VLMCX (version 1.0)

simulate: Simulate a Variable Length Markov Chain with Exogenous covariates

Description

Simulate the states of a Markov Chain based on VLMCX model.

Usage

simulate(VLMCXtree, nsim = 500, X = NULL, seed = NULL, n.start = 100)

Value

a vector with nsim simulated states.

Arguments

VLMCXtree

a VLMCX tree (a VLMCX object can also be used, in which case its tree is used).

nsim

non-negative integer, giving the length of the result.

X

A vector or matrix of exogenous variables. If vector, its length must be equal to nsim+n.start, if matrix, its first dimension must be of length nsim+n.start, if NULL a univariate independent and identically distributed Normal vector is used.

seed

random seed initializer.

n.start

the number of initial values to be discarded (because of initial effects).

Author

Adriano Zanin Zambom <[e-mail]>

Examples

# \donttest{
tree = NULL
  tree$context = "x"  ## this is the root
  tree$alpha = NULL
  tree$beta = NULL
  tree$child = list()
  this_child = NULL
  this_child$context = "left"
  this_child$alpha = 0.5
  this_child$child = list()
  tree$child[[1]] = this_child

  this_grandchild = NULL
  this_grandchild$context = c("left", "left")
  this_grandchild$alpha = 0.6
  this_grandchild$beta = array(c(1.9, 1.6, 2.6, -1.6),c(2, 2, 1)) ## steps, d, alphabet
  this_grandchild$child = list()

  tree$child[[1]]$child[[1]] = this_grandchild

  this_other_grandchild = NULL
  this_other_grandchild$context = c("left", "right")
  this_other_grandchild$alpha = -0.6
  this_other_grandchild$beta = array(c(-1.3, -1.5, 2.3, -1.2),c(2, 2, 1)) 
  this_other_grandchild$child = list()

  tree$child[[1]]$child[[2]] = this_other_grandchild

  other_child = NULL
  other_child$context = "right"
  other_child$alpha = -0.7
  other_child$beta = array(c(1,-.3),c(1, 2, 1)) ## steps, d, alphabet
  other_child$child = list()
  tree$child[[2]] = other_child

  set.seed(1)
  X = cbind(rnorm(1100), rnorm(1100))
  simulated.data = simulate(tree, nsim = 1000, X, seed = 1, n.start = 100)
  fit = VLMCX(simulated.data$y, simulated.data$X, alpha.level = 0.001, 
                 max.depth = 4, n.min = 20, trace = TRUE)
  draw(fit)
  fit
# }