MEASURING POLARIZATION IN THE COSMIC MICROWAVE BACKGROUND

Polarization induced by cosmological scalar perturbations leads to a t ypical anisotropy pattern, which can best be analyzed in a Fourier dom ain. This allows one to distinguish a cosmological signal of polarizat ion unambiguously from other foregrounds and systematics, as well as f rom polarization induced by nonscalar perturbations. The precision wit h which polarization and cross-correlation power spectra can be determ ined is limited by cosmic variance, noise, and foreground residuals. T he choice of estimator can significantly improve our capability of ext racting a cosmological signal, and in the noise-dominated limit the op timal power spectrum estimator reduces the variance by a factor of 2 c ompared to the simplest estimator. If foreground residuals are importa nt, then a different estimator can be used, which eliminates systemati c effects from foregrounds so that no further foreground subtraction i s needed. A particular combination of Stokes Q and U parameters vanish es for scalar-induced polarization, thereby allowing a direct determin ation of tensor modes. Theoretical predictions of polarization in stan dard models show that one typically expects a signal at the level of 5 -10 mu K on small angular scales and around 1 mu K on large scales (l < 200). Satellite missions should be able to reach sensitivities neede d for an unambiguous detection of polarization, which would help to br eak the degeneracies in the determination of some of the cosmological parameters.