Analysing VLBI images of astrophysical jets via cross-entropy global optimisation method

Most of jets detected in AGN blazar sources exhibit a morphological structure usually composed by a spatially unresolved core and jet knots receding from it at relativistic velocities. In some cases, the trajectories of the jet components on the plane of the sky seem to be bent, indicating the existence of some kind of acceleration in the respective motion. However, such claims depend strongly on the correct determination of the structural parameters of the jet components, usually obtained from model fitting procedures performed either in the (u, v) or in the image planes. In this work we introduce a new model fitting technique to obtain structural parameters of knots present in VLBI jet images. Our method that is based on the cross-entropy technique minimises an performance function that depends on the sum of the squared residuals obtained from the comparison of an VLBI image and a model image, constructed by summing Ns elliptical Gaussian synthetic sources. We present in this work the cross-entropy model fittings of benchmark images that were built to simulate most of the conditions encountered in typical VLBI images of active galactic nuclei. Besides recovering the parameters of the jet components in all validation tests, our method is able to point out quantitatively the number of the sources present in the image.