption and mitigation models, not an equivalent DREAM fluid-kinetic solver.", "source_urls": [ "https://github.com/chalmersplasmatheory/DREAM/releases/tag/v26.5", "https://github.com/chalmersplasmatheory/DREAM/blob/v26.5/README.md", "https://doi.org/10.1016/j.cpc.2021.108098" ] }, { "id": "process", "name": "PROCESS", "artifact_kind": "release", "assessed_version": "v3.4.2", "artifact_date": "2026-08-14", "commit_sha": "c0ae5b28649f2b20fb7efc7904628b6defe4151c", "primary_role": "UKAEA reactor systems design code.", "documented_strengths": [ "Integrated reactor physics, engineering, costing, and design optimisation.", "Versioned releases, Zenodo lineage, tests, and variable tracking." ], "evidence_boundary": "SCPN Control is not a reactor systems-design or costing code; PROCESS v3 also declares its API unstable pending v4.", "source_urls": [ "https://github.com/ukaea/PROCESS/releases/tag/v3.4.2", "https://github.com/ukaea/PROCESS/blob/v3.4.2/README.md" ] }, { "id": "omas", "name": "OMAS", "artifact_kind": "release", "assessed_version": "v0.97.3", "artifact_date": "2026-05-28", "commit_sha": "e95c785e8c4c461adb66cc130e16b8950139b103", "primary_role": "Python interoperability layer for the ITER IMAS data model.", "documented_strengths": [ "IMAS-compatible data representation across multiple file and database formats.", "Integration substrate for OMFIT and third-party fusion codes." ], "evidence_boundary": "SCPN Control exposes IMAS-facing contracts, but its current public evidence does not establish the breadth of the OMAS/OMFIT data and workflow ecosystem.", "source_urls": [ "https://github.com/gafusion/omas/releases/tag/v0.97.3", "https://github.com/gafusion/omas/blob/v0.97.3/README.md", "https://www.omfit.io/modules.html" ] }, { "id": "tcv-drl-control", "name": "TCV deep-RL magnetic control", "artifact_kind": "paper", "assessed_version": "doi:10.1038/s41586-021-04301-9", "artifact_date": "2022-02-16", "commit_sha": null, "primary_role": "Learned magnetic controller demonstrated on the TCV tokamak.", "documented_strengths": [ "10 kHz closed-loop control using real TCV magnetic and coil-current measurements.", "Zero-shot simulation-to-hardware transfer and actuation of 19 control coils.", "Peer-reviewed demonstrations across multiple plasma shapes, including two sustained plasma droplets." ], "evidence_boundary": "SCPN Control has no commissioned tokamak actuation result; simulation or offline replay is not equivalent evidence.", "source_urls": [ "https://doi.org/10.1038/s41586-021-04301-9" ] }, { "id": "efit-ai-gpu", "name": "EFIT-AI GPU reconstruction", "artifact_kind": "paper", "assessed_version": "doi:10.1145/3624062.3624607", "artifact_date": "2023-11-12", "commit_sha": null, "primary_role": "Performance-portable GPU acceleration of the EFIT inverse equilibrium solver.", "documented_strengths": [ "GPU-offloaded equilibrium-reconstruction kernels evaluated on NVIDIA, AMD, and Intel accelerators.", "Named-grid and named-hardware performance-portability study within the EFIT reconstruction lineage." ], "evidence_boundary": "SCPN Control has no admitted identical-input, full-output, named-GPU EFIT-AI comparison.", "source_urls": [ "https://doi.org/10.1145/3624062.3624607" ] } ], "quantitative_comparisons": [] }