This afternoon, I worked on a model implementation of the scale-dependent bias, following the ansatz of De Putter & Doré (2014, their Eq. 1). The goal is to constrain the fNL parameter which quantifies the amplitude of the primordial non-Gaussianity. These primordial density fluctuations are very Gaussian, but a small departure from this Gaussianity or limits on the value of fNL would provide strong constraints on models of inflation.
We are wrapping up the project in which we cross correlate the signal from the lensing of the CMB and the cosmic infrared background (CIB). At large scales, the imprint of the non-Gaussian component is particularly strong, hence large-scale data sets will be required for this.
To implement this model for the angular power spectrum, I worked with the Core Cosmology Library (CCL) which has recently been released and is very actively developed. Even though the focus are optical surveys in the context of LSST and the Dark Energy Science Collaboration, it is certainly worth checking out. While the code is C/C++ under the hood, a lot of effort goes into a stable, well-documented Python wrapper with a proper pythonic API, which I miss in other Boltzmann codes such as CAMB and Class.