0410-Planck 2018 V Planck 2015 IX

PAPER

Planck 2018 V

PCL for high l powerspectrum estimation, ignore the bias caused by point sources.

Planck 2015 IX

In addition to producing CMB maps, each method provides “confidence” masks to define the region of the sky in which the CMB solution is trusted in temperature and polarization. The procedure each method uses to define the mask is described in Appendices A–D.

Both, the standard Commander and SEVEM CMB maps are inpainted inside their corresponding confidence masks using a constrained realization (see Appendix A for details). For both methods, only the CMB maps are inpainted, not the Monte Carlo realizations as this would be too computationally expensive. The other two methods, NILC and SMICA, work on E and B modes, so it is possible to make E and B maps directly from their outputs in addition the the standard Q and U maps.

we emphasize that the Planck 2015 results parameters are not based on the high-resolution foreground-cleaned CMB maps presented in this paper, but are instead derived from the likelihood described in Planck Collaboration XIII (2015). That likelihood combines the low-resolution Commander temperature map derived from Planck, WMAP, and 408 MHz with a template-cleaned LFI 70 GHz polarization map in a pixel-based low-l likelihood, and adopts a cross-spectrum-based estimator for the high-l temperature and polarization likelihood. The parameter estimation methodology that we use here is primarily a tool to evaluate the quality of the high-resolution CMB maps and to assess overall consistency with the Planck 2015 likelihood.

While Commander performs data selection at the detector (or detector set) map level, rejecting potentially problematic maps, the other three methods employ frequency channel maps, and thereby maximize the signal-to-noise ratio. Likewise, while Commander and SEVEM perform their analyses in pixel space, NILC and SMICA perform all operations in harmonic space. These distinctions can explain many of the qualitative differences discussed in the previous sections.

As in 2013, the temperature maps presented here are therefore used for a wide range of important applications, including large-scale temperature likelihood estimation (Planck Collaboration XI 2015), gravitational lensing (Planck Collaboration XV 2015), studies of isotropy and statistics (Planck Collaboration XVI 2015), primordial nonGaussianity (Planck Collaboration XVII 2015), and non-trivial cosmological topologies (Planck Collaboration XVIII 2015). For high-l power spectrum and likelihood estimation, we recommend the cross-spectrum based methods described in Planck Collaboration XI (2015), primarily due to difficulties in establishing sufficiently accurate models of unresolved extra-galactic high-l foregrounds for the maps presented here.

BRAINSTORM

Will my template fitting better than PCL method?