Bayesian deconvolution. I

An iterative procedure to achieve spectral deconvolution by application of Bayes' postulate is presented . An analytical treatment for a Gaussian response function is used to indicate how the resolution attained depends upon the number of iterations. The damping effect produced by closely spaced doublets is explored in addition to the convergent properties for various types of response functions. The positivity constaint inherent in the algorithm and the improvement in achieved resolution indicate the method will prove to be very powerful for spectral deconvolution for a wide range of response functions. (Auth.)