Implement interface with Candl

[yomori]

This PR implements a small code that interfaces with Candl , which allows you to interact with both published ACT/SPT likelihoods.

[joezuntz]

Thanks! This looks awesome. It looks like there's a data file I need to run this - could you point me to it so I can test locally?

[yomori]

Can you let me know which file it's looking for?
I think all the data products come packaged with Candl so if you just do pip install candl-like it should know where/what to look for. I just tried installing from scratch and was able to run the test sampler.

[yomori]

(cosmosis) [yomori@midway3-login4 cosmosis-standard-library]$ cosmosis examples/candl_test.ini
Setting up pipeline from parameter file examples/candl_test.ini
---------------------------------------------------------------

Setting up module consistency
------------------------------

Setting up module camb
-----------------------

Setting up module spt3g_2018_lensing
-------------------------------------
Successfully initialised candl likelihood 'SPT-3G 2018 PP, CMB corr. (Pan et al. 2023)' (type: <class 'candl.likelihood.LensLike'>).
Data loaded from '/project/chihway/yomori/envs/cosmosis/lib/python3.12/site-packages/candl/data/SPT3G_2018_Lens_v0/'.
Functional likelihood form: gaussian
--------------------------------------------------------------------------------
It will analyse the following spectra:

pp           (12 bins, bin centres spanning ell = 58.5 - 1735.5)
--------------------------------------------------------------------------------
A data model consisting of 2 transformations has been initialised.
The following transformations will be applied to the theory spectra in this order:

(1) Name: 
    Type: <class 'candl.transformations.common_lensing.BinnedTemplateForeground'>
(2) Name: 
    Type: <class 'candl.transformations.common_lensing.ResponseFunctionM'>
--------------------------------------------------------------------------------
Setup all pipeline modules


Parameter Priors
----------------
cosmological_parameters--omch2                  ~ U(0.08, 0.16)
cosmological_parameters--ombh2                  ~ N(0.02233, 0.00036 ** 2)   [0.01 < x < 0.03]
cosmological_parameters--log1e10as              ~ U(2.9, 3.2)
cosmological_parameters--n_s                    ~ N(0.96, 0.02 ** 2)   [0.86 < x < 1.06]
cosmological_parameters--cosmomc_theta          ~ U(0.9, 1.2)
cosmological_parameters--omega_k                ~ delta(0.0)
cosmological_parameters--mnu                    ~ delta(0.06)
cosmological_parameters--nnu                    ~ delta(3.046)
cosmological_parameters--yhe                    ~ delta(0.245341)
cosmological_parameters--tau                    ~ delta(0.0543)
cosmological_parameters--num_massive_neutrinos  ~ delta(3)
nuisance_parameters--a_fg                       ~ U(0.0, 2.0)

NOTE: You set resume=T in the [runtime] section but the sampler test does not support resuming yet.  I will ignore this option.
****************************************************
* Running sampler 1/1: test
* Running in serial mode.
****************************************************
Running module consistency
Done consistency status = 0 

Running module camb
Done camb status = 0 

Running module spt3g_2018_lensing
Done spt3g_2018_lensing status = 0 

Pipeline ran okay.
Using likelihooods from first run:
 - candl.spt3g_2018_lens
Likelihood candl.spt3g_2018_lens = -5.832610246814148
Prior      =  14.244080111236407
Likelihood =  -5.832610246814148
Posterior  =  8.411469864422259
(cosmosis) [yomori@midway3-login4 cosmosis-standard-library]$ 

[joezuntz]

Ah, it auto-installed along with pip, I should have realised. Will test this out, thanks!