Do the R parallel extensions break the `apply` metaphor?

Question

Every time I see a question on parallel processing in R, it uses the foreach function. Since for loops are not very R-like, is there a parallel version of apply, and if so why isn't it more popular?

Answer 1

There are numerous parallel versions of *apply, starting with

• parLapply() in snow
• mclapply() in multicore
• mpi.apply() in Rmpi

as well as dedicated packages such as papply (possibly no longer maintained).

Answer 2

@Dirk is correct. I'd add that the plyr package now has support for a parallel backend.

In the case of the plyr package, it may be the case that little is mentioned because dropping in a parallel backend doesn't take any thought: it's just a flag.