normalize_log_posterior: Normalize log-posterior via simple numerical integration

In awstringer1/casecrossover: Approximate Bayesian Inference for Case Crossover Models

Description

Simple implementation of the trapezoid rule for integration. The key is that this function accepts grid points and LOGGED function values and computes the integral in a numerically stable manner, using matrixStats::logSumExp().

For a single hyperparameter, the function works on uniform and non-uniform grids. For a multidimensional hyperparameter, the function only works on a uniform grid (in all dimensions).

Usage

normalize_log_posterior(pp, tt)

Arguments

pp	logged function values
tt	grid points at which these function values were computed, as a NIGrid object

Value

A number giving the LOG of the integral of the function used to compute pp. This is the log-normalizing constant.

awstringer1/casecrossover documentation built on March 11, 2021, 4:41 a.m.