createISCSolver

fun createISCSolver(problemDefinition: ProblemDefinition, modelBuilder: ModelBuilderIfc, deltaC: Double = problemDefinition.indifferenceZoneParameter, deltaL: Double = deltaC, skipGlobalPhase: Boolean = false, globalBudget: Int? = null, startingPoint: MutableMap<String, Double>? = null, maxIterations: Int = ISCSolver.iscDefaultMaxIterations, replicationsPerEvaluation: Int = defaultReplicationsPerEvaluation, solutionCache: SolutionCacheIfc = MemorySolutionCache(), simulationRunCache: SimulationRunCacheIfc? = null, experimentRunParameters: ExperimentRunParametersIfc? = null, streamNum: Int = 0, streamProvider: RNStreamProviderIfc = RNStreamProvider(), name: String? = null, parallelOptions: ParallelEvaluationOptions = ParallelEvaluationOptions()): ISCSolver(source)

Creates and configures an Industrial Strength COMPASS (ISC) solver for a given problem definition. ISC is a three-phase discrete-optimization-via-simulation method: a global Niching-GA exploration phase, a local COMPASS phase per niche, and a clean-up ranking-and-selection phase.

The single indifference-zone parameter deltaC controls the statistical guarantees and follows ISC's graceful-degradation convention: with deltaC > 0 the clean-up phase runs the Rinott indifference-zone selection and reports a ±δ_C interval with a correct-selection guarantee (and, with deltaL > 0, COMPASS confirms local optimality via Kim's 2005 test); with deltaC = 0 (the default, inherited from the problem's indifference-zone parameter) the solver still returns a feasible best but the indifference-zone guarantees are dropped.

Return

An instance of ISCSolver.

Parameters

problemDefinition

The definition of the optimization problem, including constraints and objectives.

modelBuilder

The model builder interface used to create models for evaluation.

deltaC

The indifference-zone parameter δ_C. Defaults to the problem's indifference-zone parameter; 0.0 selects the degraded mode (no correct-selection guarantee), > 0 the full guarantees.

deltaL

The COMPASS local-optimality indifference zone δ_L. Defaults to deltaC.

skipGlobalPhase

When true, skip the global Niching-GA phase and run a single COMPASS search from the starting point (the paper's unimodal shortcut). Defaults to false.

globalBudget

An optional replication budget added as a soft transition rule to the global phase.

startingPoint

Optional initial coordinates seeded as ISC's first solution (and first COMPASS seed). If left null, the solver auto-generates a random feasible starting point upon initialization.

maxIterations

The maximum number of orchestration macro-steps. Defaults to 1000.

replicationsPerEvaluation

The number of replications to use during each evaluation.

solutionCache

Specifies if the evaluator uses a solution cache. By default, this is MemorySolutionCache.

simulationRunCache

Specifies if the simulation oracle will use a SimulationRunCache. The default is null (no cache).

experimentRunParameters

the run parameters to apply to the model during the building process

streamNum

the random number stream number; 0 (the default) means the next available stream

streamProvider

the provider of random number streams shared by the ISC sub-solvers; defaults to a fresh RNStreamProvider