A Pre-trained Variational Autoencoder for Gyrokinetic Plasma Turbulence Surrogate Modeling
Abstract
Machine learning surrogate models offer a promising path toward accelerating plasma turbulence simulations. We present PreVAE-Turb, a surrogate modeling framework that leverages pre-trained variational autoencoders (VAEs) from the Stable Diffusion image generation model for efficient spatial compression of turbulence fields. The pre-trained VAE is fine-tuned on turbulence data using a physics-informed loss function that includes a spectral loss operating in Fourier space to enforce spectral accuracy across scales. The VAE is combined with convolutional long short-term memory (ConvLSTM) networks to learn temporal dynamics in latent space, with a manifold consistency error metric that monitors encode--decode consistency during autoregressive rollouts. We validate the framework on two-dimensional Hasegawa-Wakatani drift-wave turbulence and extend it to gyrokinetic turbulence from the GENE code, where a four-channel adaptation simultaneously predicts electrostatic potential, density, and parallel/perpendicular temperature fluctuations without requiring architecture redesign. Once trained, inference generates thousands of time steps in seconds on a single GPU, providing substantial computational acceleration compared to direct numerical simulation. The pre-trained approach offers a transferable methodology broadly applicable to various turbulence simulation codes.
Related
- MGKDB: An IMAS-aligned multicode gyrokinetic simulation database for reproducible fusion turbulence modeling and data-driven analysis
- Mitigation of Initial Transients in Total-f Gyrokinetic Turbulence Simulations Using Neoclassically Relaxed Distribution Function
- Milestone in predicting core plasma turbulence: successful multi-channel validation of the gyrokinetic code GENE
- Global Gyrokinetic Simulations of Electromagnetic Turbulence in STEP
- Linear-Stability-Guided Two-Stage Surrogate Model for Nonlinear ITG-TEM Turbulence
BibTeX
@misc{yang2026pretrained,
title = {{A Pre-trained Variational Autoencoder for Gyrokinetic Plasma Turbulence Surrogate Modeling}},
author = {Minglei Yang and Marshall Nicholson and Diego Del-Castillo-Negrete and David Hatch and Guannan Zhang},
howpublished = {arXiv},
year = {2026},
month = sep,
doi = {10.48550/arxiv.2609.38438},
eprint = {2609.38438},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2609.38438},
}
RIS
TY - UNPB
TI - A Pre-trained Variational Autoencoder for Gyrokinetic Plasma Turbulence Surrogate Modeling
AU - Minglei Yang
AU - Marshall Nicholson
AU - Diego Del-Castillo-Negrete
AU - David Hatch
AU - Guannan Zhang
PY - 2026
DA - 2026/09/29
PB - arXiv
DO - 10.48550/arxiv.2609.38438
UR - https://arxiv.org/abs/2609.38438
AB - Machine learning surrogate models offer a promising path toward accelerating plasma turbulence simulations. We present PreVAE-Turb, a surrogate modeling framework that leverages pre-trained variational autoencoders (VAEs) from the Stable Diffusion image generation model for efficient spatial compression of turbulence fields. The pre-trained VAE is fine-tuned on turbulence data using a physics-informed loss function that includes a spectral loss operating in Fourier space to enforce spectral accuracy across scales. The VAE is combined with convolutional long short-term memory (ConvLSTM) networks to learn temporal dynamics in latent space, with a manifold consistency error metric that monitors encode--decode consistency during autoregressive rollouts. We validate the framework on two-dimensional Hasegawa-Wakatani drift-wave turbulence and extend it to gyrokinetic turbulence from the GENE code, where a four-channel adaptation simultaneously predicts electrostatic potential, density, and parallel/perpendicular temperature fluctuations without requiring architecture redesign. Once trained, inference generates thousands of time steps in seconds on a single GPU, providing substantial computational acceleration compared to direct numerical simulation. The pre-trained approach offers a transferable methodology broadly applicable to various turbulence simulation codes.
ER -