Catégorie d'emplois: Physics

M2 Internship – 2026 – Physics – Modeling – PATP/MK/1

2025-2026 M2 Internship (4-6 months)

 Applying Deep-Learning (DL) to emission spectroscopy of fusion plasmas 

Laboratory: PIIM, UMR7345, PATP team (Atomic Physics and Transport in Plasmas)

Supervisor: Mohammed KOUBITI (mohammed.koubiti@univ-amu.fr)

Address: Campus St Jérôme, Service 232, Av. Escadrille Normandie Niemen, Marseille

Phone : +33 (0)4 13 94 64 47

Research type: Theory/Numerical Modeling/Comparison with Experimental data

Subject description: Artificial intelligence (AI) is increasingly used in physics including magnetic fusion plasmas. For instance, a Machine Learning (ML) algorithm [1] was used recently to predict the plasma parameters for PISCES-B and NAGDIS linear plasma devices [2-3]. Unlike the standard line ratio technique which relies on collisional-radiative modelling [4], in [2-3] no physical model is combined with the spectroscopic measurements. More precisely, using the intensities of few neutral helium lines the electron density and temperature were predicted by the ML algorithm and compared to their values measured by independent diagnostic techniques like Langmuir probes or Thomson scattering [2-3]. In this internship proposal, we suggest applying deep-learning techniques to line spectra of hydrogen isotopes in tokamak plasmas. We will apply in particular Dense Neural Networks (DNN) and Convolutional Neural Networks (CNN) to generated spectra of hydrogen isotopes for the aim of plasma diagnostics and predictions for future experiments. Our objective of applying DL techniques to the line emission of hydrogen isotopes in tokamaks is the prediction of the hydrogen isotopic ratio (defined as D/(D+T) for a D-T mixture) whose knowledge is of great importance for safety reasons and reaction performance control [5-7]. The algorithms can be also applied to impurity spectra to predict their plasma parameters such as the electron temperature. The candidate will have the task to develop a computer program (in Python) allowing to apply DNN and CNN algorithms to Ha/Da/Ta line spectra generated by an existing code for various conditions in terms of neutral temperatures, neutral population densities, magnetic field strength and hydrogen isotopic ratio. Thanks to the involvement in the tasks of data analysis of the EUROfusion workplan Tokamak Exploitation (TE) for several tokamaks including JET, the candidate may also apply the trained deep-learning models to experimental data from devices like JET and/or WEST.

References

  1. F. Pedregosa et al, the Journal of machine Learning research 12 (2011) 2825.
  2. S. Kajita et al, AIP Advances 10 (2020) 025225.
  3. D. Nishijima et al, Rev. Sci. Instrum. 92 (2021) 023505.
  4. S. Kajita et al, Plasma Phys. Control. Fusion 63 (2021) 055018.
  5. M. Koubiti and M. Kerebel, Appl Sci 12 (2022) 9891.
  6. M. Koubiti, Eur. Phys. J. D 77 (2023)137.
  7. N. Saura, M. Koubiti, S. Benkadda, Nuclear Materials and Energy 43 (2025) 101935.

This internship can be followed by a PhD thesis with funding by the doctoral school ED352

PhD Thesis (M/F) – Physics – Modeling – JR

Thesis advisor: Joël Rosato

Email and address: joel.rosato@univ-amu.fr

Tel: +33-413945714

 

Subject’s title: Characterization of white dwarf atmospheres by spectroscopic means

 

Subject description:

The theory of stellar structure knows three final states for a star: black holes, neutron stars and white dwarfs. According to observations and current models, the vast majority (of the order of 90%) of all stars, including our sun, will evolve towards the third final state, that of white dwarf [1,2]. These stars no longer burn nuclear fuel; instead, they are slowly cooling as they radiate away their residual energy. It is known today that white dwarfs support themselves against gravity by the pressure of degenerate electrons. They are referred to as compact objects because of their high density (up to 106 g/cm3). The characteristic cooling time of a white dwarf is closely related to the structure of its atmosphere, in particular its opacity to the radiation coming from the core. Studies have shown that the majority of white dwarfs have an atmosphere of pure hydrogen as a result of gravitational setting, which removes helium and heavier elements from the atmosphere and moves them towards inner layers [3,4]. These atmospheres can be considered as hydrogen plasmas, which are similar to some created in laboratory. Such white dwarfs are classified as of DA type due to the strong hydrogen absorption lines they present. The electron density in a white dwarf atmosphere is high enough (up to 1017 cm-3, and higher) so that the line shapes are dominated by Stark broadening and, hence, can serve as a probe for the electron density Ne. The goal of the PhD thesis is to improve the accuracy of the line shape models involved in white dwarf atmosphere diagnostics. Specific issues, such as the description of ion dynamics effects in Stark broadening [5], must be addressed. The observation of Zeeman pattern on several white dwarf spectra [6,7] has prompted a specific interest in the design of models accounting for the simultaneous action of electric and magnetic fields on the structure of atomic energy levels. Investigations must be done. The problem, which is similar to the modeling of spectra in magnetic fusion experiments, will possibly be tackled using models and codes previously developed in this framework and available at the laboratory. A part of the work will be devoted to the calculation of synthetic spectra and will involve the modeling of the stellar atmosphere structure.

