log_f_theta_f: Conditional for flux field parameters

In andrewzm/atminv: Reproducible package for Section 4.1 of "Spatio-temporal bivariate statistical models for atmospheric trace-gas inversion"

Description

Returns the conditional distribution of the flux field parameters given the flux field and, possibly, other flux maps. Uniform priors are assumed over the scale and smoothness, the bounds of which can be specified.

Usage

log_f_theta_f(theta, Z, D, flux_cov_fn, X, scale_ul = 2, nu_ul = 2,
  lambda_ll = -3, lambda_ul = 3, lambda_fix = NA, uncorrelated = FALSE,
  Pericchi = TRUE)

Arguments

theta	flux field parameters – see details
Z	matrix of flux field data (of size m x n_obs, where n_obs is generally 2; an inventory and the sample derived from the mole fraction observations)
D	distance matrix of flux field spatial locations
flux_cov_fn	covariance function used to generate the transformed correlation matrix. This should be of the form flux_cov_fn(D,scale,nu), where scale and nu are scale and smoothness parameters.
X	covariates matrix (of size n_f x p)
scale_ul	upper limit of uniform distribution over scale parameter. Default of 2
nu_ul	upper limit of uniform distribution over smoothness parameter. Default of 2
lambda_fix	use a fixed value for lambda (do not sample from it). If this is set then theta should only contain two numbers
lambda_l	absolute limit of uniform distribution over Box-Cox parameter. Default of 3

Details

To be completed

Examples

n_f <- 20
s <- 1:n_f
cov_fn <- function(D,scale,nu) {
     exp(-scale*(D^nu))    
}
print(log_f_theta_f(theta = c(1,1,0),
              Z = matrix(rpois(n_f,100)),
              D = as.matrix(dist(s)),
              flux_cov_fn = cov_fn,
              X = matrix(rnorm(n_f),n_f,1)))

andrewzm/atminv documentation built on May 10, 2019, 11:14 a.m.