get_mcmc: Markov chain Monte Carlo (Gibbs) method to sample from the...

In rtlemos/scallops: Lemos and Sanso' (2012) model, edited for speed

Description

Markov chain Monte Carlo (Gibbs) method to sample from the conditional distributions of gamma and tau

Usage

get_mcmc(s, dpc_grid, y, nburn = 10, nsample = 10000, priors = NULL,
  printEvery = 1000, seed = NULL)

Arguments

s	Observation locations
dpc_grid	Discrete Process Convolution grid
y	Observation values
nburn	Number of EM iterations (burn-in) use dto estimate kernel parameters (rho)
nsample	Number of MCMC iterations
priors	Priors for gamma and tau
printEvery	Console notifications every X iterations
seed	Seed for reproducible chains

Value

List with sampled values of gamma and tau, as well as optimal estimates for rho

Examples

s = data.frame(lat = rep(0, 3), lon = c(-1:1))
y = c(1,1,0)
dpc_grid = get_grid(c(-1,1), c(0,0), spacing = 2)
priors = get_priors(dpc_grid)
iso_kernel_matrix = get_kernel_matrix(s, dpc_grid)
fit = get_mcmc(s, dpc_grid, y, 10, 1000, priors, 100, 1)
cbind(get_estimates(s, dpc_grid, fit), obs = y)

rtlemos/scallops documentation built on May 4, 2019, 7:43 p.m.