Inversion of Seismic Data by a Maximum Entropy Algorithm

In this paper we apply a maximum entropy (ME) approach to the inversion of linear geophysical data, with discrete data and continuous model parameters. This formulation is based on a probabilistic philosophy and on the concept of entropy, where we do not make use of prior information. It is defined an objective function which cont.ains the entropy of the probability density function of the model parameters. The objective function is then minimized under adequate constraints in order to give the output estimate of the model parameters. The tests with synthetic data in travel time tomography show the promissing application of the ME technique. The results with ME are better, both on qualitative and quantitative senses, when compared to SVD.