Pattern extraction from large scale plasma turbulence simulation data with principal component analysis

Asahi, Yuichi   

Phase space structures are extracted from the time series of five dimensional distribution function data computed by the flux-driven full-F gyrokinetic code GT5D. Using the principal component analysis (PCA), the dimensionality and the size of the 6D (3D space and 2D velocity space and 1D time) data is reduced from 10 TB to 10 GB. Phase space bases and the corresponding spatial coefficients (poloidal cross section) are constructed by PCA. It is shown that 83% of the variance of the original 6D data can be expressed with 64 principal components. The relationship between the avalanche-like transport phenomena and phase space structures is discussed based on the contribution of each principal component to the energy transport.