Cobaya: Using External Power Spectrum

[Umang Kumar]

I am new to Cobaya and wish to figure out how to give an external power spectrum to CAMB from Cobaya input file (.yaml).

I have been using CAMB and setting my desired power spectrum using the method set_initial_power_function(Pk, args), where Pk is the desired power spectrum defined before as a python function.

I want to extend this to Cobaya and constrain a single parameter (say T) in my power spectrum. How should I go about doing this, specifying Cobaya to use this power spectrum?

I went through the docs but couldn't figure out how to connect my python file with Cobaya's input file.

Note: I am trying to run Cobaya from the shell, using the command cobaya-run [filename].yaml

[Antony Lewis]

I would probably make a new power spectrum class in CAMB (python) inheriting from SplinedInitialPower, writing your set_params method to take in your new parameter and assign the power spectrum within that function (calling inherited set_scalar_table with a spline sampling of your function).

Or you could modify CAMB's main set_params to call set_initial_power_function with your function for the current value of the parameters.

[Umang Kumar]

Hi Antony!

Thanks for replying!

I understand that I can make a python file and get a table of k and Pk from my defined power spectrum function and then feed it into CAMB but what I don't understand is that, how should I interface this with Cobaya? Where and how does this python file go in the Cobaya input file?

[Antony Lewis]

If you modify CAMB, then you just use your new parameters just like standard CAMB parameters.

But you can also do this entirely from the Cobaya side by writing a Theory component to implement your primordial power spectrum. There's a complete example in the tests at
https://github.com/CobayaSampler/cobaya/blob/master/tests/test_cosmo_multi_theory.py#L175

[Umang Kumar]

Thanks a lot! This helps!