An Approach with the Akaike Information Criterion to the Radiation Distribution Reconstruction of Toroidal Plasma

The Akaike information criterion (AIC), that is, the well-established entropy in mathematical statistics, is adopted into the numerical analysis for toroidal-plasma image reconstruction. The bicubic spline function is taken as a linear model of the radiation intensity distribution in plasma and applied to the reconstruction problem with two-dimensional tangential projection. The criterion is used to find the optimum number of B-splines and, also, to evaluate the effect of adjusting their positions. The behavior of the estimator is examined by numerical simulation and by analyzing an X-ray pinhole image of tokamak plasma. The results suggest the versatile utility of AIC in least squares model-fitting approaches.