Chaos in Tokamak Edge Plasmas: Classification via Decision Tree and Toy Model Construction

Experiments at DIII-D, Alcator C-Mod, and KSTAR have shown that edge fluctuations can exhibit signatures of low-dimensional chaos [1–3]. How these signatures relate to transport and confinement changes across regimes remains an open question, and classifying experimental time series as chaotic is challenging. We apply a chaotic decision tree analysis [4] using surrogate testing, de-noising, and the modified 0-1 test to classify time series from DIII-D, C-Mod, and AUG. The analysis identifies chaos in C-Mod data, consistent with [2]. Reflectometer signals in the edge of I-mode at AUG are also chaotic, while CECE signals are stochastic, likely due to dominant thermal noise. Data from DIII-D negative triangularity plasmas [5] is also analyzed and chaotic signatures are found in the edge region as well. To connect observations to theory, we explore construction of a new toy model starting with curvature extended Hasegawa-Wakatani equations [6]. Connection to past successful dynamical systems models for edge plasmas, e.g. [7], is discussed.

[1] Maggs et al Plasma Phys. Control. Fusion 57, 045004 (2015)

[2] Zhu et al. Phys. Plasmas 24, 4 (2017)

[3] Choi et al. Phys. Plasmas 29, 122504 (2022)

[4] Toker et al. Commun Biol 3, 11 (2020)

[5] Stewart et al. Plasma Phys. Control. Fusion 67, 025032 (2025)

[6] Garland et al. Phys. Plasmas 27, 052304 (2020)

[7] Kim and Diamond Phys. Rev. Lett. 90, 185006 (2003)