rnorm_tails: Random generation of normal mixture deviations
In sebpardo/rtails: Simulate heavy-tailed deviations

Description Usage Arguments Details Value Examples

View source: R/rnorm_tails.R

Random generation for a mixture of normal distributions, both centered around zero and bias corrected, as well as exponentiated.

rnorm_tails(
  n,
  sigma = 0.1,
  high_sigma = 2,
  rate = 1/38,
  bias_correct = TRUE,
  sample_bias_correct = FALSE,
  ac = 0,
  log = FALSE,
  skew = NULL,
  seed = NA,
  replace = FALSE
)

`n`	sample size, or a vector of random deviates to update if 'replace = TRUE'.
`sigma`	variance of the underlying normal distribution, defaults to 0.1. Ignored if 'replace = TRUE'.
`high_sigma`	variance of the sporadic normal distribution, should be higher than 'sigma', defaults to 1.5.
`rate`	rate at which the wider sporadic normal distribution replaces the underlying distribution. This is determined by a binomial process using 'rbinom(1, n, rate)'.
`bias_correct`	logical. Should we bias correct analytically.
`sample_bias_correct`	logical. Should we bias correct using the sample mean, which results in a mean of exactly 1. Defaults to 'FALSE'. If 'bias_correct = TRUE' then this parameter is ignored (i.e. equals 'FALSE').
`ac`	auto-correlation value, between -1 and 1. If 'ac != 0' autocorrelation is incorporated in the vector using an AR(1) process, and is applied after exponentiating except when 'log = TRUE'.
`log`	logical. Whether to return the distribution before it is bias corrected and exponentiated. This is not equivalent to log-transforming the distribution when 'log = FALSE'.
`skew`	value of skewness parameter. When set, the skew-normal distribution from the 'sn' package is used (i.e. 'sn::rsn'), therefore when setting 'seed' the drawn values will probably differ to those without skew. Defaults to 'NULL', where no skewness is incorporated and the traditional 'rnorm' function is used.
`seed`	seed. Numeric for 'set.seed()'. Defaults to NA where no seed is set.
`replace`	logical. Whether to replace a vector of deviates using sporadic values or create it from scratch. If 'TRUE', then 'n' needs to be a vector of positive deviates centered around 1.

Both normal distributions are bias corrected independently so that their means in log-space equal to 1 (-σ^2/2).

a vector of random deviates of length 'n'. If there were sporadic deviates where the 'high_sigma' distribution replaced the underlying one, then the positions of these events are stored in the 'ht' attribute of the vector.