rprop.rwmh: Proposals for random walk Metropolis-Hastings

In lbelzile/mgp: Multivariate generalized Pareto models

Proposals for random walk Metropolis-Hastings

Description

This function transforms a vector cur to an unconstrained scale based on lbound and ubound, then samples draws from a multivariate Gaussian vector with covariance matrix cov centered at the current (unconstrained) value. The function then returns a list containing the proposal on the unconstrained scale (trprop), on the original scale (prop) along with the log of the jacobian of the transformation (logjac).

Usage

rprop.rwmh(cur = NULL, trcur = NULL, cov, lbound = rep(-Inf,
  ncol(cov)), ubound = rep(Inf, ncol(cov)))

Arguments

cur	vector of current parameter on the original (constrained) scale
trcur	vector of current parameter values on the unconstrained scale
cov	covariance matrix of the proposal for the trcur scale
lbound	lower bounds for the parameters
ubound	upper bounds for the parameter

Value

a list with components prop, trprop and logjac

lbelzile/mgp documentation built on Aug. 5, 2023, 2:34 a.m.