Bayesian Data Analysis for Gaussian Process Tomography

T. Wang; D. Mazon; J. Svensson; A. Liu; C. Zhou; L. Xu; L. Hu; Y. Duan; G. Verdoolaege

doi:10.1007/s10894-018-0205-y

Bayesian Data Analysis for Gaussian Process Tomography

T. Wang
, D. Mazon
, J. Svensson
, A. Liu
, C. Zhou
, L. Xu
, L. Hu
, Y. Duan
, G. Verdoolaege

Laboratory for Plasma Physics

CNNC
Commissariat à l'Énergie Atomique (CEA)
University of Ghent
Max-Planck-Institut für Plasmaphysik
University of Science and Technology of China
Institute of Plasma Physics Chinese Academy of Sciences

Research output: Contribution to journal › Article › peer-review

13 Citations (Scopus)

Abstract

Bayesian inference is used in many scientific areas as a conceptually well-founded data analysis framework. In this paper, we give a brief introduction to Bayesian probability theory and its application to the tomography problem in fusion research by means of a Gaussian process prior. This Gaussian process tomography (GPT) method is used for reconstruction of the local soft X-ray (SXR) emissivity in WEST and EAST based on line-integrated data. By modeling the SXR emissivity field in a poloidal cross-section as a Gaussian process, Bayesian SXR tomography can be carried out in a robust and extremely fast way. Owing to the short execution time of the algorithm, GPT is an important candidate for providing real-time feedback information on impurity transport and for fast MHD control. In addition, the Bayesian formulism allows for uncertainty analysis of the inferred emissivity.

Original language	English
Pages (from-to)	445-457
Number of pages	13
Journal	Journal of Fusion Energy
Volume	38
Issue number	3-4
DOIs	https://doi.org/10.1007/s10894-018-0205-y
Publication status	Published - 1 Aug 2019

Keywords

Bayesian inference
Data analysis
Gaussian process
Nuclear fusion
Plasma physics
Soft X-ray
Tokamak
Tomography

Access to Document

10.1007/s10894-018-0205-y

Cite this

@article{203ff3806dab4403b33d577b8228c065,

title = "Bayesian Data Analysis for Gaussian Process Tomography",

abstract = "Bayesian inference is used in many scientific areas as a conceptually well-founded data analysis framework. In this paper, we give a brief introduction to Bayesian probability theory and its application to the tomography problem in fusion research by means of a Gaussian process prior. This Gaussian process tomography (GPT) method is used for reconstruction of the local soft X-ray (SXR) emissivity in WEST and EAST based on line-integrated data. By modeling the SXR emissivity field in a poloidal cross-section as a Gaussian process, Bayesian SXR tomography can be carried out in a robust and extremely fast way. Owing to the short execution time of the algorithm, GPT is an important candidate for providing real-time feedback information on impurity transport and for fast MHD control. In addition, the Bayesian formulism allows for uncertainty analysis of the inferred emissivity.",

keywords = "Bayesian inference, Data analysis, Gaussian process, Nuclear fusion, Plasma physics, Soft X-ray, Tokamak, Tomography",

author = "T. Wang and D. Mazon and J. Svensson and A. Liu and C. Zhou and L. Xu and L. Hu and Y. Duan and G. Verdoolaege",

note = "Publisher Copyright: {\textcopyright} 2018, Springer Science+Business Media, LLC, part of Springer Nature.",

year = "2019",

month = aug,

day = "1",

doi = "10.1007/s10894-018-0205-y",

language = "English",

volume = "38",

pages = "445--457",

journal = "Journal of Fusion Energy",

issn = "0164-0313",

number = "3-4",

}

TY - JOUR

T1 - Bayesian Data Analysis for Gaussian Process Tomography

AU - Wang, T.

AU - Mazon, D.

AU - Svensson, J.

AU - Liu, A.

AU - Zhou, C.

AU - Xu, L.

AU - Hu, L.

AU - Duan, Y.

AU - Verdoolaege, G.

PY - 2019/8/1

Y1 - 2019/8/1

N2 - Bayesian inference is used in many scientific areas as a conceptually well-founded data analysis framework. In this paper, we give a brief introduction to Bayesian probability theory and its application to the tomography problem in fusion research by means of a Gaussian process prior. This Gaussian process tomography (GPT) method is used for reconstruction of the local soft X-ray (SXR) emissivity in WEST and EAST based on line-integrated data. By modeling the SXR emissivity field in a poloidal cross-section as a Gaussian process, Bayesian SXR tomography can be carried out in a robust and extremely fast way. Owing to the short execution time of the algorithm, GPT is an important candidate for providing real-time feedback information on impurity transport and for fast MHD control. In addition, the Bayesian formulism allows for uncertainty analysis of the inferred emissivity.

AB - Bayesian inference is used in many scientific areas as a conceptually well-founded data analysis framework. In this paper, we give a brief introduction to Bayesian probability theory and its application to the tomography problem in fusion research by means of a Gaussian process prior. This Gaussian process tomography (GPT) method is used for reconstruction of the local soft X-ray (SXR) emissivity in WEST and EAST based on line-integrated data. By modeling the SXR emissivity field in a poloidal cross-section as a Gaussian process, Bayesian SXR tomography can be carried out in a robust and extremely fast way. Owing to the short execution time of the algorithm, GPT is an important candidate for providing real-time feedback information on impurity transport and for fast MHD control. In addition, the Bayesian formulism allows for uncertainty analysis of the inferred emissivity.

KW - Bayesian inference

KW - Data analysis

KW - Gaussian process

KW - Nuclear fusion

KW - Plasma physics

KW - Soft X-ray

KW - Tokamak

KW - Tomography

UR - https://www.scopus.com/pages/publications/85057003838

U2 - 10.1007/s10894-018-0205-y

DO - 10.1007/s10894-018-0205-y

M3 - Article

AN - SCOPUS:85057003838

SN - 0164-0313

VL - 38

SP - 445

EP - 457

JO - Journal of Fusion Energy

JF - Journal of Fusion Energy

IS - 3-4

ER -

Bayesian Data Analysis for Gaussian Process Tomography

Abstract

Keywords

Access to Document

Other files and links

Fingerprint

Cite this