incl/future.R

In HenrikBengtsson/future: Unified Parallel and Distributed Processing in R for Everyone

## Evaluate futures in parallel
plan(multisession)

## Data
x <- rnorm(100)
y <- 2 * x + 0.2 + rnorm(100)
w <- 1 + x ^ 2


## EXAMPLE: Regular assignments (evaluated sequentially)
fitA <- lm(y ~ x, weights = w)      ## with offset
fitB <- lm(y ~ x - 1, weights = w)  ## without offset
fitC <- {
  w <- 1 + abs(x)  ## Different weights
  lm(y ~ x, weights = w)
}
print(fitA)
print(fitB)
print(fitC)


## EXAMPLE: Future assignments (evaluated in parallel)
fitA %<-% lm(y ~ x, weights = w)      ## with offset
fitB %<-% lm(y ~ x - 1, weights = w)  ## without offset
fitC %<-% {
  w <- 1 + abs(x)
  lm(y ~ x, weights = w)
}
print(fitA)
print(fitB)
print(fitC)


## EXAMPLE: Explicitly create futures (evaluated in parallel)
## and retrieve their values
fA <- future( lm(y ~ x, weights = w) )
fB <- future( lm(y ~ x - 1, weights = w) )
fC <- future({
  w <- 1 + abs(x)
  lm(y ~ x, weights = w)
})
fitA <- value(fA)
fitB <- value(fB)
fitC <- value(fC)
print(fitA)
print(fitB)
print(fitC)

\dontshow{
## Make sure to "close" an multisession workers on Windows
plan(sequential)
}