The Challenge

Inferring magnetic geometry without magnetic diagnostics

A deceptively simple but scientifically rigorous inverse problem — and a probe of how far machine learning can be pushed toward truly machine-agnostic plasma state estimation.

Graphical abstract: mapping coil currents and Thomson-scattering profiles to a 2D poloidal flux map without magnetic diagnostics.
The whole challenge in one diagram: non-magnetic inputs on the left, the learned surrogate in the middle, and the full equilibrium — flux map, boundary, and scalars — on the right.
Why this matters

The diagnostics that built tokamak control are about to fail

Magnetic confinement fusion is transitioning from physics-exploration tokamaks to a new generation of reactor-class devices — SPARC, ARC, CFETR — designed to demonstrate net energy gain. These machines will deliver neutron fluxes more than an order of magnitude higher than any existing facility.

In that environment the very diagnostics that have underpinned tokamak control for half a century — Mirnov coils, pickup loops, Rogowski coils mounted near the vessel walls — face severe degradation. Radiation-induced EMF and conductivity produce drift voltages that exceed the signal of interest in long pulses, and structural transmutation slashes sensor lifetime to a fraction of a commercial plant’s operating hours.

The equilibrium-reconstruction problem doesn’t disappear — it becomes more important. Strip the magnetic signals from the classic EFIT code and it loses essentially all of its constraints: the capability we will most demand of next-generation devices is the one our existing toolkit cannot deliver.

Example poloidal flux map ψ(R,Z) with nested magnetic flux surfaces.
A poloidal flux map ψ(R,Z): nested contours are magnetic flux surfaces; the outermost closed surface is the plasma boundary.

“Can machine learning, trained on present-day machines, learn an equilibrium representation that survives the loss of magnetic diagnostics?”

The research question
The task

What you are given, and what you predict

For a held-out shot, you receive non-magnetic diagnostic inputs at every EFIT timestamp and predict the equilibrium.

Inputs (non-magnetic)

  • Coil currents — 18 F-coils, ohmic solenoid (ECOILA) and toroidal-field coil (bcoil) on DIII-D; 10 P-coils, solenoid, toroidal-field (TF) and error-field (EFPS) coils on MAST. 21 signals on DIII-D, 14 on MAST.
  • Thomson scattering — electron temperature Te and density ne profiles (core & edge views).
  • Plasma current Ip.

Outputs (the equilibrium)

  • Full flux map ψ̂(R,Z) on a 65×65 grid — the primary target; you submit it plus two scalars (below).
  • Two predicted scalars — edge safety factor q95 and normalized beta βN, the only quantities a flux map cannot contain.
  • Everything else is derived from your flux map — the LCFS boundary, the magnetic axis (Raxis, Zaxis), elongation and triangularity, volume, and internal inductance li are all computed from your ψ̂ by the scorer and scored for consistency with the same computation on the true ψ.
Two challenges, two awards

What you can win, and how

Challenge 1 Award #1 · $500

Intra-machine (DIII-D)

Reconstruct the plasma’s shape on the same machine you trained on.

Scored on
S_model S_model The composite score that decides Challenge 1: S = 0.55·R²ψ + 0.15·R²_q95,βN + 0.10·(1 − D_LCFS) + 0.20·Consistency. It blends flux-map accuracy, the two predicted scalars, boundary alignment, and the consistency of the eight ψ-derived scalars into one number in [0, 1]. Higher is better.
Wins
Best DIII-D intra-machine reconstruction — $500
Leaderboard
sort by Ch1: DIII-D S
Challenge 2 Award #2 · $500

Cross-machine (DIII-D → MAST)

Take that model to a completely different machine it has never seen — with zero training data.

Scored on
G_ratio G_ratio The score that decides Challenge 2: G = S_model(MAST) / S_model(DIII-D) — the fraction of your DIII-D performance you keep when moving zero-shot to MAST. 1.0 would mean perfect transfer. It needs predictions for both machines, and only entries with S_model(DIII-D) ≥ 0.85 are eligible — below that gate (or with DIII-D missing) the scorer reports 0.
Wins
Best DIII-D → MAST cross-machine generalization — $500
Leaderboard
sort by Ch2: G_ratio
The honest stress test

Why Challenge 2 is the hard one

Comparison of DIII-D conventional D-shaped plasma and MAST spherical kidney-bean plasma cross-sections.
DIII-D’s conventional D-shape vs. MAST’s spherical, low-aspect-ratio geometry.

Challenge 2 asks you to train on DIII-D and transfer zero-shot Zero-shot Applying a model to a new machine with no training data from it at all — no fine-tuning, no examples. MAST has zero training shots on purpose; that is the whole of Challenge 2. to MAST — a machine with a different shape, a different coil set, and no training data at all. Nothing about MAST is available to learn from.

Our pilot baseline makes the difficulty explicit: a naïve coil-mapping transfer scores SSIM SSIM Structural Similarity Index — an image-quality metric that rewards getting the structure right rather than each pixel exactly. Used to describe baseline results, not to score submissions. 0.83 on DIII-D but only 0.10 on MAST. We frame that collapse as a failure on purpose — it is a far more honest measure of generalization than a within-machine split, and a direct test of whether a model captures the universal physics of the Grad–Shafranov equation Grad–Shafranov equation The equation governing a tokamak equilibrium. It is the same physics on every machine — which is exactly why a model that has truly learned it should transfer from DIII-D to MAST. or merely one device’s engineering minutiae.

No GPU? No problem

The compute-light badge

What it is

Applies to both challenges
A badge Compute-light badge A self-declared badge for models that train end-to-end in under two hours on a single GPU or CPU. It is a recognition that applies across both challenges — not a separate track, and not a separate leaderboard. for models that train end-to-end in under two hours on a single GPU or CPU. Tick a box on your submission to claim it. It is a recognition for accessible solutions — for educators and resource-constrained teams — not a separate competition, and not a quarantine.

What it is not

Not a separate leaderboard
There is no compute limit on anyone. Transformer-scale models, diffusion approaches, and neural operators are all welcome and compete on the same board. The badge simply marks which of those entries got there cheaply — and our reference ridge regression and small CNN both qualify.
Why it’s useful

Where a learned equilibrium pays off

Diagnostic-loss redundancy

A backup ML reconstruction provides a safe shutdown trajectory when a present-day machine suffers a transient magnetic-sensor failure.

Reactor primary inference

In SPARC and successors, kinetic diagnostics may outlast magnetic ones — making coil-and-profile inference the dominant equilibrium pipeline.

Real-time control

EFIT runs at ~10–100 ms latencies. A learned surrogate could plausibly run at sub-millisecond inference, fast enough for the modes EFIT can’t catch.

Physics interpretation

Kinetic-equilibrium reconstruction already fuses Thomson and magnetic data; this challenge explores the limit where the magnetic terms are removed entirely.