Skip to contents

Benchmarking fastpeRmute

benchmark.Rmd

In this vignette, we benchmark the performance of fastpeRmute against the gtools package, which provide similar functionalities. For this, we utilize the microbenchmark package to measure execution times for various functions under different scenarios.

Combinations with Repetition 

mb_comb_with_rep <- microbenchmark(
  fastpeRmute = fastpeRmute:::combinationsWithRepetition(20, 5),
  gtools = gtools::combinations(n = 20, r = 5, repeats.allowed = TRUE),
  times = 100L
)

df_comb_with_rep <- as.data.frame(mb_comb_with_rep)

ggplot(df_comb_with_rep, aes(x = expr, y = time)) + 
  geom_boxplot() + 
  theme_minimal() + 
  labs(title = "Combinations with Repetition", y = "Time (nanoseconds)", x = "Function") + 
  coord_flip()

Combinations without Repetition 

mb_comb_without_rep <- microbenchmark(
  fastpeRmute = fastpeRmute:::combinationsWithoutRepetition(20, 5),
  combn = combn(x = 20, m = 5),
  gtools = gtools::combinations(n = 20, r = 5, repeats.allowed = FALSE),
  times = 100L
)

df_comb_without_rep <- as.data.frame(mb_comb_without_rep)

ggplot(df_comb_without_rep, aes(x = expr, y = time)) + 
  geom_boxplot() + 
  theme_minimal() + 
  labs(title = "Combinations without Repetition", y = "Time (nanoseconds)", x = "Function") + 
  coord_flip()

Permutations without Repetition 

mb_perm_without_rep <- microbenchmark(
  fastpeRmute = fastpeRmute:::permutationsWithoutRepetition(15,5),
  gtools = gtools::permutations(n = 15, r = 5, repeats.allowed = FALSE),
  times = 100L
)

df_perm_without_rep <- as.data.frame(mb_perm_without_rep)

ggplot(df_perm_without_rep, aes(x = expr, y = time)) + 
  geom_boxplot() + 
  theme_minimal() + 
  labs(title = "Permutations without Repetition", y = "Time (nanoseconds)", x = "Function") + 
  coord_flip()

Permutations with Repetition 

mb_perm_with_rep <- microbenchmark(
  fastpeRmute = fastpeRmute:::permutationsWithRepetition(15,5),
  gtools = gtools::permutations(n = 15, r = 5, repeats.allowed = TRUE),
  times = 100L
)

df_perm_with_rep <- as.data.frame(mb_perm_with_rep)

ggplot(df_perm_with_rep, aes(x = expr, y = time)) + 
  geom_boxplot() + 
  theme_minimal() + 
  labs(title = "Permutations with Repetition", y = "Time (nanoseconds)", x = "Function") + 
  coord_flip()

From the benchmark results, we can see the performance differences between fastpeRmute, gtools, and base-r functions when they exist.