A programmable stellarator-tokamak hybrid for million-scale magnetic-configuration discovery

This paper proposes a programmable hybrid fusion device that combines a tokamak-like coil set with 288 dipole-field coils to generate over 1.66 million optimized magnetic configurations, enabling rapid discovery of diverse stellarator and tokamak geometries without requiring hardware redesign.

The Gist

Imagine trying to build the perfect oven to cook the sun's energy (nuclear fusion). For decades, scientists have been stuck choosing between two very different oven designs: the Tokamak and the Stellarator.

• The Tokamak is like a simple, round donut oven. It's easy to build and works great, but it needs a constant, powerful electric current running through the fuel to keep it stable. If that current hiccups, the whole oven shuts down (a "disruption").
• The Stellarator is like a twisted, pretzel-shaped oven. It doesn't need that internal electric current, so it can run forever without shutting down. But to make the twisted shape work, you need to build incredibly complex, custom-shaped metal coils that are a nightmare to manufacture and assemble.

The Big Problem:
Until now, if you wanted to test a new, better shape for your fusion oven, you had to tear down your old machine and build a brand new one with different, custom-shaped coils. It's like having to rebuild your entire kitchen every time you wanted to try a new recipe.

The New Solution: The "Programmable" Hybrid
This paper introduces a new device that acts like a programmable kitchen. Instead of building new coils for every new idea, the researchers propose a machine with a fixed set of simple, flat metal rings (coils).

Here is how it works, using simple analogies:

1. The "Pixel" Coils

Imagine the machine has a grid of 288 small, flat coils (like pixels on a giant screen) wrapped around the donut.

• In a traditional machine, these coils are hard-wired to do one specific job.
• In this new machine, you can program the electricity flowing through each of these 288 coils individually.
• By turning the "volume" (current) up or down on different coils, you can change the shape of the magnetic field inside the machine instantly.

2. The "Magic Menu" of Shapes

Because you can program the electricity, this single machine can instantly switch between being a Tokamak or a Stellarator, or anything in between.

• The researchers used a computer to "program" these coils into 1.66 million different shapes.
• They found shapes that are perfectly round (Tokamak style), shapes that are twisted like a pretzel (Stellarator style), and millions of variations in between.
• It's as if you could take one physical oven and, by pressing a button, instantly turn it into a pizza oven, a bagel maker, or a deep-fryer without changing a single part of the hardware.

3. Why This Matters

• Speed: Instead of spending years designing and building a new machine to test a new idea, scientists can just "download" a new configuration and test it immediately.
• Discovery: They found that this machine can create "hidden symmetries" (special magnetic patterns) that trap heat and particles very efficiently, which is the holy grail of fusion energy.
• Flexibility: They showed that the same machine can handle the "twisted" needs of a stellarator and the "round" needs of a tokamak, and even test specific tweaks to fix problems in existing tokamak designs.

The Bottom Line:
This paper proposes a "universal translator" for fusion energy. It separates the idea (the magnetic shape) from the hardware (the metal coils). By using a fixed set of simple coils that can be reprogrammed with electricity, they can explore over a million different fusion designs on a single machine, speeding up the path to clean, limitless energy.

Technical Summary

Technical Summary: A Programmable Stellarator-Tokamak Hybrid for Million-Scale Magnetic-Configuration Discovery

Problem Statement
Magnetic confinement fusion relies primarily on two concepts: tokamaks and stellarators. Tokamaks utilize simple axisymmetric coils and plasma current to achieve excellent confinement but suffer from current-driven disruptions and the need for sustained current drive. Conversely, stellarators enable steady-state, disruption-free operation using external three-dimensional (3D) magnetic fields; however, generic non-axisymmetric fields often suffer from large neoclassical transport. While "hidden symmetries" such as quasi-axisymmetry (QA), quasi-helical symmetry (QH), and quasi-isodynamicity (QI) can mitigate these transport losses, optimized stellarators typically require complex, non-planar 3D coils that are difficult to manufacture and assemble. Previous hybrid approaches have attempted to combine these advantages but have generally been limited to specific configurations or required hardware redesigns to explore new magnetic topologies.

