Comparison of the three tokamak plasma tomography methods for high spatial resolution and fast calculation

Abstract

High spatial resolution tomographic reconstructions with several thousand pixels were performed by three different reconstruction algorithms. To overcome the increased ill posedness due to the use of a large number of pixels, the fast maximum entropy method (MEM) and the Phillips-Tikhonov method were used without any biasing information. The developed spatially resolved fast tomography codes were tested with the KSTAR plasma-like emissivity phantom and the TCV soft x-ray data. The reconstruction results were compared with those by the minimum Fisher information method, and these three methods shows the reliable reconstruction results. Speedup of the calculation was attempted by implementing the parallel computing technique via MATHLINK in the fast MEM code written in the MATHEMATICA language. The calculation speed of the fast MEM code was improved by about ten times without loss of accuracy despite the large number of pixels (2700 and 3500 pixels). (c) 2006 American Institute of Physics.