A non-parametric method to detect clusters in Planck data

In this work we show how to get a very significant number of SZ detections in the future Planck data without doing any of the typical assumptions needed in present component separation methods. That is, we do not need to assume anything about the power spectrum nor the frequency dependence of any one of the components, circular symmetry or a typical scale of the clusters. Our method predict a number of detections $\approx 14.000$ in all the sky with an unbiased estimation of the total flux for clusters with $S > 150$ mJy (at 353 GHz), although with some scatter around the real flux, this proving the robustness of our method. This large number of SZ detections will allow a robust and consistent analysis of the evolution of the cluster population with redshift and will have important implications on the estimation of the best cosmological model.