R函数在向量上使用FOR循环

Question

Surely a facepalm question, sorry. I have been trying to RTFM and google that but no luck.

I am trying to make a silly function in R that takes a vector as argument, loops through it with a for loop and does a very simple addition to each component.

func1 <- function(vector) {
  vr <- c() ### an empty vector 

  for (i in 1:length(vector)) { 
    vr <- vector[i]+i
  }
}

To me this looks correct, but I am always getting a NULL vector. Why?

Answer 1

You could write that much more efficiently like this:

func1 <- function(vector){
    vector + seq_along(vector)
}

This way you are taking advantage of vector addition.

For the sake of completeness, let's benchmark these two solutions. Here is the general loop solution:

func2 <- function(vector) {
    vr <- c() ### an empty vector 

    for (i in 1:length(vector)) { 
        vr[i] <- vector[i]+i
    }
    vr
}

As a step up, there is also the in-place solution

func3 <- function(vector) {    
    for (i in 1:length(vector)) { 
        vector[i] <- vector[i]+i
    }
    vector
}

Using microbenchmark we can see that the vectorized solution is by far the most efficient.

vec <- sample(100000, 10000)

library(microbenchmark)
microbenchmark(func1(vec), func3(vec), func2(vec))

Unit: microseconds
       expr       min        lq         mean    median       uq         max neval cld
 func1(vec)    29.998    36.984     44.78312    42.736    44.38     399.006   100  a 
 func3(vec) 12845.823 13666.432  14452.02863 14060.712 14708.53   25025.950   100  a 
 func2(vec) 84898.055 87354.750 110046.26659 88634.566 91193.38 1042819.269   100   b

Answer 2

Your code is close but has two problems. First, unless specified with a return() call, functions in R return the last expression. In your function this is the for loop itself. But the for loop expression returns nothing in and of itself.

Second, the last line doesn't index on vr , so you are overwriting it completely with the last result in the loop.

Thus, this code will (I think) do what you want. But agreed with the other answer that there are easier ways to do this.

func1 <- function(vector) {
  vr <- c() ### an empty vector 

  for (i in 1:length(vector)) { 
    vr[i] <- vector[i]+i
  }
  vr
}

This gives this output:

> func1(c(1,2))
[1] 2 4