Soft x-ray tomography using L1 regularization for MHD modes with limited sight lines in JT-60SA.

Abstract

Soft x-ray (SX) tomography is a useful diagnostic in fusion research, and a multi-channel SX diagnostic will be installed in JT-60SA, the largest elongated tokamak in the world. However, in the SX diagnostic of JT-60SA, plasmas will be only viewed from the low field side and the upper side of plasmas; the sight lines are limited, which would be common in future devices as well as JT-60SA. This kind of limited sight lines is not preferred for SX tomography to investigate the spatial structure of magnetohydrodynamics (MHD) modes because inadequate information of plasmas makes artifacts in the reconstructed SX profiles. One of the solutions to reduce the artifacts is to employ L1 regularization, which gives the essential and sparse contributions [Kaptanoglu et al., Phys. Plasmas 30, 033906 (2023)]. In this study, as a first topic, the applicability of L1 regularization to reduce the artifacts in SX tomography with limited sight lines is investigated with traditional L2 regularization for a high beta scenario of JT-60SA where MHD modes would occur. Here, as a series of basis functions, the Fourier-Bessel series (FBS) is employed because FBS has the poloidal Fourier modes explicitly. A disadvantage of FBS is that the accurate equilibrium inside the last closed flux surface (LCFS) is needed; interior measurement such as the motional Stark effect measurement is required, which is not always available during a whole discharge. The second topic of this study is to investigate other appropriate basis functions to study the spatial structure of MHD modes in elongated tokamak plasmas. Here, we introduce Saito's Laplacian eigenfunction (LEF). Saito's LEF can be calculated if LCFS is given and the LEF is expected to show the explicit poloidal Fourier mode. Because the calculation of LCFS with magnetic measurements is a basic task of plasma operations, Saito's LEF may be used anytime. Our investigation showed that L1 regularization can strongly improve the SX tomography with the traditional L2 regularization having FBS/LEF and would be effective against other tomographic problems in fusion devices.