Fast way to return matching elements as list-CodePudding

I have two large vectors:

a <- sample(1:1000000, 300000)

b <- sample(1:1000000, 40000)

I want to find the elements in a that match b:

a[(which(a %in% b))]

but instead of returning a vector I would like to have each matching element stored as a separate entry in a list, e.g.:

sapply(b, function(x) a[(which(a %in% x))])

this would to the job but it takes very long and the speed difference between the two is huge.

Is there a way to store the results in a list that is actually fast?

CodePudding user response：

If you are happy with NA instead of elements with lengths zero, you can coerce the vector as list and replace.

replace(as.list(a), !a %in% b, NA)

CodePudding user response：

You can try to use Map and set the length of the not matching to 0 or use [<- to replace the non matched.

a <- c(1,4,5)
b <- c(4,7)

Map(\(a,b) {`length<-`(a, b)}, b,  (b %in% a))
#[[1]]
#[1] 4
#
#[[2]]
#numeric(0)

`[<-`(as.list(b), !b %in% a, list(b[0]))
#replace(as.list(b), !b %in% a, list(b[0])) #Alternative
#[[1]]
#[1] 4
#
#[[2]]
#numeric(0)

sapply(b, function(x) a[(which(a %in% x))]) #Method form the question
#[[1]]
#[1] 4
#
#[[2]]
#numeric(0)