Stokes inversion based on convolutional neural networks

被引:43
|
作者
Asensio Ramos, A. [1 ,2 ]
Baso, C. J. Diaz [3 ]
机构
[1] Inst Astrofis Canarias, C Via Lactea S-N, Tenerife 38205, Spain
[2] Univ La Laguna, Dept Astrofis, E-38206 Tenerife, Spain
[3] Stockholm Univ, Alballova Univ Ctr, Inst Solar Phys, Dept Astron, S-10691 Stockholm, Sweden
基金
欧洲研究理事会; 瑞典研究理事会;
关键词
Sun: photosphere; Sun: magnetic fields; methods: data analysis; techniques: polarimetric; methods: numerical; MAGNETIC-FIELDS; SOLAR PHOTOSPHERE; GRANULATION; PROFILES; VECTOR; NM;
D O I
10.1051/0004-6361/201935628
中图分类号
P1 [天文学];
学科分类号
0704 ;
摘要
Context. Spectropolarimetric inversions are routinely used in the field of solar physics for the extraction of physical information from observations. The application to two-dimensional fields of view often requires the use of supercomputers with parallelized inversion codes. Even in this case, the computing time spent on the process is still very large. Aims. Our aim is to develop a new inversion code based on the application of convolutional neural networks that can quickly provide a three-dimensional cube of thermodynamical and magnetic properties from the interpreation of two-dimensional maps of Stokes profiles. Methods. We trained two different architectures of fully convolutional neural networks. To this end, we used the synthetic Stokes profiles obtained from two snapshots of three-dimensional magneto-hydrodynamic numerical simulations of different structures of the solar atmosphere. Results. We provide an extensive analysis of the new inversion technique, showing that it infers the thermodynamical and magnetic properties with a precision comparable to that of standard inversion techniques. However, it provides several key improvements: our method is around one million times faster, it returns a three-dimensional view of the physical properties of the region of interest in geometrical height, it provides quantities that cannot be obtained otherwise (pressure and Wilson depression) and the inferred properties are decontaminated from the blurring effect of instrumental point spread functions for free. The code, models, and data are all open source and available for free, to allow both evaluation and training.
引用
收藏
页数:18
相关论文
共 50 条
  • [1] Stokes inversion based on convolutional neural networks
    Asensio Ramos, A.
    Díaz Baso, C.J.
    [J]. Astronomy and Astrophysics, 2019, 626
  • [2] Inversion of stokes profiles with artificial neural networks
    Carroll, TA
    Balthasar, H
    Muglach, K
    Nickelt, I
    [J]. ADVANCED SOLAR POLARIMETRY: THEORY, OBSERVATION, AND INSTRUMENTA TION, 2001, 236 : 511 - 519
  • [3] Convolutional Neural Networks and Stokes Response Functions
    Centeno, Rebecca
    Flyer, Natasha
    Mukherjee, Lipi
    Egeland, Ricky
    Casini, Roberto
    Aleman, Tanausu del Pino
    Rempel, Matthias
    [J]. ASTROPHYSICAL JOURNAL, 2022, 925 (02):
  • [4] The inversion of Stokes profiles with artificial neural networks
    Carroll, TA
    Staude, J
    [J]. ASTRONOMY & ASTROPHYSICS, 2001, 378 (01) : 316 - 326
  • [5] Deep learning electromagnetic inversion with convolutional neural networks
    Puzyrev, Vladimir
    [J]. GEOPHYSICAL JOURNAL INTERNATIONAL, 2019, 218 (02) : 817 - 832
  • [6] The Face Inversion Effect in Deep Convolutional Neural Networks
    Tian, Fang
    Xie, Hailun
    Song, Yiying
    Hu, Siyuan
    Liu, Jia
    [J]. FRONTIERS IN COMPUTATIONAL NEUROSCIENCE, 2022, 16
  • [7] Magnetotelluric inversion based on convolutional neural network
    Liao X.
    Zhang Z.
    Yao Y.
    Lu R.
    Fan X.
    Cao Y.
    Feng T.
    Shi Z.
    [J]. Zhang, Zhihou (logicprimer@163.com), 1600, Central South University of Technology (51): : 2546 - 2557
  • [8] Micro motion parameter inversion of spatial cone target based on Convolutional Neural Networks
    Li Bo
    Ren Hongmei
    Xiao Zhihe
    [J]. SEVENTH ASIA PACIFIC CONFERENCE ON OPTICS MANUFACTURE (APCOM 2021), 2022, 12166
  • [9] Multiscale Dilated Convolutional Neural Networks for Transient Electromagnetic Inversion
    Cheng, Kai
    Yang, Xiaodong
    Wu, Xiaoping
    [J]. IEEE Transactions on Geoscience and Remote Sensing, 2024, 62
  • [10] Understading Image Restoration Convolutional Neural Networks with Network Inversion
    Protas, Eglen
    Bratti, Jose
    Gaya, Joel
    Drews, Paulo, Jr.
    Botelho, Silvia
    [J]. 2017 16TH IEEE INTERNATIONAL CONFERENCE ON MACHINE LEARNING AND APPLICATIONS (ICMLA), 2017, : 215 - 220