Package 'simmer.bricks'

simmer.bricks: Helper Methods for simmer Trajectories

Description

Provides wrappers for common activity patterns in simmer trajectories.

Author(s)

Iñaki Ucar

See Also

simmer's homepage https://r-simmer.org and GitHub repository https://github.com/r-simmer/simmer.bricks.

---

Delayed Release of a Resource

Description

This brick encapsulates a delayed release: the arrival releases the resource and continues its way immediately, but the resource is busy for an additional period of time.

Usage

delayed_release(
  .trj,
  resource,
  task,
  amount = 1,
  preemptive = FALSE,
  mon_all = FALSE
)

delayed_release_selected(
  .trj,
  task,
  amount = 1,
  preemptive = FALSE,
  mon_all = FALSE
)

Arguments

.trj	the trajectory object.
resource	the name of the resource.
task	the timeout duration supplied by either passing a numeric or a callable object (a function) which must return a numeric (negative values are automatically coerced to positive).
amount	the amount to seize/release, accepts either a numeric or a callable object (a function) which must return a numeric.
preemptive	whether arrivals in the server can be preempted or not based on seize priorities.
mon_all	if TRUE, get_mon_arrivals will show one line per clone.

Value

Returns the following chain of activities: clone > synchronize (see examples below).

Examples

## These are equivalent for a non-preemptive resource:
trajectory() %>%
  delayed_release("res1", 5, 1)

trajectory() %>%
  clone(
    2,
    trajectory() %>%
      set_capacity("res1", -1, mod="+") %>%
      release("res1", 1),
    trajectory() %>%
      timeout(5) %>%
      set_capacity("res1", 1, mod="+")
  ) %>%
  synchronize(wait=FALSE)

## These are equivalent for a preemptive resource:
trajectory() %>%
  delayed_release("res2", 5, 1, preemptive=TRUE)

trajectory() %>%
  clone(
    2,
    trajectory() %>%
      release("res2", 1),
    trajectory() %>%
      set_prioritization(c(rep(.Machine$integer.max, 2), 0)) %>%
      seize("res2", 1) %>%
      timeout(5) %>%
      release("res2", 1)
  ) %>%
  synchronize(wait=FALSE)

---

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)

---

Interleaved Resources

Description

This brick encapsulates a chain of interleaved resources, i.e., the current resource is not released until the next one in the chain is available. An interesting property of such a pattern is that, if one resource is blocked for some reason, the whole chain stops.

Usage

interleave(.trj, resources, task, amount = 1)

Arguments

.trj	the trajectory object.
resources	character vector of resource names.
task	the timeout duration supplied by either passing a numeric or a callable object (a function) which must return a numeric (negative values are automatically coerced to positive).
amount	the amount to seize/release, accepts either a numeric or a callable object (a function) which must return a numeric.

Details

Both task and amount accept a list of values/functions, instead of a single one, that should be of the same length as resources, so that each value/function is applied to the resource of the same index.

The transition to the second and subsequent resources is guarded by a token, an auxiliary resource whose capacity must be equal to the capacity + queue size of the guarded resource, and its queue size must be infinite. For example, if two resources are provided, c("A", "B"), the auxiliary resource will be named "B_token". If capacity=2 and queue_size=1 for B, then capacity=3 and queue_size=Inf must be the values for B_token. But note that the user is responsible for adding such an auxiliary resource to the simulation environment with the appropriate parameters.

Value

Returns the following chain of activities: seize (1) > timeout > [seize (token to 2) > release (1) > seize (2) > timeout > release (2) > release (token to 2) > ... (repeat) ] (see examples below). Thus, the total number of activities appended is length(resources) * 3 + (length(resources)-1) * 2.

Examples

## These are equivalent:
trajectory() %>%
  interleave(c("A", "B"), c(2, 10), 1)

trajectory() %>%
  seize("A", 1) %>%
  timeout(2) %>%
  seize("B_token", 1) %>%
  release("A", 1) %>%
  seize("B", 1) %>%
  timeout(10) %>%
  release("B", 1) %>%
  release("B_token", 1)

---

Visit a Resource

Description

These bricks encapsulate a resource visit: seize, spend some time and release.

Usage

visit(.trj, resource, task, amount = 1)

visit_selected(.trj, task, amount = 1, id = 0)

Arguments

.trj	the trajectory object.
resource	the name of the resource.
task	the timeout duration supplied by either passing a numeric or a callable object (a function) which must return a numeric (negative values are automatically coerced to positive).
amount	the amount to seize/release, accepts either a numeric or a callable object (a function) which must return a numeric.
id	selection identifier for nested usage.

Value

Returns the following chain of activities: seize > timeout > release (see examples below).

Examples

## These are equivalent:
trajectory() %>%
  visit("res", 5, 1)

trajectory() %>%
  seize("res", 1) %>%
  timeout(5) %>%
  release("res", 1)

## These are equivalent:
trajectory() %>%
  visit_selected(5, 1)

trajectory() %>%
  seize_selected(1) %>%
  timeout(5) %>%
  release_selected(1)

---

Wait a Number of Signals

Description

These bricks encapsulate n stops: wait for a sequence of n signals. wait_until also traps and untraps the required signals.

Usage

wait_n(.trj, n = 1)

wait_until(.trj, signals, n = 1)

Arguments

.trj	the trajectory object.
n	number of wait activities to chain.
signals	signal or list of signals, accepts either a string, a list of strings or a callable object (a function) which must return a string or a list of strings.

Value

wait_n returns n times wait. wait_until also adds trap and untrap at the beginning and end, respectively, of the chain of waits (see examples below).

Examples

## These are equivalent:
trajectory() %>%
  wait_n(3)

trajectory() %>%
  wait() %>%
  wait() %>%
  wait()

## These are equivalent:
trajectory() %>%
  wait_until("green")

trajectory() %>%
  trap("green") %>%
  wait() %>%
  untrap("green")

## These are equivalent:
trajectory() %>%
  wait_until(c("one", "another"), 2)

trajectory() %>%
  trap(c("one", "another")) %>%
  wait() %>%
  wait() %>%
  untrap(c("one", "another"))

---