How to run a function multiple times and write the results to a list?

问题

I have a function that creates a matrix and I want to call this function a thousand times such that in the end I have a list of 1000 matrices. Here is an example:

set.seed(1)
gen_mat <- function(x) matrix(c(1, 1, 1, x + rnorm(1)), nrow = 2)

Now, I tried replicate(10, gen_mat(1)), but this returns an array and not a list. How to do it?

回答1:

Combination of above answer, comment, and my own answer. Naturally, I like mine better. Also, I think there is a mistake in the above answer for base R.

n <- 10

# give 1 to gen_mat n-times
lapply(rep(1, n), gen_mat)

# replicate n-times 
replicate(n, gen_mat(1), simplify=FALSE)

# lapply returns error if FUN is not function or 
# the function is not taking an argument. Hence a dummy function.
lapply(seq_len(n), function(x) gen_mat(1))

microbenchmarking the three methods

I used a larger value for n, but the results are similar in my desktop with smaller n as well. For this, replicate takes longer than the other two methods.

set.seed(1)
gen_mat <- function(x) matrix(c(1, 1, 1, x + rnorm(1)), nrow = 2)
n <- 1000 

library(microbenchmark)
library(ggplot2)

mb <- microbenchmark(
  lap1 = {lapply(rep(1, n), gen_mat)},
  repl = {replicate(n, gen_mat(1), simplify=FALSE)},
  lap2 = {lapply(seq_len(n), function(x) gen_mat(1))},
  times = 10000L
)

mb
# Unit: milliseconds
# expr      min       lq     mean   median       uq      max neval cld
# lap1 2.839435 3.494157 4.806954 3.854269 5.611413 103.0111 10000  a 
# repl 3.091829 3.777199 5.140789 4.165856 6.013591 101.4318 10000   b
# lap2 3.131491 3.761274 5.089170 4.140316 5.939075 101.1983 10000   b

autoplot(mb)

回答2:

base R

n <- 10
lapply(seq_len(n), gen_mat(1))

purrr package

library(purrr)
map(seq_len(n), ~gen_mat(1))

来源：https://stackoverflow.com/questions/46104176/how-to-run-a-function-multiple-times-and-write-the-results-to-a-list

标签

list

matrix

replicate