Benchmarks - Vectorized Recursive Subset Operations On Nested Lists

library(squarebrackets)
#> Run `?squarebrackets::squarebrackets_help` to open the introduction help page of 'squarebrackets'.

 

Introduction

Operating on recursive sub-sets of a nested lists requires the [[ and [[<- operators. These operators can only select a single subset at once. Thus, operating on multiple recursive sub-sets requires a for-loop, and is thus not vectorized.

The lst_untree() function unnests a tree-like nested list into a recursive matrix, in such a way that recursive sub-set operations can be performed in a vectorized manner.

Below are the benchmarks on operating on the original list, and the list transformed by lst_untree().

x <- lapply(1:1000, \(x) list(
  list(sample(letters), sample(LETTERS)),
  list(sample(month.abb), sample(month.name)),
  list(sample(1:100), rnorm(100))
))
y <- lst_untree(x, 1)

foo <- bench::mark(
  original = for(i in seq_along(x)) {x[[c(i, c(1,2))]]},
  "untree'd" = sb2_x(n(seq_len(nrow(y)), 2), 1:2),
  check = FALSE,
  min_iterations = 1000
)
summary(foo)
#> # A tibble: 2 × 6
#>   expression      min   median `itr/sec` mem_alloc `gc/sec`
#>   <bch:expr> <bch:tm> <bch:tm>     <dbl> <bch:byt>    <dbl>
#> 1 original     1.09ms   1.26ms      750.    38.4KB     16.9
#> 2 untree'd      7.8µs    9.2µs    98036.    86.9KB     19.6
ggplot2::autoplot(foo)
#> Loading required namespace: tidyr