Solver interface

AbstractStructuredOptimizer

Abstract supertype for structure-exploiting optimizers.

AbstractSampledOptimizer

Abstract supertype for sample-based optimizers.

optimize!(optimizer::AbstractStructuredOptimizer)

Start the solution procedure for optimizer after a call to load_structure!.

See also: load_structure!

check_loadable(optimizer::AbstractStructuredOptimizer, structure::AbstractStochasticStructure)

Throws an UnloadableStructure exception if structure is not loadable by optimizer.

See also: [`load_structure!`](@ref)

load_model!(optimizer::AbstractSampledOptimizer, model::StochasticModel, x₀::AbstractVector)

Instantiate the optimizer with the stochastic model and initial decision x₀.

See also: optimize!

load_structure!(optimizer::AbstractStructuredOptimizer, structure::AbstractStochasticStructure, x₀::AbstractVector)

Instantiate the optimizer with the stochastic program represented in memory by the given structure and initial decision x₀.

See also: optimize!

master_optimizer(optimizer::AbstractStructuredOptimizer)

Return a MOI optimizer constructor

num_iterations(optimizer::AbstractStructuredOptimizer)

Return the number of iterations ran by the structured optimizer

See also: optimize!

optimal_instance(optimizer::AbstractSampledOptimizer)

Return a stochastic programming instance of the stochastic model after a call to optimize!.

optimizer_name(optimizer::AbstractSampledOptimizer)

Optionally, return a string identifier of AbstractSampledOptimizer.

optimizer_name(optimizer::AbstractStructuredOptimizer)

Optionally, return a string identifier of AbstractStructuredOptimizer.

restore_structure!(optimizer::AbstractStructuredOptimizer)

Restore the stochastic program to the state it was in before a call to optimize!

See also: load_structure!

subproblem_optimizer(optimizer::AbstractStructuredOptimizer)

Return a MOI optimizer constructor for solving subproblems

supports_structure(optimizer::StochasticProgramOptimizerType, structure::AbstractStochasticStructure)

Return a Bool indicating whether optimizer supports the stochastic structure. That is, load_structure!(optimizer, structure) will not throw UnsupportedStructure

AbstractStructuredOptimizerAttribute

Abstract supertype for attribute objects that can be used to set or get attributes (properties) of the structure-exploiting optimizer.

AbstractSampledOptimizerAttribute

Abstract supertype for attribute objects that can be used to set or get attributes (properties) of the sample-based optimizer.

Confidence

An optimizer attribute for specifying the confidence level of sample-based procedures.

Execution

An optimizer attribute for specifying an execution policy for a structure-exploiting algorithm. Options are:

• Serial: Classical L-shaped (default)
• Synchronous: Classical L-shaped run in parallel
• Asynchronous: Asynchronous L-shaped ?Asynchronous for parameter descriptions.

ExecutionParameter

Abstract supertype for execution-specific attributes.

InstanceCrash

An optimizer attribute for specifying a crash start to be used when solving sampled instances.

InstanceOptimizer

An optimizer attribute for specifying the AbstractStructuredOptimizer/AbstractOptimizer used to solve sampled problems in sample-based procedures.

InstanceOptimizerAttribute

An optimizer attribute used for setting attributes of the instance optimizer.

MasterOptimizer

An optimizer attribute for specifying the MathOptInterface optimizer used to solve master problems arising in a structure-exploiting procedure.

MasterOptimizerAttribute

An optimizer attribute used for setting attributes of the master optimizer.

NumEEVSamples

An optimizer attribute for specifying the sample size of expected-value evaluation instances in sample-based procedures.

NumEWSSamples

An optimizer attribute for specifying the sample size of wait-and-see instances in sample-based procedures.

NumEvalSamples

An optimizer attribute for specifying the sample size of evaluation instances of sample-based procedures.

NumLowerTrials

An optimizer attribute for specifying the number of trials used to compute lower bounds of confidence intervals in sample-based procedures.

NumSamples

An optimizer attribute for specifying the sample size of sample-based procedures.

NumUpperTrials

An optimizer attribute for specifying the number of trials used to compute upper bounds of confidence intervals in sample-based procedures.

RawInstanceOptimizerParameter

An optimizer attribute used for raw parameters of the instance optimizer. Defers to RawOptimizerAttribute.

RawMasterOptimizerParameter

An optimizer attribute used for raw parameters of the master optimizer. Defers to RawOptimizerAttribute.

RawSubProblemOptimizerParameter

An optimizer attribute used for raw parameters of the subproblem optimizer. Defers to RawOptimizerAttribute.

RelativeTolerance

An optimizer attribute for specifying the relative tolerance of a structure-exploiting algorithm.

