Conclusions

Abstract

In this final short chapter we summarise the results of the thesis. The contamination from the intrinsic bispectrum generated by the non-linear effects generally leads to a small bias in the estimates of primordial non-Gaussianity. This is good news for the prospect of using CMB data to probe primordial non-Gaussianity, especially for experiments such as ESA Planck space telescope. While the precise answer depends on the terms included, the biases for local templates of non-Gaussianity are below the level of primordial $f_\text {NL}$ detectable by the Planck satellite. The biases from the intrinsic bispectrum for other primordial templates, equilateral and orthogonal, also appear to be small. The intrinsic non-Gaussianity can be searched for directly, using the predicted signal as a template; our calculations suggest this signal is just beyond what is possible with Planck, with a signal-to-noise rising to unity only for an angular resolution of about 4 arc minutes (${\ell _\text {max}} =3000$). In this concluding chapter we also discuss interesting extensions to our work and to our code, SONG, such as: computing the intrinsic power spectrum of the B polarisation; quantifying the effect of modified gravity on the intrinsic bispectrum; studying the generation of magnetic fields due to non-linear effects before and during recombination.