Likelihood Module

MGLensing.likelihood

class MGLensing.likelihood.MGLike(Model, Data, use_jax=False)

Bases: object

loglikelihood(param_dic)

Calculate the log-likelihood for a given set of parameters using the difference between model’s and data’s vectors.

loglikelihood_binned_wrapper(param_dic)

Binned log-likelihood: use JAX path if enabled and available, else Python.

loglikelihood_det_3x2pt(param_dic)

Compute the log-likelihood for a 3x2pt analysis using the determinant method from https://arxiv.org/pdf/1210.2194. Applicable only for the same scale-cuts per probe in all redshift bins. The \(\chi^2\)-value is computed as

\[\chi^2 = \sum_{\ell_{\rm min}}^{\ell_{\rm max}} (2\ell+1) f_{\rm sky} \left( \frac{d^{\rm mix}}{d^{\rm theo}} + \ln{\frac{d^{\rm theo}}{d^{\rm obs}}} \right)\, ,\]

where the determinants are computed for each \(\ell\)-value from \(N \times N\) matrices with \(C_\ell\)-values, \(N\) is the number of photo-z bins. The superscripts meaning: “obs” stays for the observed/mock data; “theo” – theoretical model prediction; “mix” – mix from theoretical and data \(C_\ell\) (see Eq. 3.2 in https://arxiv.org/pdf/2404.11508); “high” corresponds to the WL \(C_\ell\)’s with \(\ell > \rm{max}(\ell_{\rm GC})\).

loglikelihood_det_gc(param_dic)

Compute the log-likelihood for a given set of parameters using the determinant method.

loglikelihood_det_wl(param_dic)

Compute the log-likelihood for a given set of parameters using the determinant method.