Forecasting performance of CMB experiments in the presence of complex foreground contaminations

We present a new, semianalytic framework for estimating the level of residuals present in cosmic microwave background (CMB) maps derived from multifrequency CMB data and forecasting their impact on cosmological parameters. The data are assumed to contain non-negligible signals of astrophysical and/or Galactic origin, which we clean using a parametric component separation technique. We account for discrepancies between the foreground model assumed during the separation procedure and the true one, allowing for differences in scaling laws and/or their spatial variations. Our estimates and their uncertainties include both systematic and statistical effects and are averaged over the instrumental noise and CMB signal realizations. The framework can be further extended to account self-consistently for existing uncertainties in the foreground models. We demonstrate and validate the framework on simple study cases which aim at estimating the tensor-to-scalar ratio, r. The proposed approach is computationally efficient permitting an investigation of hundreds of setups and foreground models on a single CPU.