STELLARATOR DESIGN / QI v1.0
QI Design Benchmark
Public Fourier boundaries. Verified feasibility. Pareto hypervolume.
Leaderboard
| Rank | Participant / Method | Feasible | Status | Received (UTC) |
|---|
No reviewed submissions in this NFP track yet.
Pareto comparison
Lower aspect ratio, higher L_delta_b. Only feasible points contribute to HV.
Design inspection
Submit a design set
Public submission
HF account, method name and all Fourier coefficients will be public. Scores appear after maintainer evaluation and review.
No file selected.
Upload the validated file under submissions/ and open a pull request. Your HF PR account identifies the entry; this page does not upload automatically.
Evaluation rules
Submission
One submission contains one or more stellarator-symmetric boundaries with the same NFP (2, 3 or 4). Each r_cos / z_sin matrix has shape 5 × 9, with m = 0…4 and n = −4…4. Coefficients are public. No benchmark-imposed design-count or file-size cap; HF platform and browser/local resource limits still apply.
Leaderboard
Each NFP is ranked separately. A participant's highest-HV reviewed submission is shown; every reviewed attempt remains in history. Different submissions are never combined.
Scoring
Hypervolume of the feasible non-dominated designs, minimizing aspect ratio and maximizing L_delta_b. Fixed reference: (20, −1), without objective normalization.
Evaluation
qi / qi_frontier v1.0.0, constellaration 0.2.6, fixed high fidelity. Failed designs contribute no area. Partial evaluations are labelled; infrastructure faults are retried before review. An all-failed submission is unranked.
Feasibility
Numerical tolerances follow the problem definitions, not the optimizer's internal feasibility gate. Geometry previews are not physics evaluations.