Methodology
The authors propose a "programmable stellarator–tokamak hybrid" device designed to decouple magnetic-configuration discovery from hardware redesign. The device architecture consists of a standard tokamak-like coil set (toroidal field coils, poloidal field coils, and a central solenoid) augmented by a fixed set of 288 dipole-field (DF) coils. These DF coils are mounted on a toroidal support structure with an aspect ratio of 2, arranged in a $24 \times 12$ toroidal-by-poloidal grid. Due to symmetry, only six unique coil geometries are required. The DF coils are fabricated from high-temperature superconducting (HTS) tapes to carry large currents.

The core methodology involves programming the currents in these DF coils to generate a vast range of magnetic configurations without altering the physical hardware. The optimization workflow utilizes a two-stage process:

1. Seed Generation: A quasi-single-stage optimization using the SIMSOPT framework iteratively adjusts plasma boundaries and coil currents to satisfy physics targets (QA, QH, or QI properties) and engineering constraints (aspect ratio, edge rotational transform, and maximum coil current).
2. Global Search: A differential evolution algorithm performs a large-scale global search over the TF and DF coil current space, initialized from the seed configurations.

The resulting configurations are validated using free-boundary equilibrium codes (VMEC, TokaMaker) and transport analysis tools (NEO for neoclassical transport, SIMPLE for collisionless $\alpha$-particle losses).

Key Contributions and Results

• Million-Scale Configuration Database: By varying only the coil currents, the authors generated a database of over 1.66 million optimized stellarator configurations. This database spans QA, QH, and QI configurations with varying field periodicities ($n_{fp} = 2, 3, 4$) and rotational transforms ranging from 0.1 to 1.3. All selected configurations meet criteria for low neoclassical transport ($\epsilon_{eff}^{3/2} < 10^{-2}$) and practical compactness ($A_p < 15$).
• Representative Configurations: The study presents specific optimized examples:
  • QA: Aspect ratio $A_p=6$, edge rotational transform $\iota_{edge}=0.19$, and $\epsilon_{eff}^{3/2} = 2.6 \times 10^{-4}$.
  • QI: $A_p=8$, $\iota_{edge}=0.47$, and $\epsilon_{eff}^{3/2} = 1.4 \times 10^{-4}$.
  • QH: $A_p=10$, $\iota_{edge}=0.92$, and $\epsilon_{eff}^{3/2} = 3.4 \times 10^{-3}$.
  • Particle Confinement: Collisionless $\alpha$-particle loss fractions for reactor-scaled configurations were evaluated. The four-period QI configuration exhibited the lowest loss fraction (below 1%), indicating effective energetic-particle confinement even with discrete-coil ripple.
• Advanced Tokamak Control: The device demonstrates unprecedented selectivity in 3D field control for tokamak modes. The authors show that resonant magnetic perturbations (RMPs) can be highly isolated to specific rational surfaces (e.g., $(m, n) = (2, 2)$) while minimizing non-resonant responses. Specifically, the design enables the creation of quasi-symmetric magnetic perturbations (QSMPs) that reduce neoclassical toroidal viscosity (NTV) by five orders of magnitude, a level of control beyond the capability of present tokamak coil sets.

Significance and Claims
The paper claims that this approach fundamentally separates magnetic-configuration discovery from hardware redesign. By utilizing a single, fixed hardware platform capable of accessing over a million optimized stellarator fields and advanced tokamak configurations, the device enables rapid experimental testing of new concepts without the need for costly and time-consuming coil reconstruction.

The authors modestly note that the demonstrated QH and QI configurations are not exact hidden-symmetry fields but rather combine these structures with piecewise-omnigenity (pwO) features while maintaining good confinement. This suggests that high-quality confinement may be achievable in a broader configuration space than that defined by exact symmetry alone. The proposed platform offers a practical path to experimentally validate a large family of optimized stellarator and tokamak-relevant 3D magnetic fields, potentially accelerating the development of future fusion reactors.