Talk detail

Participant	Wandelt, Benjamin
Institution	Institut d'Astrophysique de Paris - 98 bis, blvd Arago - Paris - Ile-de-France - France

Session	GL2	Accepted		Order	Time
Talk	Oral abstract	Title	Beyond The Quadratic Estimator - Efficient Weak Lensing Inference For The Cosmic Microwave Background
Coauthors
Abstract	It is well-known that the cosmic mass distribution lenses the cosmic microwave background anisotropies. The standard methodology for extracting this signal from the CMB map uses an estimator based on quadratic combinations of the pixels. This estimator is known to be sub-optimal in the case of high S/N data -- the relevant regime for the next generation of CMB experiments that go after lensing B-modes either to exploit them scientifically or to remove them in the search for primordial gravitational waves. We show how to circumvent the main problem that has prevented an efficient, principled Bayesian approach so far and demonstrate the feasibility and optimality of a full statistical analysis using simulated skys.
Pdf file

Session	GL2	Accepted		Order	Time
Talk	Oral abstract	Title	Beyond The Quadratic Estimator: Full Information Inference of Cosmic Microwave Background Lensing
Coauthors
Abstract	The age of quantitative scientific exploitation of the gravitational lensing distortions of the cosmic microwave background (CMB) is upon us: the Planck satellite, along with several ground based telescopes, have mapped the CMB at sufficient resolution and signal-to-noise so as to allow a detection of the subtle distortions due to the gravitational influence of the intervening matter distribution. All current analyses use the so-called quadratic estimator. This approach is known to be sub-optimal for the high signal-to-noise, high resolution data of next generation experiments. A natural and optimal modeling approach is to write a Bayesian hierarchical model for the lensed CMB in terms of the unlensed CMB and the lensing potential, but inference based on this Bayesian approach has been computationally unfeasible. We solve this problem using a reparametrization to allow efficient Markov Chain Monte Carlo sampling from the joint posterior of the lensing potential and the unlensed CMB map using the Hamiltonian Monte Carlo technique and the messenger algorithm. We demonstrate a fast implementation on simulated data including noise and a sky cut that uses a further acceleration based on a very mild approximation of the inverse lensing potential. The improved resulting lensing reconstructions are making our approach promising for application to the high resolution CMB data sets of the near future.
Pdf file