Osnova / Outline:
A natural way of diagnosing fusion devices is to collect the visible light radiated by the interaction of the plasma and the neutral gas present in the vacuum vessel by high definition (~ up to 1 million pixels) cameras. It provides a very intuitive mean to understand the plasma edge phenomena, similarly to eye observations. Moreover, it is non-perturbative, as it does not require to probe the burning plasma itself. On the other hand, interpretation of such data is usually not trivial as the observed events are generally three-dimensional objects projected onto the two-dimensional plane of the camera chip and there is, therefore, a missing dimension. A solution to overcome this problem is to assume a symmetry in the system and to perform tomographic inversion [1] in order to retrieve a 2D emissivity poloidal plane cut from the camera data.

We propose to perform tomographic inversion of COMPASS camera data during RE experiments (an experiment during which relativistic electrons are created in the tokamak). The student will work with two already developed algorithms [2]. The aim of the study is to assess the feasibility of doing RE tomography towards the Japanese/EU JT-60SA tokamak.

Skills that the student will acquire:

• Working on already existing data from a relevant fusion device (COMPASS)
• Getting familiar with the visible camera diagnostic, a diagnostic widely used in the
  fusion community
• Getting familiar with tomographic inversion that gains more and more interest in the
  community

The work of the bachelor student will be the following:

• Learn how to model the camera view thanks to the Calcam software
• Perform tomographic inversion of camera data with existing Python programs
• Compare the output from two different algorithms
• Investigate the possibility to improve existing inversion algorithms
• Possibly analysing different plasma scenarios, depending on the student’s skills

Literatura / reference:
[1] J. Cavalier et al. Nucl. Fusion 59 (2019) 056025 (16pp).
https://iopscience.iop.org/article/10.1088/1741-4326/ab0d4c
[2] J. Svoboda et al 2021 JINST 16 C12015.
https://iopscience.iop.org/article/10.1088/1748-0221/16/12/C12015

Outline:

With increasing capabilities of high-performance computing kinetic modelling of plasma edge has become a fast-growing branch of the magnetic confinement fusion plasma study. Our day kinetic simulations of the plasma edge, performed via PIC (Particle-in Cell) codes, incorporate nonlinear dynamics of plasma, neutral and impurity particles and significantly contribute to optimization of design of future tokamaks (see [1-3] and references their). One of the bottlenecks of these PIC codes is the requirement of equal weights of simulation particles necessary for used collision operators. Therefore, the treatment of very low concentration species, such as the highly ionized states of impurity particles, becomes inaccurate in these models.

The aim of this PhD work is to develop new collision operators, not requiring same weights of collided particles; their implementation into the existing PIC codes (e.g. BIT, SPICE etc.) and performing simulations of the tokamak Scrape-off Layer (i.e. edge plasma layer with open magnetic surfaces) including highly ionized states of impurity particles. The obtained results will be used for design and optimization plasma performance in future fusion devices (COMPASS-U, ITER and DEMO).

References:

[1] D. Tskhakaya, Plasma Phys. Control. Fusion, 59 (2017) 114001

[2] T. Takizuka, et al., Contrib. Plasma Phys., 59 (2017) 034008

[3] D. Tskhakaya, et al., Nucl. Mater. and Energy, 26 (2021) 100893

Outline:

The PIC (Particle-in Cell) codes represent powerful tools for plasma transport study [1,2]. They are frequently used in fusion plasma applications and contribute to design of next generation fusion devices. As it was demonstrated in [3], PIC simulations with Cartesian grid do conserve momentum and energy simultaneously only for uniform grids. This requirement represents very heavy condition for plasma edge modelling, when nonuniform space resolution is favorable. Some time ago a solution for 1D codes has been found [4], but so far up to now no multidimensional generalization of this approach has been done.

The aim of this PhD work is to develop a multidimensional energy and momentum conserving PIC code with nonuniform Cartesian grid. We expected significant speedup, by order of magnitude, of such code against the conventional codes, which will save tens of millions of CPU hours per year for average PIC user on high performance computer facilities.

References:

[1] D. Tskhakaya, Plasma Phys. Cont. Fus., 59 (2017) 114001

[2] T. Takizuka, et al., Contrib. Plasma Phys., 59 (2017) 034008

[3] D. Tskhakaya, et al., Contrib. Plasma Phys., 47 (2007) 563

[4] J. Duras, et al., Contrib. Plasma Phys., 54 (2014) 697

Osnova:

PIC (Particle-in Cell) kódy jsou výkonnými nástroji pro studium transportu v plazmatu [1,2]. Jsou často aplikovány na fúzní plazma a přispívají k projektování příští generace fúzních experimentů. Jak bylo ukázáno v [3], PIC simulace s kartézskými souřadnicemi zachovávají zároveň hybnost a energii pouze pro pravidelné sítě. Tento požadavek představuje velmi silný předpoklad pro modelování okraje plazmatu, kdy je mnohem výhodnější nepravidelné rozlišení prostorových elementů. Před několika lety bylo nalezeno řešení pro 1D případ [4], ale až dosud nebyla tato metoda zobecněna na vícedimenzionální sítě.

Cílem této doktorské práce bude vyvinout vícerozměrný PIC kód s nepravidelnou kartézskou sítí. Očekáváme podstatné zrychlení takového kódu, až o řád, oproti obdobným konvenčním kódům. To by ušetřilo desítky miliónů CPU hodin ročně pro průměrného uživatele PIC modelů ve vysokovýkonných výpočetních centrech.

Literatura:

[1] D. Tskhakaya, Plasma Phys. Cont. Fus., 59 (2017) 114001

[2] T. Takizuka, et al., Contrib. Plasma Phys., 59 (2017) 034008

[3] D. Tskhakaya, et al., Contrib. Plasma Phys., 47 (2007) 563

[4] J. Duras, et al., Contrib. Plasma Phys., 54 (2014) 697