select_inverse_temp: Select inverse temperatures in RMC3 and RTMC3

In kkdey/tmcmcR: tmcmcR : a R/C++ package for Transformation based Markov Chain Monte Carlo

Description

The function selects the inverse temperatures using a Stochastic approximation algorithm for RMC3 or RTMC3 chains.

Usage

select_inverse_temp(pdf_component, minbeta = 0.01, L_iter = 50,
  sim_method = c("RWMH", "TMCMC"), inv_temp_scheme = c("randomized",
  "fixed"), rho_start = 0, scale = 0.1)

Arguments

pdf_component	The univariate marginal distribution of the iid product component target distribution
minbeta	the cut off of inverse temperature which when crossed, we stop at that value of inverse temperature. Helps to check the lower bound of the inverse temperatures. Default is 0.01.
L_iter	The number of sub-iterations required to fix each iterate of the inverse temperatures
sim_method	The method used to simulate from the marginal pdf component. Choices include TMCMC and RWMH.
inv_temp_scheme	The selection of inverse temperatures scheme followed, affects the stochastic approximation driving the inverse temperature updates. Choices include "fixed" and "randomized".
rho_start	The scale update for the inverse temperature selection. Default is 0.
scale	The stochastic approximation scaling parameter. Default is 0.1. The smaller this value, more inverse temperatures will be selected and algorithm will be more fine-grained.

Value

Returns a vector of inverse temperatures starting from 1 and ending at the first value of inverse temperature smaller than minbeta.

Author(s)

Kushal K Dey

kkdey/tmcmcR documentation built on May 20, 2019, 10:39 a.m.