A Study of High-Resolution X-ray Scattering Data Evaluation by the Maximum-Entropy Method

The ability of the maximum-entropy method (in the program MAXENT) to estimate the distancedistribution function from high-resolution X-ray scattering data is studied. It is demonstrated that a key element for the successful application of M A X E N T is the use of a good prior estimate for the distance-distribution function. For simulated as well as experimental data, the effects of different priors, noise levels, smearing and measuring intervals are investigated. For practical applications of MAXENT, various methods for the calculation of priors are treated and a principle for the subsequent choice between the priors is suggested. It is demonstrated that, when the construction of the prior is given sufficient consideration, M A X E N T provides a very useful method for estimating the distance distribution from the scattering data.