Random seed

with_seed() runs code with a specific random seed and resets it afterwards.

with_preserve_seed() runs code with the current random seed and resets it afterwards.

with_seed(
  seed,
  code,
  .rng_kind = "default",
  .rng_normal_kind = "default",
  .rng_sample_kind = "default"
)

local_seed(
  seed,
  .local_envir = parent.frame(),
  .rng_kind = "default",
  .rng_normal_kind = "default",
  .rng_sample_kind = "default"
)

with_preserve_seed(code)

local_preserve_seed(.local_envir = parent.frame())

Arguments

seed	[integer(1)] The random seed to use to evaluate the code.
code	[any] Code to execute in the temporary environment
.rng_kind	[character(1)] Kind of (uniform) RNG to use.
.rng_normal_kind	[character(1)] Kind of normal RNG to use.
.rng_sample_kind	[character(1)] Kind of RNG to use for sampling.
.local_envir	[environment] The environment to use for scoping.

Value

[any]
The results of the evaluation of the code argument.

See also

withr for examples

Examples

# Same random values:
with_preserve_seed(runif(5))
#> [1] 0.10169262 0.76999328 0.08906143 0.53659237 0.26213892
with_preserve_seed(runif(5))
#> [1] 0.10169262 0.76999328 0.08906143 0.53659237 0.26213892

# Use a pseudorandom value as seed to advance the RNG and pick a different
# value for the next call:
with_seed(seed <- sample.int(.Machine$integer.max, 1L), runif(5))
#> [1] 0.5885083 0.3289879 0.4357050 0.2184870 0.3697503
with_seed(seed, runif(5))
#> [1] 0.5885083 0.3289879 0.4357050 0.2184870 0.3697503
with_seed(seed <- sample.int(.Machine$integer.max, 1L), runif(5))
#> [1] 0.6099326 0.6076366 0.4936592 0.6276273 0.5535448