Spatially resolved EEDFs/IEDFs provide essential kinetic information about low-temperature plasmas (LTPs) and play a central role in determining transport, chemical reaction rates, and plasma surface interactions. While kinetic simulations directly resolve these distributions, experimental measurements remain challenging and are often invasive, spatially limited, or require assumptions regarding the distribution shape such as a Maxwellian.
arXiv:2608. 19377v1 Announce Type: cross Abstract: Bayesian inference via Markov Chain Monte Carlo (MCMC) provides effective parameter estimation, but its real-time application in complex physical systems is hindered by heavy computational bottlenecks and extreme sensitivity to statistical noise.
By Liyun Zhang, Naoya Mamada, Kentaro Sakai, Takeo Hoshi, Toru Aonishi
GyroSwin is a scalable 5‑dimensional neural surrogate that models nonlinear gyrokinetic plasma turbulence, a key challenge for nuclear fusion research. It extends Vision Transformers to 5D, incorporates cross‑attention and latent 3D↔5D interactions, and uses channelwise mode separation inspired by nonlinear physics. The model outperforms traditional reduced numerics in heat‑flux prediction, captures turbulent energy cascades, and cuts the computational cost of full gyrokinetic simulations by three orders of magnitude while remaining physically verifiable.
By Fabian Paischer, Gianluca Galletti, William Hornsby, Paul Setinek, Lorenzo Zanisi, Naomi Carey, Stanislas Pamela, Johannes Brandstetter
arXiv:2601. 10885v2 Announce Type: replace-cross Abstract: We propose a methodology to infer collision operators from phase space data of plasma dynamics.
By Diogo D. Carvalho, Pablo J. Bilbao, Warren B. Mori, Luis O. Silva, E. Paulo Alves
arXiv:2609.38438v1 Announce Type: cross
Abstract: Machine learning surrogate models offer a promising path toward accelerating plasma turbulence simulations. We present PreVAE-Turb, a surrogate model...
By Minglei Yang, Marshall Nicholson, Diego Del-Castillo-Negrete, David Hatch, Guannan Zhang
The paper presents an inverse Physics-Informed Neural Network (PINN) framework that infers the scrape‑off layer (SOL) perpendicular heat conductivity κ_π(n,T) from radial electron density and temperature measurements combined with a reduced one‑dimensional transport equation. Three neural networks are trained simultaneously: two reconstruct the temperature and density profiles, while the third models the conductivity as a function of local density and temperature. The method is validated on synthetic data, achieving errors below 10 % in the data‑constrained region, and is then applied to experimental data from the TJ‑II stellarator using a helium‑beam diagnostic to estimate the effective SOL conductivity.
By J. Gallego (Departamento de Tecnolog\'ia, CIEMAT, Spain), P. Protopapas (Harvard John A. Paulson School of Engineering and Applied Sciences, USA), A. Bustos (Departamento de Tecnolog\'ia, CIEMAT, Spain), A. Alonso (Laboratorio Nacional de Fusi\'on, CIEMAT, Spain), S. Barquero (Laboratorio Nacional de Fusi\'on, CIEMAT, Spain), A. Baciero (Laboratorio Nacional de Fusi\'on, CIEMAT, Spain), I. Rivera (Laboratorio Nacional de Fusi\'on, CIEMAT, Spain), J. A. Mor\'i\~nigo (Departamento de Tecnolog\'ia, CIEMAT, Spain), R. Mayo-Garc\'ia (Departamento de Tecnolog\'ia, CIEMAT, Spain)