1602.02167
Firedec: a two-channel finite-resolution image deconvolution algorithm
Cantale, Courbin, Tewes, Jablonka, Meylan
Present a two-channel deconvolution method that decomposes images into a parametric point-source channel and a pixelized extended-source channel. Based on the central idea of the deconvolution algorithm proposed by Magain, Courbin & Sohy (1998), the method aims at improving the resolution of the data rather than at completely removing the PSF. Improvements over the original method include a better regularization of the pixel channel of the image, based on wavelet filtering and multiple analysis, and a better controlled separation of the point source vs. the extended source. In addition, the method is able to simultaneously deconvolve many individual frames of the same object taken with different instruments under different PSF conditions. For this purpose, introduce a general geometric transformation between individual images. This transformation allows the combination of the images without having to interpolate them. Illustrate the capability of the algorithm using real and simulated images with complex diffraction-limited PSF.
1602.02960
Discrepancies between CFHTLenS cosmic shear & Planck: new physics or systematic effects?
Kitching, Verde, Heavens, Jimenez
There is currently a discrepancy in the measured value of the amplitude of matter clustering, parameterized using sigma8, inferred from WL and CMB data. Investigate what could mitigate this discrepancy. Consider systematic effects in the WL data and include IA, and biases in photometric z estimates. Also apply a non-parametric approach to model the baryonic feedback on the DM clustering, which is flexible enough to reproduce the OWLS and Illustris sims. Finally, extend the cosmo analysis of the WL data to include the effect of massive neutrinos. The statistic used, 3D cosmic shear, is a method that extracts cosmo info from WL data using a spherical-Bessel function PS approach. There are several advantages that this affords, in particular that the method does not rely on binning in z, or covariance estimation from sims. It also allows for a robust scale-dependent analysis of data. Analyse the CFHTLenS WL data and, assuming best fit cosmo parameters from the Planck CMB experiment, find that there is no evidence for baryonic feedback on the DM PS; if the IA amplitude is close to zero, then there is evidence for a bias in the photo-z in CFHTLenS data. Also find an upper limit to the sum of neutrino masses, directly from the shape of the matter PS, <0.28 eV (1 sigma), similar to the one obtained from other indirect probes of the matter distribution.
No comments:
Post a Comment