Types

Struct for user callbacks.

Fields

• name::String : User's name for the callback. Defaults to string(h).
• h::Function : Callback function. See notes below for calling signature.
• initialize::Function : Function to initialize ext after subproblem creation has occurred.
• ext::Dict{Symbol,Any} : Dictionary to store data between callback calls or needed parameters.

The callback function h must have the signature h(ext::Dict{Symbol,Any}, phd::PHData, winf::WorkerInf, niter::Int)::Bool where ext is the same dictionary given to the Callback constructor, phd is the standard PH data structure (see PHData), winf is used for communicating with subproblems (see apply_to_subproblem) and niter is the current iteration. The callback may return false to stop PH.

The initialize function must have the signature initialize(ext::Dict{Symbol,Any}, phd::PHData) where ext is the same dictionary given to the Callback constructor and phd is the standard PH data structure (see PHData).

Struct for user supbroblem callbacks.

Fields

• name::String : User's name for the callback. Defaults to string(h).
• h::Function : Callback function. See notes below for calling signature.
• ext::Dict{Symbol,Any} : Dictionary to store data between callback calls or needed parameters.

The callback function h must have the signature h(ext::Dict{Symbol,Any}, sp::T, niter::Int, scenario_id::ScenarioID) where T <: AbstractSubproblem where ext is the same dictionary given to the Callback constructor, sp is a concrete type of AbstractSubproblem (see AbstractSubproblem), niter is the current iteration and scenario_id is a scenario identifier (see ScenarioID).

Exception indicating that the specified method is not implemented for an interface.

Unique, type-safe identifier for variables associated with the same scenario tree node.

Unique identifier for a ScenarioNode in a ScenarioTree.

Unique, type-safe identifier for a scenario.

Unique, type-safe identifier for a stage.

Unique, type-safe identifier for any variable in a (multi-stage) stochastic programming problem. Composed of a ScenarioID, a StageID and an Index.

Unique identifier for consensus variables. The scenario variables being driven to consensus with this variable is given by convert_to_variable_ids.

Variable dependent penalty parameter given by k * c_i where c_i is the linear coefficient of variable i in the objective function. If c_i == 0 (that is, the variable has no linear coefficient in the objective function), then the penalty value is taken to be k.

Requires subproblem type to have implemented report_penalty_info for this type. This implementation should return the linear coefficient in the objective function for each variable.

Constant scalar penalty parameter.

Penalty parameter set with Watson-Woodruff SEP method. See (Watson and Woodruff 2011) for more details.

Requires subproblem type to have implemented report_penalty_info for this type. This implementation should return the linear coefficient in the objective function for each variable.

Implementation of the AbstractSubproblem interface using JuMP.

Type representing a consensus variable.

The following functions are available to the user to interact with consensus variables

• is_integer
• value
• variables

Data structure used to store information and results for a stochastic programming problem.

See the following functions make use of this object:

• apply_to_subproblem
• branch_value
• consensus_variables
• convert_to_variable_ids
• convert_to_xhat_id
• get_callback
• get_callback_ext
• is_leaf
• name
• probability
• scenario_bundle
• scenarios
• stage_id
• value
• w_value
• xhat_value

The following post solution functions are also available:

• leaf_value
• retrieve_soln
• retrieve_aug_obj_value
• retrieve_obj_value
• retrieve_no_hats
• retrieve_w

If the corresponding save options are enabled, the saved terms may be accessed with one of the following:

• lower_bounds
• residuals
• retrieve_xhat_history
• retrieve_no_hat_history
• retrieve_w_history

Struct representing a node in a scenario tree.

Struct representing the scenario structure of a stochastic program.

Can be built up by the user using the functions add_node and add_leaf.

Constructor

ScenarioTree()

Default constructor generates the root node of the tree. Can get the root node with root.