Create smooth fiducial data for testing

[mgerbino]

We should create a set of smooth simulated data (i.e., direct predictions from Boltzmann solvers) that share the same set of input cosmo params for testing purposes. This is mostly relevant when testing combined likelihoods.

[mgerbino]

This is slowly advancing. Current status can be viewed here: https://www.overleaf.com/read/bqjffrkryfps

Currently available, see notebooks create_xxx_dataset.ipynb at https://github.com/simonsobs/SOLikeT/tree/dev-smoothTheory/notebooks/dev:

• MFLike
• lensing

Models currently tested:

• LCDM
• LCDM+sumnu

Bottom line: agreement with Fisher forecast presented in Science goals paper (https://arxiv.org/pdf/1808.07445.pdf)

To do:

• add new smooth dataset (xcorr, clusters)
• explore more LCDM extensions