Plasma and Fusion Research

Volume 20, 1201059 (2025)

Rapid Communications


Predicting Spatio-Temporal Dynamics in Multi-Dimensional Turbulent Flows via Hankel Sparsity-Promoting Dynamic Mode Decomposition
Naoya MITSUHASHI, Makoto SASAKI, Takumi KODAHARA
College of Industrial Technology, Nihon University, Narashino 275-8575, Japan
(Received 15 September 2025 / Accepted 31 October 2025 / Published 24 December 2025)

Abstract

In this study, we apply Hankel Sparsity-Promoting Dynamic Mode Decomposition (Hankel-SP-DMD) to dynamic turbulent flows in two-dimensional space to predict their long-term spatial structures. The target turbulence data are obtained from numerical simulations based on the extended Hasegawa-Wakatani model. The proposed method successfully extracts dominant modes and predicts their temporal evolution. We further evaluate the impact of hyperparameter settings required for training on prediction performance, and provide guidance for appropriate parameter selection based on the correlation with turbulence and limit-cycle periods.


Keywords

DMD, sparse modeling, plasma turbulence, long-term prediction

DOI: 10.1585/pfr.20.1201059


References

  • [1] J.W. Connor, Plasma Phys. Control. Fusion 40, 191 (1998).
  • [2] P.H. Diamond et al., Plasma Phys. Control. Fusion 47, R35 (2005).
  • [3] M. AizawaCaranza et al., Plasma Fusion Res. 17, 1201048 (2022).
  • [4] M. Priyanka et al., E3S Web Conf. 477, 00039 (2024).
  • [5] P.J. Schmid, J. Fluid Mech. 656, 5 (2010).
  • [6] M. Sasaki et al., J. Plasma Fusion Res. 97, 79 (2021).
  • [7] A. Kusaba et al., Jpn. J. Appl. Phys. 61, SA1011 (2021).
  • [8] M. Sasaki et al., Plasma Phys. Control. Fusion 61, 112001 (2019).
  • [9] M. Sasaki et al., Phys. Plasmas 26, 042305 (2019).
  • [10] T. Kodahara et al., Plasma Phys. Control. Fusion 67, 065012 (2025).
  • [11] J.H. Tu et al., J. Comput. Dyn. 1, 391 (2014).
  • [12] M.R. Jovanović et al., Phys. Fluids 26, 024103 (2014).
  • [13] N. Kasuya et al., Phys. Plasmas 15, 052302 (2008).
  • [14] M. Sasaki et al., Phys. Plasmas 24, 112103 (2017).