rmh_colwise_new: Function to implement the Metropolis step.

In yanbowisc/SIMP: Bayesian Simultaneous Partial Envelope Model

Function to implement the Metropolis step.

Description

Function to implement the Metropolis step.

Usage

rmh_colwise_new(
  A_start,
  lpd_func,
  method = "RW",
  tau = NULL,
  eps,
  eps_bar,
  H,
  mu,
  LL = 10,
  ColSigma.A.used,
  alpha,
  grad_lpd_func,
  M_adapt,
  M_diag,
  delta,
  iter,
  samp.size,
  ...
)

Arguments

A_start	A or B in the current iteration.
lpd_func	log full conditional density function of A or B.
method	Method for the metropolis. We have three choices: RW (default), HMC, NUTS. We recommend to use RW method always if no other reasons.
tau	The standard deviation for the proposal distribution used in the RW-Metropolis.
eps	Initial value for estimating the stepsize in NUTS-metropolis.
eps_bar	eps_bar parameter for NUTS-metropolis.
H	The H_m parameter for NUTS-metropolis.
mu	The mu parameter for NUTS-metropolis.
LL	Number of leapfrog steps in the trajectory
ColSigma.A.used	Covariance matrix for our matrix parameter A or B
alpha	Tempering parameter
grad_lpd_func	The function to calculate the gradients of the log full conditional posterior density of A or B.
M_adapt	The M_adapt parameter used in the Algorithm 6 of NUTS.
M_diag	Covariance matrix for our matrix parameter A or B.
delta	The delta parameter for NUTS-metropolis.
iter	The iteration index.
samp.size	Sample size.
...	Other parameters that may be useful.

yanbowisc/SIMP documentation built on Oct. 30, 2022, 1:33 a.m.