Vectorization

Most of the ‘magic’ in R expressions is done by ‘vectorized’ operations and the vector recycling. You’ve already seen vectorized operations several times, like:

1:10 > 5

• 1:10 creates an interger vector of length 10 (check yourself)
• 5 is recycled 10 times to match the length of the longer vector
• > comparison is performed, returning logical vector of length 10

Try this:

1:10 + 1:2

Can you explain the result?

1:10 + rep(c(1, 2), times=5)

Now try this:

1:10 + 1:3

Vectorized functions

Many of the basic R functions are vectorized - they work the same for a single argument and for a vector of arguments.

sum(4)
sum(1:10)

mean(4)
mean(1:10)

Single value is usually treated a vector of length 1. Hence the [1] when it prints to console. It shows you the value at position 1 in the resulting vector. When you print enough to fill one row in your console, you’ll see the next index.

1:200

Because many standard operations are vectorized, the functions you write are vectorized without any additional work:

fib <- function(n) {
  ifelse(n > 1, 
         fib(n - 1) + fib(n - 2),
         n)
}

fib(5)

fib(1:10)

But what if you use something that does not like vector arguments - like grep (which means ‘search for string’ because of historical reasons).

count_in_vec <- function(what, where) {
  grep(what, where) %>% length
}

# counting various species of Streptococcus
count_in_vec("Streptococcus", dtaxons$taxon_name)

# now I'd like to search for both Streptococcus and Veillonella
count_in_vec(c("Veillonella", "Streptococcus"), dtaxons$taxon_name) 

The cure here is the sapply function, which calls a function for each element in a vector and return the results as a new vector.

add_one <- function(x) x + 1
sapply(1:5, add_one)

# you can pass additional arguments like this
add_num <- function(x, num) x + num
sapply(1:5, add_num, num = 3)

# or you don't have to name your function
sapply(1:5, function(x) x + 1)

Now we can vectorize our function easily:

count_in_vec <- function(what, where) {
  sapply(what, function(x) grep(x, where) %>% length)
}

# as an extra we get the names..
count_in_vec("Streptococcus", dtaxons$taxon_name)
count_in_vec(c("Veillonella", "Streptococcus"), dtaxons$taxon_name)