rvar_factor: Factor random variables of arbitrary dimension

View source: R/rvar-factor.R

rvar_factorR Documentation

Factor random variables of arbitrary dimension


Random variables backed by factor-like arrays of arbitrary dimension.


  x = factor(),
  dim = NULL,
  dimnames = NULL,
  nchains = NULL,
  with_chains = FALSE,

  x = ordered(NULL),
  dim = NULL,
  dimnames = NULL,
  nchains = NULL,
  with_chains = FALSE,



(multiple options) The object to convert to an rvar:

  • A vector of draws from a distribution.

  • An array where the first dimension represents draws from a distribution. The resulting rvar will have dimension dim(x)[-1]; that is, everything except the first dimension is used for the shape of the variable, and the first dimension is used to index draws from the distribution (see Examples). Optionally, if with_chains == TRUE, the first dimension indexes the iteration and the second dimension indexes the chain (see with_chains).

  • An rvar.


(integer vector) One or more integers giving the maximal indices in each dimension to override the dimensions of the rvar to be created (see dim()). If NULL (the default), dim is determined by the input. NOTE: This argument controls the dimensions of the rvar, not the underlying array, so you cannot change the number of draws using this argument.


(list) Character vectors giving the names in each dimension to override the names of the dimensions of the rvar to be created (see dimnames()). If NULL (the default), this is determined by the input. NOTE: This argument controls the names of the dimensions of the rvar, not the underlying array.


(positive integer) The number of chains. The if NULL (the default), 1 is used unless x is already an rvar, in which case the number of chains it has is used.


(logical) Does x include a dimension for chains? If FALSE (the default), chains are not included, the first dimension of the input array should index draws, and the nchains argument can be used to determine the number of chains. If TRUE, the nchains argument is ignored and the second dimension of x is used to index chains. Internally, the array will be converted to a format without the chain index. Ignored when x is already an rvar.


Arguments passed on to base::factor


an optional vector of the unique values (as character strings) that x might have taken. The default is the unique set of values taken by as.character(x), sorted into increasing order of x. Note that this set can be specified as smaller than sort(unique(x)).


either an optional character vector of labels for the levels (in the same order as levels after removing those in exclude), or a character string of length 1. Duplicated values in labels can be used to map different values of x to the same factor level.


a vector of values to be excluded when forming the set of levels. This may be factor with the same level set as x or should be a character.


logical flag to determine if the levels should be regarded as ordered (in the order given).


an upper bound on the number of levels; see ‘Details’.


A subtype of rvar() that represents a (possibly multidimensional) sample of a factor or an ordered factor. It is otherwise very similar to the basic rvar(): it is backed by a multidimensional array with draws as the first dimension. The primary difference is that the backing array has class "factor" (for rvar_factor()) or c("ordered", "factor") (for rvar_ordered()). If you pass a factor or ordered factor to rvar() it will automatically return an object with the classes "rvar_factor" or c("rvar_ordered", "rvar_factor").

See rvar() for more details on the internals of the random variable datatype.


An object of class "rvar_factor" representing a factor-like random variable.

See Also

as_rvar_factor() to convert objects to rvar_factors. See rdo(), rfun(), and rvar_rng() for higher-level interfaces for creating rvars.



# To create a "scalar" `rvar_factor`, pass a one-dimensional array or a vector
# whose length (here `4000`) is the desired number of draws:
x <- rvar(sample(c("a","a","a","b","c"), 4000, replace = TRUE))

# Create random vectors by adding an additional dimension:
x_array <- array(c(
    sample(c("a","a","a","b","c"), 4000, replace = TRUE),
    sample(c("a","a","b","c","c"), 4000, replace = TRUE),
    sample(c("b","b","b","b","c"), 4000, replace = TRUE),
    sample(c("d","d","b","b","c"), 4000, replace = TRUE)
  ), dim = c(4000, 4))

# You can also create ordered factors

# arguments of factor() and ordered() are passed down by the constructor
# e.g. we can reorder levels of an ordered factor:
rvar_ordered(x_array, levels = c("d","c","b","a"))

# Unlike base factors, rvar factors can be matrices or arrays:
rvar_factor(x_array, dim = c(2, 2))

# If the input to rvar_factor() is an array with a `"levels"` attribute, it
# will use those as the levels of the factor
y_array <- t(array(rbinom(3000, 1, c(0.1, 0.5, 0.9)) + 1, dim = c(3, 1000)))
# with levels
attr(y_array, "levels") = c("a", "b")

posterior documentation built on July 4, 2024, 1:07 a.m.