This class defines the common application programming interface (API) for the following classes that allow you to control the MIP search:

• IloCplex::SolveCallbackI
• IloCplex::UserCutCallbackI
• IloCplex::LazyConstraintCallbackI
• IloCplex::HeuristicCallbackI
• IloCplex::BranchCallbackI

An instance of one of these classes represents a user-written callback that intervenes in the search for a solution at a given node in an application that uses an instance of IloCplex to solve a mixed integer program (MIP). Control callbacks are tied to a node. They are called at each node during IloCplex branch-and-cut search. The user never subclasses the IloCplex::ControlCallbackI class directly; it defines only the common interface of those listed callbacks.

In particular, SolveCallbackI is called before solving the node relaxation and optionally allows substitution of its solution. IloCplex does this by default. Then the node relaxation is solved, either by an instance of SolveCallbackI or by IloCplex. If the solution is integer feasible or the relaxation is unbounded IloCplex::LazyConstraintCallbackI is invoked; otherwise, the other control callbacks are called in the following order:

1. IloCplex::UserCutCallbackI
2. IloCplex::HeuristicCallbackI
3. IloCplex::BranchCallbackI

If the cut callback added new cuts to the node relaxation, the node relaxation will be solved again using the solve callback, if used. The same is true if IloCplex generated its own cuts.

The methods of this class are protected and its constructor is private; you cannot directly subclass this class; you must derive from its subclasses.

If an attempt is made to access information not available to an instance of this class, an exception is thrown.

In the case where the node LP is unbounded, the methods getValue return a vector that corresponds to an unbounded direction. The vector is scaled in such a way that the maximum absolute value of one of its elements is CPX_INFBOUND. Thus, often the vector can be used directly, for example to separate lazy constraints. However, due to the large values, care must be taken to deal with potential numerical errors. If in doubt, rescale the vector, and use it as an unbounded ray rather than as a primal vector.

This method returns the current pseudo cost for branching downward on the variable var.

This method returns the current pseudo cost for branching downward on the variable var.

This method specifies whether each of the variables in the array vars is integer feasible, integer infeasible, or implied integer feasible by putting the status in the corresponding element of the array stats.

Before you query the feasibility of a solution from a solve callback, a solution must exist. That is, you must first create the solution by calling the CPLEX optimization method solve, and then you must verify that the method solve generated a solution by checking its return value before you invoke the method getFeasibilities.

This method specifies whether each of the variables in the array vars is integer feasible, integer infeasible, or implied integer feasible by putting the status in the corresponding element of the array stats.

Before you query the feasibility of a solution from a solve callback, a solution must exist. That is, you must first create the solution by calling the CPLEX optimization method solve, and then you must verify that the method solve generated a solution by checking its return value before you invoke the method getFeasibilities.

This method specifies whether the variable var is integer feasible, integer infeasible, or implied integer feasible in the current node solution.

Before you query the feasibility of a solution from a solve callback, a solution must exist. That is, you must first create the solution by calling the CPLEX optimization method solve, and then you must verify that the method solve generated a solution by checking its return value before you invoke the method getFeasibility.

This method specifies whether the variable var is integer feasible, integer infeasible, or implied integer feasible in the current node solution.

Before you query the feasibility of a solution from a solve callback, a solution must exist. That is, you must first create the solution by calling the CPLEX optimization method solve, and then you must verify that the method solve generated a solution by checking its return value before you invoke the method getFeasibility.

This method specifies whether the Special Ordered Set sos is integer feasible, integer infeasible, or implied integer feasible in the current node solution.

Before you query the feasibility of a solution from a solve callback, a solution must exist. That is, you must first create the solution by calling the CPLEX optimization method solve, and then you must verify that the method solve generated a solution by checking its return value before you invoke the method getFeasibility.

This method specifies whether the Special Ordered Set sos is integer feasible, integer infeasible, or implied integer feasible in the current node solution.

Before you query the feasibility of a solution from a solve callback, a solution must exist. That is, you must first create the solution by calling the CPLEX optimization method solve, and then you must verify that the method solve generated a solution by checking its return value before you invoke the method getFeasibility.

This method returns the lower bound of var at the current node. This bound is likely to be different from the bound in the original model because an instance of IloCplex tightens bounds when it branches from a node to its subnodes. The corresponding solution value from getValue may violate this bound at a node where a new incumbent has been found because the bound is tightened when an incumbent is found.

Unbounded Variables

If a variable lacks a lower bound, then getLB returns -IloInfinity (that is, the system-specific least value).

This method returns the lower bound of var at the current node. This bound is likely to be different from the bound in the original model because an instance of IloCplex tightens bounds when it branches from a node to its subnodes. The corresponding solution value from getValue may violate this bound at a node where a new incumbent has been found because the bound is tightened when an incumbent is found.

Unbounded Variables

If a variable lacks a lower bound, then getLB returns -IloInfinity (that is, the system-specific least value).

For each element of the array vars, this method puts the lower bound at the current node into the corresponding element of the array vals. These bounds are likely to be different from the bounds in the original model because an instance of IloCplex tightens bounds when it branches from a node to its subnodes. The corresponding solution values from getValues may violate these bounds at a node where a new incumbent has been found because the bounds are tightened when an incumbent is found.

Unbounded Variables

If a variable lacks a lower bound, then getLBs returns -IloInfinity (that is, the system-specific least value).

