Comparative investigation of unfolding methods

The problem of reconstructing the function S(x) from the finite number of measured functionals anti Yn = ∫dxAn(x)S(x) is studied in the class of linear estimates. Functions An(x) characterise the resolution of the experiment or the response function of the detector. The bias and error operator of estimate S(x) are studied as well. The least squares method, the spectral window method, the iterative method. Tikhanov's regularization method and others are investigated as unfolding (deconvolution, unsmearing) methods. These methods are compared for two numerical examples. The problem of choosing the regularization parameters and some related problems are also discussed. (orig.)