SubProblemOptimizer

An optimizer attribute for specifying the MathOptInterface optimizer used to solve subproblems arising in a structure-exploiting procedure.

get_instanceoptimizer_attribute(stochasticmodel::StochasticModel, attr::MOI.AbstractOptimizerAttribute)

Return the value of the solver-specific attribute attr of the instance optimizer in stochasticmodel.

See also: set_instanceoptimizer_attribute, set_instanceoptimizer_attributes.

get_instanceoptimizer_attribute(stochasticprogram::StochasticProgram, name::String)

Return the value associated with the solver-specific attribute named name of the instance optimizer in stochasticprogram.

See also: set_instanceoptimizer_attribute, set_instanceoptimizer_attributes.

get_masteroptimizer_attribute(stochasticprogram::StochasticProgram, attr::MOI.AbstractOptimizerAttribute)

Return the value of the solver-specific attribute attr of the master optimizer in stochasticprogram.

See also: set_masteroptimizer_attribute, set_masteroptimizer_attributes.

get_masteroptimizer_attribute(stochasticprogram::StochasticProgram, name::String)

Return the value associated with the solver-specific attribute named name of the master optimizer in stochasticprogram.

See also: set_masteroptimizer_attribute, set_masteroptimizer_attributes.

get_suboptimizer_attribute(stochasticprogram::StochasticProgram, attr::MOI.AbstractOptimizerAttribute)

Return the value of the solver-specific attribute attr of the subproblem optimizer in stochasticprogram.

See also: set_suboptimizer_attribute, set_suboptimizer_attributes.

get_suboptimizer_attribute(stochasticprogram::StochasticProgram, name::String)

Return the value associated with the solver-specific attribute named name of the subproblem optimizer in stochasticprogram.

See also: set_suboptimizer_attribute, set_suboptimizer_attributes.

set_instanceoptimizer_attribute(stochasticmodel::StochasticModel, attr::MOI.AbstractOptimizerAttribute, value)

Sets solver-specific attribute attr of the instance optimizer to value in stochasticmodel.

set_instanceoptimizer_attribute(stochasticmodel::StochasticModel, name::Union{Symbol, String}, value)

Sets solver-specific attribute of the instance optimizer identified by name to value in stochasticmodel.

set_instanceoptimizer_attributes(stochasticmodel::StochasticModel, pairs::Pair...)

Given a list of attribute => value pairs or a collection of keyword arguments, calls set_instanceoptimizer_attribute(stochasticmodel, attribute, value) for each pair.

set_masteroptimizer_attribute(stochasticprogram::StochasticProgram, attr::MOI.AbstractOptimizerAttribute, value)

Sets solver-specific attribute attr of the master optimizer to value.

set_masteroptimizer_attribute(stochasticprogram::StochasticProgram, name::Union{Symbol, String}, value)

Sets solver-specific attribute of the master optimizer identified by name to value.

set_masteroptimizer_attributes(stochasticprogram::StochasticProgram, pairs::Pair...)

Given a list of attribute => value pairs or a collection of keyword arguments, calls set_masteroptimizer_attribute(stochasticprogram, attribute, value) for each pair.

set_suboptimizer_attribute(stochasticprogram::StochasticProgram, attr::MOI.AbstractOptimizerAttribute, value)

Sets solver-specific attribute attr of the subproblem optimizer to value.

set_suboptimizer_attribute(stochasticprogram::StochasticProgram, name::Union{Symbol, String}, value)

Sets solver-specific attribute of the subproblem optimizer identified by name to value.

set_suboptimizer_attributes(stochasticprogram::StochasticProgram, pairs::Pair...)

Given a list of attribute => value pairs or a collection of keyword arguments, calls set_suboptimizer_attribute(stochasticprogram, attribute, value) for each pair.

Serial

Factory object for LShaped.SerialExecution/ProgressiveHedging.SerialExecution. Pass through the Execution attribute.

Synchronous

Factory object for LShaped.SynchronousExecution/ProgressiveHedging.SynchronousExecution. Pass through the Execution attribute.

Asynchronous

Factory object for LShaped.AsynchronousExecution/ProgressiveHedging.AsynchronousExecution. Pass through the Execution attribute.

...

Parameters

• max_active::Int = 3: Maximum number of active iterations that run asynchronously.
• κ::T = 0.5: Relative amount of finished subproblems required to start a new iterate. Governs the amount of asynchronicity.

...

Quadratic

Functor object for using a quadratic 2-norm penalty term. Requires an AbstractMathProgSolver capable of solving QP problems. Passed by default through penalty where applicable.

Linearized

Functor object for using an approximately quadratic penalty term, through linearization. Pass through penalty where applicable.

...

Parameters

• nbreakpoints::Int: Number of cutting planes used to approximate quadratic term

...

InfNorm

Functor object for using a linear ∞-norm penalty term. Pass through penalty where applicable.

ManhattanNorm

Functor object for using a linear 1-norm penalty term. Pass through penalty where applicable.