This method puts the lower bound at the current node of each element of the array vars into the corresponding element of the array vals. These bounds are likely to be different from the bounds in the original model because an instance of IloCplex tightens bounds when it branches from a node to its subnodes. The corresponding solution values from getValues may violate these bounds at a node where a new incumbent has been found because the bounds are tightened when an incumbent is found.

Unbounded Variables

If a variable lacks a lower bound, then getLBs returns -IloInfinity (that is, the system-specific least value).

This method retrieves the NodeData object that may have previously been assigned to the current node by the user with the method IloCplex::BranchCallbackI::makeBranch. If no data object has been assigned to the current node, the method returns 0 (zero).

Returns the NodeId of the current node.

See MIPCallbackI::NodeId.

This method returns the objective value of the solution at the current node.

If you need the object representing the objective itself, consider the method IloCplex::getObjective instead.

This method returns the slack value for the constraint specified by rng in the solution of the relaxation at the current node.

For each of the constraints in the array of ranges rngs, this method puts the slack value in the solution of the relaxation at the current node into the corresponding element of the array vals.

This method returns the upper bound of the variable var at the current node. This bound is likely to be different from the bound in the original model because an instance of IloCplex tightens bounds when it branches from a node to its subnodes. The corresponding solution value from getValue may violate this bound at a node where a new incumbent has been found because the bound is tightened when an incumbent is found.

Unbounded Variables

If a variable lacks an upper bound, then getUB returns IloInfinity (that is, the system-specific greatest value).

This method returns the upper bound of the variable var at the current node. This bound is likely to be different from the bound in the original model because an instance of IloCplex tightens bounds when it branches from a node to its subnodes. The corresponding solution value from getValue may violate this bound at a node where a new incumbent has been found because the bound is tightened when an incumbent is found.

Unbounded Variables

If a variable lacks an upper bound, then getUB returns IloInfinity (that is, the system-specific greatest value).

For each element in the array vars, this method puts the upper bound at the current node into the corresponding element of the array vals. The bounds are those in the relaxation at the current node. These bounds are likely to be different from the bounds in the original model because an instance of IloCplex tightens bounds when it branches from a node to its subnodes. The corresponding solution values from getValues may violate these bounds at a node where a new incumbent has been found because the bounds are tightened when an incumbent is found.

Unbounded Variables

If a variable lacks an upper bound, then getUBs returns IloInfinity (that is, the system-specific greatest value).

For each element in the array vars, this method puts the upper bound at the current node into the corresponding element of the array vals. The bounds are those in the relaxation at the current node. These bounds are likely to be different from the bounds in the original model because an instance of IloCplex tightens bounds when it branches from a node to its subnodes. The corresponding solution values from getValues may violate these bounds at a node where a new incumbent has been found because the bounds are tightened when an incumbent is found.

Unbounded Variables

If a variable lacks an upper bound, then getUBs returns IloInfinity (that is, the system-specific greatest value).

This method returns the current pseudo cost for branching upward on the variable var.

This method returns the current pseudo cost for branching upward on the variable var.

This method returns the value of the variable var in the solution of the relaxation at the current node.

For special considerations about integrality, truncation, and rounding applicable to this method, see the concept Integer values, integrality tolerance, and round-off in CPLEX Optimizers.

This method returns the value of the variable var in the solution of the relaxation at the current node.

This method returns the value of the expression expr in the solution of the relaxation at the current node.

For each variable in the array vars, this method puts the value in the solution of the relaxation at the current node into the corresponding element of the array vals.

For special considerations about integrality, truncation, and rounding applicable to this method, see the concept Integer values, integrality tolerance, and round-off in CPLEX Optimizers.

For each variable in the array vars, this method puts the value in the solution of the relaxation at the current node into the corresponding element of the array vals.

This method returns IloTrue if the solution of the LP at the current node is SOS feasible for the special ordered set specified in its argument. The SOS set passed as a parameter to this method may be of type 2. See the CPLEX User's Manual for more explanation of types of special ordered sets.

Before you query the feasibility of a solution from a solve callback, a solution must exist. That is, you must first create the solution by calling the CPLEX optimization method solve, and then you must verify that the method solve generated a solution by checking its return value before you invoke the method isSOSFeasible.

This method returns IloTrue if the solution of the LP at the current node is SOS feasible for the special ordered set specified in its argument. The SOS set passed as a parameter to this method may be of type 1. See the CPLEX User's Manual for more explanation about these types of special ordered sets.

Before you query the feasibility of a solution from a solve callback, a solution must exist. That is, you must first create the solution by calling the CPLEX optimization method solve, and then you must verify that the method solve generated a solution by checking its return value before you invoke the method isSOSFeasible.

This method sets the NodeData object assigned to the the current node. If a NodeData object has already been set for the current node, then the method replaces this object and returns the old object. Otherwise, the method returns 0 (zero).

The enumeration IloCplex::ControlCallbackI::IntegerFeasibility is an enumeration limited in scope to the class IloCplex::ControlCallbackI. This enumeration is used by IloCplex::ControlCallbackI::getFeasibility to represent the integer feasibility of a variable or SOS in the current node solution:

• Feasible specifies the variable or SOS is integer feasible.
• ImpliedFeasible specifies the variable or SOS has been presolved out. It will be feasible when all other integer variables or SOS are integer feasible.
• Infeasible specifies the variable or SOS is integer infeasible.

This type defines an array-type for IloCplex::ControlCallbackI::IntegerFeasibility. The fully qualified name of an integer feasibility array is IloCplex::ControlCallbackI::IntegerFeasibility::Array.