How to use list-argument in microbenchmark

Question

How do one use the list-argument in the microbenchmark function. I want to microbenchmark the same function with different inputs as in

microbenchmark(j1 = {sample(1e5)},
               j2 = {sample(2e5)},
               j3 = {sample(3e5)})

The following is not going to fly, as the list will just contain vectors and not unevaluated expressions.

microbenchmark(list = list(j1 = {sample(1e5)},
                          j2 = {sample(2e5)},
                          j3 = {sample(3e5)))

I would also want to generate the list using e.g. lapply.

Answer 1

Just use alist:

microbenchmark(list = alist(a = Sys.sleep(0.005), b = Sys.sleep(0.01)))
#> Unit: milliseconds
#>  expr      min       lq      mean    median        uq       max neval cld
#>     a  5.02905  5.15946  5.447538  5.446029  5.612429  6.030764   100  a 
#>     b 10.04997 10.18264 10.431011 10.459569 10.547814 11.058911   100   b

alist handles its arguments as if they described function arguments. So the values are not evaluated, and tagged arguments with no value are allowed whereas list simply ignores them.

Answer 2

We need to use the substitute or bquote function to get the unevaluated expressions in the list, e.g.

microbenchmark(list = list(j1 = bquote({sample(1e5)}),
                           j2 = bquote({sample(2e5)}),
                           j3 = bquote({sample(3e5)})))

The jobs can be generated using lapply, but we have to be careful with environments

jobs = lapply(1000*1:3, function(s) local({s = s; bquote(sample(.(s)))}) )