Internals

Objective{F,V <: AbstractVector}

A wrapped objective function for solving a CDCProblem. This facilitates maintaining an internal interface for dealing with objective functions that is independent from user interface.

Users are not required to construct Objective unless there is a need for fine-grained control.

Fields

• f::F: objective function
• ℒ::V: the decision vector at which to evaluate the objective function
• fcall::Int = 0: the number of times the objective has been called, used to track whether the maximum number of objective calls has been reached (see also CDCProblem)

margin(obj::Objective, ℓ::Int, z)

Evaluate the change in obj with optional parameter z when the ℓth item is included or not. This corresponds to the D_j function in earlier implementation.

CDCPSolver

Abstract type for all solution algorithms for a CDCProblem.

See also Naive, Squeezing, SqueezingPolicy.

Naive{Z} <: CDCPSolver

A type for solving a CDCProblem with exhaustion.

Squeezing{V <: AbstractVector,Z} <: CDCPSolver

A type for solving a CDCProblem with single-agent squeezing.

See also solve.

Fields

• scdca::Bool: does the problem obey single crossing differences in choices from above? (if false, then from below is assumed)
• branching::Vector{V}: container holding branching outcomes
• z: the parameter (e.g. productivity) of the problem
• branch::Bool = true: whether or not to branch

SqueezingPolicy{Z,V<:AbstractVector{ItemState},F1,F2,S,O} <: CDCPSolver

A type for solving a CDCProblem with policy function squeezing.

See also solve.

Fields

• scdca::Bool
• intervalchoices::Vector{IntervalChoice{Z,V}}
• squeezing_indices::Vector{Int}
• branching_indices::Vector{Int}
• zero_margin::F1
• equal_obj::F2
• singlecdcp::S
• obj2::O
• skiprefinement::Bool
• maxfcall::Int