如何在R中基于system.time用户时间绘制三个函数的运行时间曲线？

Got it, let's walk through how to do this step by step in R. Here's a complete, practical approach that'll get you those runtime comparison curves:

Step 1: Prep Your Functions #

First, make sure your helper functions a(), b(), and c() are properly defined (you didn't include them in your question, so I'll use placeholder examples here):

# Example helper functions (replace with your actual ones)
a <- function(x) x^2
b <- function(x) sqrt(x)
c <- function(x) lgamma(x)

# Your original functions
f1 <- function(n) { 
  sum = 0 
  for(i in 1:n) { sum <- sum + a(i) } 
  return(sum) 
}

f4 <- function(n) { 
  sum = 0 
  for(i in 1:n) { sum <- sum + b(i) } 
  return(sum) 
}

f3 <- function(n) { 
  # Done using lgamma 
  sum = 0 
  for(i in 1:n) { sum = sum + c(i) } 
  return(sum) 
}

Step 2: Define Test Parameters #

Choose a range of n values to test, and decide how many times to replicate each run (to smooth out random timing noise):

# Range of n values to test (adjust based on your function speed)
n_values <- seq(1, 1000, by = 50)

# Number of replications per n to get reliable average times
n_reps <- 10

Step 3: Measure Runtime for Each Function #

Loop through each n value, measure the user CPU time (using system.time's user.self component) for each function, and store the averages:

# Initialize vectors to store average runtime
time_f1 <- numeric(length(n_values))
time_f4 <- numeric(length(n_values))
time_f3 <- numeric(length(n_values))

# Run timing loop
for (i in seq_along(n_values)) {
  current_n <- n_values[i]
  
  # Measure f1's average user time
  time_f1[i] <- mean(replicate(n_reps, system.time(f1(current_n))["user.self"]))
  
  # Measure f4's average user time
  time_f4[i] <- mean(replicate(n_reps, system.time(f4(current_n))["user.self"]))
  
  # Measure f3's average user time
  time_f3[i] <- mean(replicate(n_reps, system.time(f3(current_n))["user.self"]))
}

Step 4: Create a Plotable Data Frame #

Combine the results into a tidy data frame for easier plotting:

runtime_data <- data.frame(
  n = rep(n_values, 3),
  avg_user_time = c(time_f1, time_f4, time_f3),
  function = rep(c("f1", "f4", "f3"), each = length(n_values))
)

Step 5: Plot the Runtime Curves #

You can use either base R or ggplot2 for plotting. Here are both options:

Option 1: Using ggplot2 (Cleaner, More Customizable)

# Install ggplot2 if you haven't already
# install.packages("ggplot2")
library(ggplot2)

ggplot(runtime_data, aes(x = n, y = avg_user_time, color = function)) +
  geom_line(linewidth = 1.2) +
  geom_point(size = 2) +
  labs(
    title = "Function Runtime Comparison",
    x = "Input Size (n)",
    y = "Average User CPU Time (seconds)",
    color = "Function"
  ) +
  theme_minimal()

Option 2: Using Base R (No Extra Packages Needed)

plot(n_values, time_f1, type = "l", col = "red", lwd = 2,
     xlab = "Input Size (n)", ylab = "Average User CPU Time (seconds)",
     main = "Function Runtime Comparison")
lines(n_values, time_f4, col = "blue", lwd = 2)
lines(n_values, time_f3, col = "green", lwd = 2)
legend("topleft", legend = c("f1", "f4", "f3"),
       col = c("red", "blue", "green"), lwd = 2)

Quick Notes #

• Adjust n_values and n_reps based on your functions: if your functions are very fast, use larger n (e.g., 10,000) and more reps (e.g., 100) to get measurable times. If they're slow, stick to smaller n and fewer reps.
• The user.self value from system.time isolates the CPU time used directly by your function, ignoring system overhead or waiting time.
• If your functions have side effects (like modifying global variables), reset their state between replications to avoid skewed results.

内容的提问来源于stack exchange，提问作者Suhail Gupta