apply function to igraph object

Question

I would like to learn how to code more efficiently using apply, sapply or lapply. While I understand the basics of it, I am yet lacking intuition. Specifically, I would like to use apply on a graph object. Is it possible to implement the example below with one of these functions and how could it be done? Any more efficient solution is welcome as the graph I am interested has more than 2 Mio. nodes, each having one edge.

# data
g <- make_graph(c(1, 2, 2, 3, 3, 4, 5, 6), directed = T)

# loop 
for(i in 1:length(V(g))){
  neigbours[[i]] <- adjacent_vertices(g, v= c(i), mode = "in")
}

score 2 · Accepted Answer · answered Mar 26 '20 at 12:08

2

lapply is not necessarily more efficient. But you can improve efficiency by creating the object neigbours first, instead of extending it in each for loop iteration, forcing calls to R's memory management routins.

neigbours <- vector("list", length = length(V(g)))
for(i in 1:length(V(g))){
  neigbours[[i]] <- adjacent_vertices(g, v= c(i), mode = "in")
} 

neib2 <- lapply(seq_along(V(g)), function(i){
  adjacent_vertices(g, v = i, mode = "in")
})
all.equal(neigbours, neib2)
# [1] TRUE

answered Mar 26 '20 at 12:08

Rui Barradas

70,273
8
34
66

I have been told, that for loops are less efficient than ```lapply``` and that I should consider learning ```lapply``` for this reason. – S Front Mar 26 '20 at 12:12
@SFront That statement is a [myth](https://stackoverflow.com/questions/42393658/lapply-vs-for-loop-performance-r). – Rui Barradas Mar 26 '20 at 12:14
also comparing system.time() shows that lapply is indeed faster – S Front Mar 26 '20 at 12:14
1

@SFront Yes, but in my test it's only 5% faster. – Rui Barradas Mar 26 '20 at 12:21

apply function to igraph object

1 Answers1