arXiv subjects

Yuetao Meng

Publications and source records attributed to Yuetao Meng.

2 recordsLinked to original sources

arXiv2026

DKEKAN: A single-parameterized KAN surrogate for Drift Kinetic Equation Toward Fast Neoclassical Toroidal Viscosity Torque Modeling in Tokamaks

The neoclassical toroidal viscosity (NTV) torque is a critical driver of toroidal rotation in tokamaks, profoundly influencing plasma stability and performance. Consequently, incorporating NTV effects is essential for modern integrated modeling frameworks that aim to self-consistently unify multiple physical processes. However, the high computational cost of NTV modeling precludes its self-consistent integration within such frameworks. This bottleneck arises because NTV calculation requires solving its governing equation--the drift kinetic equation (DKE)--in high-dimensional phase space. To address this issue, this study develops DKEKAN, a single-parameterized Kolmogorov-Arnold Network (SKAN) surrogate for solving DKE, to realize fast NTV modeling in tokamaks. The research process consists of the following steps: Firstly, a large dataset mapping DKE equation parameters to solutions is generated based on first-principle simulations under plasma parameters of the Experimental Advanced Superconducting Tokamak (EAST); Secondly, a surrogate model for solving DKE is developed based on the SKAN framework, which also incorporates a modular expert network design; Finally, the DKEKAN surrogate model is integrated with the NTV modeling framework to realize fast NTV calculation. With its physics-grouped expert layer and SKAN backbone, DKEKAN outperforms the tested MLP, KAN, and neural-operator baselines in overall prediction accuracy, while reducing the standalone DKE-solving time from 35.85s to 3.74s, corresponding to a speedup of approximately 9.6x, and reducing the total coupled NTVTOK runtime from 38.24s to 5.58s, corresponding to an overall speedup of approximately 6.9x. This work effectively overcomes the computational bottleneck in NTV simulations, thus supporting further integrated modeling that incorporates NTV effects.

physics.plasm-ph↗

arXiv2026

A Data-Free, Physics-Informed Surrogate Solver for Drift Kinetic Equation: Enabling Fast Neoclassical Toroidal Viscosity Torque Modeling in Tokamaks

Toroidal rotation is crucial for maintaining stable and high performance plasmas in tokamak fusion reactors. Among its driving mechanisms, the neoclassical toroidal viscosity (NTV) torque--induced by three-dimensional magnetic perturbations--is particularly significant due to its strong impact and controllability, especially for reactor-scale devices like ITER where conventional momentum injection method becomes less effective. However, traditional first-principle NTV modeling is computationally expensive, as it requires solving the drift kinetic equation (DKE) in high-dimensional phase space, therefore precluding any real-time applications such as active control or nonlinear integrated modeling of tokamak plasma. Although surrogate solver shows promising ability for accelerating scientific computations, obtaining the data required to train such model is still very challenging. In this work, we present a novel, data-free approach for developing fast surrogate solver of DKE, by training neural network solely based on physical constraints. Such physical constraints are implemented in two ways: First, the loss function is defined based on physical governing equations; Second, the boundary condition is hard-coded into the predicting model. The proposed model is validated against the dataset generated by first-principle numerical solver, which is found to achieve accurate DKE solution with significantly reduced time consuming. In particular, physics-driven surrogate shows higher physical consistency than data-driven surrogate. In general, our study provides a new idea for developing surrogate solvers in data-scarce scenarios, and demonstrates the potential of purely physics-driven neural networks to accelerate demanding scientific computations.

physics.plasm-ph↗