Skip to content

Lightweight package for utilizing the Laplace Approximation to compare Bayesian models

Notifications You must be signed in to change notification settings

CCBeard/LApprox

Repository files navigation

LApprox

Lightweight package for utilizing the Laplace Approximation to compare Bayesian models.

LA

Project documentation can be found here: LApprox Documentation

LApprox.py is the only file required to run this package, though we include a Likelihood_Functions.py file for convenience. This project was designed for the purpose of calculating the Laplace Approximation for RV likelihood functions, and these are non-trivial to work with. Hence, we share our implementation.

An example of a simple, exact computation done for a multivariate Gaussian distribution is available in the Example_Multivariate_Gaussian.ipynb file.

A more complicated example, utilizing our Radial Velocity (RV) likelihood functions, is visible in Example_RV_Likelihood.ipynb.