do_parallel: Perform Parallel Tasks

In r-simmer/simmer.bricks: Helper Methods for 'simmer' Trajectories

Perform Parallel Tasks

Description

This brick encapsulates the activity of n workers running parallel sub-trajectories.

Usage

do_parallel(.trj, ..., .env, wait = TRUE, mon_all = FALSE)

Arguments

.trj	the trajectory object.
...	sub-trajectories or list of sub-trajectories to parallelise.
.env	the simulation environment.
wait	if TRUE, the arrival waits until all parallel sub-trajectories are finished; if FALSE, the arrival continues as soon as the first parallel task ends.
mon_all	if TRUE, get_mon_arrivals will show one line per clone.

Value

Returns the following chain of activities: clone > synchronize (> wait > untrap if wait=FALSE) (see examples below).

Examples

env <- simmer()
signal <- function() get_name(env)

task.1 <- trajectory("task 1") %>%
  timeout(function() rexp(1))
task.2 <- trajectory("task 2") %>%
  timeout(function() rexp(1))

## These are equivalent:
trajectory() %>%
  do_parallel(
    task.1,
    task.2,
    .env = env, wait = TRUE
  )

trajectory() %>%
  clone(
    n = 3,
    trajectory("original") %>%
      trap(signal) %>%
      wait() %>%
      wait() %>%
      untrap(signal),
    task.1[] %>%
      send(signal),
    task.2[] %>%
      send(signal)) %>%
  synchronize(wait = TRUE)

## These are equivalent:
trajectory() %>%
  do_parallel(
    task.1,
    task.2,
    .env = env, wait = FALSE
  )

trajectory() %>%
  clone(
    n = 3,
    trajectory("original") %>%
      trap(signal),
    task.1[] %>%
      send(signal),
    task.2[] %>%
      send(signal)) %>%
  synchronize(wait = FALSE) %>%
  wait() %>%
  untrap(signal)

r-simmer/simmer.bricks documentation built on July 19, 2023, 8:13 p.m.