Cosmic microwave background power spectrum estimation with non-circular beam and incomplete sky coverage

Over the last decade, measurements of the cosmic microwave background (CMB) anisotropy have spearheaded the remarkable transition of cosmology into a precision science. However, addressing the systematic effects in the increasingly sensitive, high-resolution, 'full' sky measurements from different CMB experiments poses a stiff challenge. The analysis techniques must not only be computationally fast to contend with the huge size of the data, but the higher sensitivity also limits the simplifying assumptions which can then be invoked to achieve the desired speed without compromising the final precision goals. While maximum likelihood is desirable, the enormous computational cost makes the suboptimal method of power spectrum estimation using pseudo-C-l unavoidable for high-resolution data. The debiasing of the pseudo-C-l needs account for non-circular beams, together with non-uniform sky coverage. We provide a (semi) analytic framework to estimate bias in the power spectrum due to the effect of beam non-circularity and non-uniform sky coverage, including incomplete/masked sky maps and scan strategy. The approach is perturbative in the distortion of the beam from non-circularity, allowing for rapid computations when the beam is mildly non-circular. We advocate that it is computationally advantageous to employ 'soft' azimuthally apodized masks whose spherical harmonic transform die down fast with m. We numerically implement our method for non-rotating beams. We present preliminary estimates of the computational cost to evaluate the bias for the upcoming CMB anisotropy probes (l(max) similar to 3000), with angular resolution comparable to the Planck surveyor mission. We further show that this implementation and estimate are applicable for rotating beams on equal declination scans, and can possibly be extended to simple approximations to other scan strategies.