search

andreufont / LyaCosmoParams

Notes on cosmological parameters and Lyman alpha simulations
2 stars 2 forks source link

CAMB speedup bug & likelihood parameter bug #68

Closed Chris-Pedersen closed 4 years ago

Chris-Pedersen commented 4 years ago

When checking which parameters to build a new camb cosmology from, CAMB_model was using likelihood_parameter.is_same_parameter(test_parameter), which will only pass the new value if both the parameter limits and the name are the same. We only care if the name is the same, as we only pass the value, not the value in the unit cube. This meant that n_s was not being changed, only A_s (I noticed this as when I relaxed the priors in sampler tests on the primordial power spectrum we had no constraining power over n_s). My fix is to change this to only check if the parameter names are the same.

This also revealed a bug in the CAMB speedup which is now fixed.

Chris-Pedersen commented 4 years ago

This is ready to merge now - fixing the fact that n_s wasn't being changed when using the likelihood object will probably give different results for the likelihood_test notebooks. It would be worth looking at those next week, I will not have time to go over this today.