set_seed: A 'set.seed'-wrapper to simplify reproducible coding.
In LAJordanger/leanRcoding: Leaner R Coding

Description Usage Arguments Details Value Examples

A set.seed-wrapper to simplify reproducible coding.

1
2
3

set_seed(seed, expr = NULL, envir = parent.frame(), kind = NULL,
  normal.kind = NULL, vstr = NULL, create_kind_vstr_list = FALSE,
  create_code = FALSE)

`seed`	The integer to be used as the seed value. This argument will be given to the ordinary `set.seed`-function.
`expr`	An expression, default `NULL`, to be evaluated after the seed has been set. The evaluation will be within the frame given by the three arguments `kind`, `normal.kind` and `vstr`, and thus gives an extra level of protection for those cases where the internal workings of an expression use `set.seed` with a fixed value for the seed. Note that the seed and the evaluation will be performed at the level of the calling function.
`envir`	The environment in which `expr` is to be evaluated, default value `parent.frame()`. See the documentation of `eval` for further details.
`kind`	Character or `NULL`. If `kind` is a character string, set R's RNG to the kind desired.
`normal.kind`	Character string or `NULL`. If it is a character string, set the method of Normal generation.
`vstr`	A character string containing a version number of R. The default value `NULL` will use the present version number. Note that the function `RNGversion` works as a wrapper that calls `RNGkind` based on the version number, and that it can be conflicts between the settings specified by `vstr` and the setting specified by `kind` and `normal.kind`. The settings indicated by `vstr` will be used if `kind` and `normal.kind` both are `NULL`, but if one or both of them are given, then settings will be based upon them.
`create_kind_vstr_list`	Logical value, default `FALSE`. This can be used to create a list with the values for `kind`, `normal.kind` and `vstr`. If this is used when the before mentioned arguments are `NULL`, then the present values from the active R-session will be stored. This can be used by functions that need to store detailed information in an updated call, to ensure reproducibility later on. This argument will be ignored if `create_code` is `TRUE`.
`create_code`	Logical value, default `FALSE`. This can be used to create a chunk of code to be inserted into e.g. a script. The idea is that this should be done when the script is created, in order to ensure that the present settings of the system is properly recorded for reproducible results if the script is to be used later on by other machines, versions of R.

This function wraps around the function set.seed, RNGversion-function such that all the details related to kind, normal.kind and vstr are fixed by this single function. Moreover, it's also possible to return the code for calling this function (with the default values used) in order to ensure that computational scripts properly reflect the settings under which they initially were used. Finally, it's also possible to use this function to create a list of the present settings, such that these can be integrated into call of functions that are stored in order to ensure reproducibility. See the help-page of Random for details about the three arguments kind, normal.kind and vstr.

The result depends on the value given to create_code. It will either be a setting of the seed, or it will be the code needed to set the seed. NOTE: The function will ensure that the settings for kind and normal.kind will be reverted to the original ones. The value of the .Random.seed-vector will also depend upon this. If kind and normal.kind was different from the existing values, the value of .Random.seed will be reverted back to what it was when the function was called. Otherwise, the Random.seed-vector will be similar to what it would have been if set.seed and the expression expr had been evaluated in the workspace.

##  Gives the same effect as 'set.seed'.
set.seed(seed = 1)
rnorm(3)
.a <- .Random.seed
set_seed(seed = 1)
rnorm(3)
.b <- .Random.seed
identical(.a, .b)
## The 'expr'-argument is evaluated in the global environment,
## unless otherwise specified by the 'envir'-argument.
set_seed(seed = 1,
         expr = {.tmp <- rnorm(3)})
.tmp
.c <- .Random.seed
identical(.a, .c)

## No footprint on workspace when a change of RNG has been used.
set.seed(NULL)
.a <- .Random.seed
set_seed(seed =1,
         expr = {.tmp <- rnorm(3)},
         vstr = "1.0.0")
.tmp
.b <- .Random.seed
identical(.a, .b)

##  No footprint on workspace when a list is produced.
set.seed(NULL)
.a <- .Random.seed
set_seed(seed =1,
         expr = {.tmp <- rnorm(3)},
         vstr = "1.0.0",
         create_kind_vstr_list = TRUE)
.tmp
.b <- .Random.seed
identical(.a, .b)

## No footprint on workspace when code is produced.
set.seed(NULL)
.a <- .Random.seed
.seed <- 1
set_seed(seed = .seed,
         expr = {.tmp__ <- rnorm(3)},
         vstr = "1.0.0",
         create_code = TRUE)
.b <- .Random.seed
identical(.a, .b)
## Note: The creation of the code does not evaluate
## the 'expr'-argument.