宇宙科学広報・普及主幹付

長谷川 洋

ハセガワ ヒロシ  (Hiroshi Hasegawa)

基本情報

所属
国立研究開発法人宇宙航空研究開発機構 宇宙科学研究所 助教

J-GLOBAL ID
200901040603353226
Researcher ID
A-1192-2007
researchmap会員ID
6000000662

外部リンク

論文

 172
  • Eunjin Choi, Hiroshi Hasegawa, Kyoung-Joo Hwang, Kyunghwan Dokgo, James L. Burch
    Physics of Plasmas 33(11) 2026年11月1日  
    This study demonstrates the effectiveness of physics-informed neural networks (PINNs) for reconstructing Grad–Shafranov (GS) equilibrium structures from synthetic, noise-free trajectory-like data representative of satellite measurements under both GS and magnetohydrostatic constraints. The proposed framework focuses on inferring continuous two-dimensional equilibrium structures from limited and sparsely sampled measurements along trajectories analogous to satellite sampling using static equilibrium solutions without explicitly provided boundary conditions. The sinusoidal representation network-enhanced architecture played a key role in capturing fine-scale structures. By incorporating uncertainty weighting, the model ensured balanced optimization, leading to higher accuracy and more effective enforcement of physical constraints. Additionally, the study highlights the importance of diverse training data, including multi-component measurements and data from multiple trajectories. The integration of such data significantly improved model performance by reducing errors and uncertainty while stabilizing training dynamics. These findings suggest that, even in the absence of explicit boundary conditions, which are commonly incorporated in PINN research, multi-component and multi-trajectory data can provide additional directional information that guides the solution process, reduces reconstruction ambiguity, and improves the reliability of the reconstructed solution.
  • I-Hsiang Chiu, Jih-Hong Shue, Hiroshi Hasegawa, Yasuhito Narita, Jun Zhong, Zhaojin Rong, Masafumi Hirahara
    Earth, Planets and Space 78(1) 2026年10月9日  査読有り
  • Masaki N. Nishino, Keisuke Hosokawa, Lei Cai, Hiroshi Hasegawa, Yukinaga Miyashita, Yongliang Zhang, Larry J. Paxton
    Polar Science 101443-101443 2026年10月  査読有り
  • Hiroshi Hasegawa, Eunjin Choi, Kyoung‐Joo Hwang, Kyunghwan Dokgo
    Journal of Geophysical Research: Machine Learning and Computation 3(2) 2026年4月3日  査読有り筆頭著者責任著者
    Abstract We present a novel data analysis technique based on physics‐informed neural networks (PINNs) to reconstruct two‐dimensional (2D) magnetohydrodynamic (MHD) and Hall MHD equilibria in a space plasma from in situ spacecraft measurements. Our method incorporates the steady‐state MHD or Hall MHD equations—a set of partial differential equations (PDEs) as physical constraints—into a deep learning framework. In contrast to traditional reconstruction techniques relying on explicit spatial integration of the PDEs from spacecraft trajectories, the PINN approach allows us to obtain a physically consistent equilibrium by minimizing a composite loss function that includes both prediction errors and PDE residuals. We validate the method through benchmark tests using exact solutions of axially symmetric MHD and Hall MHD equilibria, and further demonstrate its utility through application to an extensively investigated magnetotail reconnection event observed by the Magnetospheric Multiscale mission. The resulting reconstruction reproduces key structural features, including the X‐type current sheet geometry and quadrupolar Hall magnetic field, consistent with prior results and numerical simulations. We also present empirical estimates of the reconnection rate and spatially varying resistivity based on the reconstructed fields. Our results demonstrate that the PINN‐based method provides a flexible and powerful new tool for revealing plasma structures around observing spacecraft, particularly in regions where model assumptions may be only approximately valid.
  • K.-J. Hwang, K. Dokgo, E. Choi, J. L. Burch, N. Bessho, H. Hasegawa, R. Nakamura, C. P. Escoubet, H. S. Fu, W. D. Fu, C. J. Pollock
    The Astrophysical Journal 998(1) 17-17 2026年2月2日  査読有り
    Abstract Magnetic reconnection is a fundamental physical process operating throughout the Universe, yet its onset in planetary magnetotails remains elusive. Current-sheet flapping is a common dynamic phenomenon in planetary magnetotails, with its roles in driving reconnection largely underexplored. This paper reports the first observation of electron-dominant and ion-coupled reconnection accompanied by current sheet flapping. The Magnetospheric Multiscale mission (MMS) spacecraft was located in the premidnight sector current sheet, which was undergoing kink-mode flapping that propagated duskward, away from the midnight sector, with a frequency of 0.06–0.1 Hz and a wavelength of around 0.5 R E . The flapping led MMS to cross the current sheet repeatedly, from north to south and vice versa. During the initial crossing of the current sheet, distinct electron outflows were observed, while ion responses were weak. This indicates either electron-only reconnection, suggesting an early stage of reconnection (temporal effect) or MMS passing through the exhaust region close to an X-line (spatial effect). Signatures of the electron diffusion region’s limited length, yielding an unusually high aspect ratio, suggest that the former scenario is more likely. Regular ion-coupled reconnection that involves ion outflows became more prominent until MMS eventually crossed the X-line from tailward to earthward. This event indicates a potential causal relationship between electron-only reconnection, current sheet flapping, and subsequent ion reconnection. We discuss feasible mechanisms for current sheet flapping and the feedback between flapping and reconnection, providing new insights into the onset of magnetotail reconnection and cross-scale processes of importance to other heliospheric and astrophysical systems.

MISC

 25

書籍等出版物

 2

講演・口頭発表等

 168

所属学協会

 1

共同研究・競争的資金等の研究課題

 8