sim_glmmfields: Simulate a random field with a MVT distribution

In glmmfields: Generalized Linear Mixed Models with Robust Random Fields for Spatiotemporal Modeling

Simulate a random field with a MVT distribution

Description

Simulate a random field with a MVT distribution

Usage

sim_glmmfields(
  n_knots = 15,
  n_draws = 10,
  gp_theta = 0.5,
  gp_sigma = 0.2,
  mvt = TRUE,
  df = 1e+06,
  seed = NULL,
  n_data_points = 100,
  sd_obs = 0.1,
  covariance = c("squared-exponential", "exponential", "matern"),
  matern_kappa = 0.5,
  obs_error = c("normal", "gamma", "poisson", "nb2", "binomial", "lognormal"),
  B = c(0),
  phi = 0,
  X = rep(1, n_draws * n_data_points),
  g = data.frame(lon = runif(n_data_points, 0, 10), lat = runif(n_data_points, 0, 10))
)

Arguments

n_knots	The number of knots
n_draws	The number of draws (for example, the number of years)
gp_theta	The Gaussian Process scale parameter
gp_sigma	The Gaussian Process variance parameter
mvt	Logical: MVT? (vs. MVN)
df	The degrees of freedom parameter for the MVT distribution
seed	The random seed value
n_data_points	The number of data points per draw
sd_obs	The observation process scale parameter
covariance	The covariance function of the Gaussian process ("squared-exponential", "exponential", "matern")
matern_kappa	The optional matern parameter. Can be 1.5 or 2.5. Values of 0.5 equivalent to exponential model.
obs_error	The observation error distribution
B	A vector of parameters. The first element is the intercept
phi	The auto regressive parameter on the mean of the random field knots
X	The model matrix
g	Grid of points

Examples

s <- sim_glmmfields(n_draws = 12, n_knots = 12, gp_theta = 1.5,
  gp_sigma = 0.2, sd_obs = 0.2)
names(s)

glmmfields documentation built on Oct. 21, 2023, 1:06 a.m.