bbo_benchmarking_optimizers: Optimizer for BBo_Benchmarking

Description Usage Arguments Details

Description

Optimizers ready for usage in bbob_custom or bbob_custom parallel

Usage

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
optimizerCMAES(dimension, instance, function_id, maxit, maxFE, stopFitness,
  path, debug.logging = FALSE, max_restarts, restart_multiplier,
  restart_triggers, OCD = FALSE, varLimit = NULL, nPreGen = NULL,
  maxGen = NULL, fitnessValue = FALSE, dispersion = FALSE,
  evolutionPath = FALSE)

optimizerCMAESWithoutDef(dimension, instance, function_id, maxit, maxFE,
  stopFitness, path, debug.logging = FALSE, max_restarts, restart_multiplier,
  restart_triggers, OCD = FALSE, varLimit = NULL, nPreGen = NULL,
  maxGen = NULL, fitnessValue = FALSE, dispersion = FALSE,
  evolutionPath = FALSE)

optimizerRS(dimension, instance, function_id, maxit, maxFE, stopFitness, path,
  OCD = FALSE, debug.logging = FALSE, max_restarts = 0,
  restart_multiplier = 1, restart_triggers = character(0), ...)

optimizerGA(dimension, instance, function_id, maxit, maxFE, stopFitness, path,
  debug.logging = FALSE, max_restarts, restart_multiplier, restart_triggers,
  OCD = FALSE, varLimit = NULL, nPreGen = NULL, maxGen = NULL,
  fitnessValue = FALSE, dispersion = FALSE, evolutionPath = FALSE)

Arguments

maxit

[integer]
If maxFE is not passed, the number of iterations, maxit, serves as an upper execution limit for optimizing one function.

maxFE

[interger]
maxFE is another upper execution limit for an optimizer, i.e. if the specified number of function evaluations is reached, the optimizer terminates the optimization.

stopFitness

[numeric]
If stopFitness is specified, the optimization is terminated if the gap between the current function value and the global optimum of the function is below the value of stopFitness.

debug.logging

[logical]
debug.logging ... Default is FALSE.

OCD

[logical]
OCD indicates if Online Convergence Detection (OCD) should be used as a stopping condition of the optimizer under test. OCD can only be activated if the optimizer has implemented the functionality necessary for OCD. See stopOnOCD for further necessary parameters. Default is FALSE.

varLimit

[logical]
varLimit: OCD Parameter (See stopOnOCD)

nPreGen

[logical]
nPreGen: OCD Parameter (See stopOnOCD)

maxGen

[logical]
maxGen: OCD Parameter (See stopOnOCD)

fitnessValue

[logical]
fitnessValue: OCD Parameter (See stopOnOCD)

dispersion

[logical]
disperion: OCD Parameter (See stopOnOCD)

evolutionPath

[logical]
evolutionPath: OCD Parameter (See stopOnOCD)

dimensions

[integer]
The dimensions parameter determines the dimensions to be passed to the optimizer under test. Possible values are between 2 and 40.

instances

[integer]
Every smoof_function can be instantiated in different ways. Could be a single intergerish value greater of equals 1.

function_ids

[integer(1)]
The function_ids are the unique identifier of the 24 noiseless bbob function.

max.restarts

[integer]
If max.restarts the optimizer restarts the optimization if one of the passed restart_triggers has fired. In order to use this feature, the optimizer under test must be able to restart the optimization. Default is FALSE.

restart.triggers

[string]
If restart.triggers are passed, the optimizer restarts the optimization if one of those triggers has fired. In order to use this feature, the optimizer under test must be able to restart the optimization. Default is FALSE.

Details

The optimizers are ready to be set up in a benchmarking experiment with bbob_custom or bbob_custom_parallel. Every input parameter is required as it passed by bbob_custom or bbob_custom_parallel to the particular optimizer.


andreas-he/cmaesbenchmarking documentation built on May 10, 2019, 10:30 a.m.