A new method of solving noisy Abel-type equations (Q5935551)

The authors present and analyze a method of inverting the classical Abel integral equation of the first kind with order of singularity between 0 and 1. Assuming the data function to be perturbed by noise, they make for the solution an ansatz in form of a linear combination of orthogonal functions (in particular sine functions) and take statistical samples of the perturbed data function. Under a few special assumptions on the quality of the solution and the properties of noise they give a method of estimating the coefficients in their ansatz and show optimal convergence of the mean integrated squared error. They also compare their method with an alternative using wavelets.