 

Bibliography:

[1] S. L. Shapiro and S. A. Teukolsky, Black Holes, White Dwarfs, and Neutron Stars – The Physics of Compact Objects (Wiley, 2004).

[2] D. Koester and G. Chanmugam, Rep. Prog. Phys. 53, 837 (1990).

[3] G. Fontaine and G. Michaud, Astrophys. J. 231, 826 (1979).

[4] R. D. Rohrmann, Mon. Not. R. Astron. Soc. 323, 699 (2001).

[5] R. Stamm and D. Voslamber, J. Quant. Spectrosc. Radiat. Transfer 22, 599 (1979).

[6] B. Külebi et al., Astron. Astrophys. 506, 1341 (2009).

[7] S. O. Kepler et al., Mon. Not. R. Astron. Soc. 429, 2934 (2013).

PhD Thesis (M/F) – Physics – Modeling – MK

Thesis supervisor : Mohammed KOUBITI

Laboratory : PIIM, UMR7345 (http://piim.univ-amu.fr/)

Email & address : mohammed.koubiti@univ-amu.fr, Campus de St-Jérôme, Marseille, France.

Phone : +33 (0)4 13 94 64 47

Exploring Fusion Plasmas by combining spectroscopy and Scientific Artificial Intelligence

Subject description : Artificial intelligence tools are taking an important place in plasma science [1] and particularly in plasma physics [2-3]. Concerning plasma spectroscopy, several applications of machine-learning can be found including 2D beam emission spectroscopy in the DIII-D tokamak for real-time inference of plasma dynamics [4], neutral helium emission in linear plasma devices to predict plasma parameters using a support vector regression model (SVM) algorithm [5], or a more recent work on the Balmer-β line (Hβ /Dβ) in the WEST tokamak [6].

As non-invasive method, emission spectroscopy is widely used for diagnostics of magnetic fusion plasmas. Several parameters are diagnosed, e.g, the densities and temperatures of the electrons and main plasma ions, the impurity densities, and the temperatures and concentrations of the hydrogen isotope neutrals. Concerning hydrogen isotopes, the knowledge of the isotopic ratio D/(D+T) is of great importance since tritium inventory is mandatory in magnetic fusion devices operated with DT mixtures for obvious safety reasons. To infer the hydrogen isotopic ratio, we have built a predictive model based on the application of 1D Convolutional Neural Networks (1D-CNN) algorithms to theoretical Balmer Hα/Dα line spectra generated for neutral temperatures and magnetic field strengths typical of tokamak edge plasmas composed of HD mixtures [7].

In this thesis, it is proposed to develop predictive models based on different architecture of neural networks for more realistic conditions by considering plasmas composed of HD, DT as well as HDT mixtures but also to apply the models to experimental spectra from different tokamaks. We consider exploring experimental data measured in WEST (HD), JET (DT) through the EUROfusion WPTE involvement. Beyond their usefulness for future fusion devices like ITER, the development of predictive models will not be limited to Hα/Dα/Tα spectra but extended to other emission lines of H isotopes and impurity spectra like tungsten EUV spectra [8] in collaboration with spectroscopists from IRFM CEA or from other groups. The selected candidate will have the task to develop, test and validate computer programs coupling different neural network-based architectures to various emission spectra. Such programs should be more general to be extended to various diagnostics. Python and machinelearning/deep-learning skills as well as fusion plasma knowledges are greatly appreciated.

References

[1] E. Anirudh et al, IEEE Transactions on Plasma Science 51 1750 (2023)

[2] C. M. Samuell et al, Rev. Sci. Instrum. 92 043520 (2021)

[3] B. Dorland, Machine-Learning for Plasma Physics and fusion energy, Journal of Plasma Physics (2022)

[4] L. Malhorta et al, 4th IAEA Technical Meeting on fusion data processing, validation and analysis (2021)

[5] S. Kajita et al, AIP Advances, 10 025225 (2020)

[6] G. Ronchi et al, JQSRT 318 108925 (2024)

[7] N. Saura, M. Koubiti, S. Benkadda, Nucl. Materials Energy, 43 101935 (2025)

[8] N. Saura, R. Guirlet, M. Koubiti et al, Phys. Plasmas 32 083901 (2025).