Non-parametric estimates of high energy gamma-ray source distributions

We discuss the application of non-parametric techniques to EGRET data, to estimate source intensity distributions. The methods used are model independent, or they use the model as the null hypothesis in a non-parametric testing framework. Two particular techniques are applied, both employing wavelets. TIPSH denoising operates directly on the data, recovering the statistically significant residual based on a null hypothesis. Basis Pursuit Inversion also accounts for the instrumental response, and estimates the “unblurred” source distribution. In addition, the Basis Pursuit Inversion framework allows us to estimate point and diffuse contributions separately.