Researchers have developed a computational method that solves collisionless gyrokinetic eigenvalue problems in sub-second timescales, representing a significant acceleration for the analysis of microscopic drift-wave instabilities. The technique, detailed in a new preprint, is called orbit-invariant decomposition and has been implemented in the MGK code with both CPU and GPU support. For specific test cases involving ion temperature gradient (ITG) and trapped electron modes (TEM), the solver achieved solution times in the 0.01 to 0.1 second range. This performance is more than three orders of magnitude faster than the widely used CGYRO code when benchmarked on identical hardware, according to the paper. Source: arXiv
The method's efficiency stems from its novel approach to discretizing velocity space. Instead of a standard grid, it discretizes along orbit invariants—specifically, particle energy and magnetic moment. This decomposition separates the full eigenvalue matrix into smaller, independent orbit blocks that are coupled only through the field equation. The result is a substantial reduction in the matrix dimension and the associated computational demand required to find a solution, without compromising the underlying physics of the simulation. The current implementation supports collisionless electrostatic linear simulations using both s-alpha and Miller equilibrium models. Source: arXiv
The method's efficiency stems from its novel approach to discretizing velocity space.
The primary application for this accelerated solver is the modeling of anomalous transport, a critical factor limiting confinement in magnetic fusion devices. Understanding and predicting transport driven by microinstabilities is essential for designing high-performance plasmas. The speed of the new MGK solver enables large-scale parameter scans that were previously computationally prohibitive. Such scans allow for a more thorough exploration of how plasma performance is affected by variables like temperature gradients, magnetic shear, and collisionality, leading to more robust and optimized operational scenarios for future tokamaks and stellarators. Source: arXiv
The authors validated the method by comparing its outputs—eigenfrequencies and mode structures—against established codes, finding strong agreement. This confirms that the computational speedup does not come at the cost of physical accuracy for the tested regimes. The ability to rapidly generate stability analyses is a key enabler for integrated modeling workflows, where physics from different domains are coupled to predict overall device performance. This tool could accelerate the design cycle for next-generation fusion power plants and improve the interpretation of experimental results from existing machines by providing faster, more detailed theoretical comparisons. Source: arXiv