E/B mode mixing

TL;DR

This paper discusses how finite sky coverage and pixelization cause E/B mode mixing in CMB polarization maps, affecting B mode detection, and proposes methods to separate pure and ambiguous modes for improved analysis.

Contribution

It introduces a practical approach to decompose polarization maps into pure and ambiguous modes, aiding accurate B mode characterization in future experiments.

Findings

Pure B mode information can be estimated from survey geometry.

Separation into pure and ambiguous modes can be achieved via differential equations.

The method is compatible with power spectrum estimation techniques.

Abstract

In future microwave background polarization experiments, particularly those that aim to characterize the B component, careful attention will have to be paid to the mixing of E and B components due to finite sky coverage and pixelization. Any polarization map can be decomposed into "pure E," "pure B," and "ambiguous" components. In practice, since the B component is expected to be much weaker than the E component, nearly all of the recoverable $B$ information is contained in the pure B component. The amount of B information lost to ambiguous modes can be estimated in simple ways from the survey geometry and pixelization. Separation of any given map into pure and ambiguous components can be done by finding a basis of pure and ambiguous modes, but it is often more efficient to "purify" the map directly in real space by solving a certain differential equation to find the ambiguous component. This method may be useful in conjunction with power spectrum estimation techniques such as the pseudo-Cl method.