$B$-mode forecast of CMB-Bh$\overline{a}$rat

TL;DR

This paper evaluates the potential of the proposed ECHO space mission to detect primordial gravitational wave signals via CMB B-mode polarization, analyzing component separation methods and foreground models to assess sensitivity and accuracy.

Contribution

It provides a detailed forecast of ECHO's capability to measure the tensor-to-scalar ratio r around 10^{-3}, comparing NILC and Commander pipelines under various foreground conditions.

Findings

Both pipelines can achieve the target sensitivity for simple foreground models.

NILC outperforms Commander with complex foreground models in bias reduction.

Delensing improves sensitivity by approximately 50%.

Abstract

Exploring Cosmic History and Origins (ECHO), popularly known as `CMB-Bh$\overline{a}$rat', is a space mission that has been proposed to the Indian Space Research Organisation (ISRO) for the scientific exploitation of the Cosmic Microwave Background (CMB) at the next level of precision and accuracy. The quest for the CMB polarization $B$-mode signals, generated by inflationary gravitational waves in the very early universe, is one of the key scientific goals of its experimental design. This work studies the potential of the proposed ECHO instrumental configuration to detect the target tensor-to-scalar ratio $r \sim 10^{-3}$ at $3\sigma$ significance level, which covers the predictions of a large class of inflationary models. We investigate the performance of two different component separation pipelines, NILC and Commander, for the measurement of $r$ in presence of different physically motivated models of astrophysical foregrounds. For a simplistic foreground model (only polarized dust and synchrotron), both component separation pipelines can achieve the desired sensitivity of ECHO, i.e. $\sigma (r =0) \sim (0.4 - 0.7)\times 10^{-3}$. NILC performs better than Commander in terms of bias on recovered $r$ for complex spectral models (power-law and curved power-law) of the synchrotron emission and complex dust models (dust decorrelation). Assuming 84 % delensing, we can achieve an improvement of $\sigma (r = 0)$ by approximately 50 % as compared to the results obtained for the same configuration without any lensing correction.