Function GMP::SolverSession::GetAttributeValue(GMP, attr, objNo)

GMP::SolverSession::GetAttributeValue

The function GMP::Instance::GetAttributeValue can be used to get the value of certain solver attributes, related to an executed solver session.

GMP::SolverSession::GetAttributeValue(
     solverSession,  ! (input) a solver session
     attr,           ! (input) a string expression
     [objNo]         ! (input, optional) an integer expression
     )

Arguments

solverSession

An element in the set AllSolverSessions.

attr

A string that holds the name of the attribute.

objNo

A nonnegative scalar value specifying the objective number in a multi-objective optimization problem. The default value is 0.

Return Value

In case of success, the function returns the attribute value. Otherwise it returns UNDF.

Note

• The attribute name is case-insensitive.

• This function is only supported by CPLEX and Gurobi.

• This function can be called inside an incumbent, lazy, cuts or heuristic callback procedure, but only for Gurobi.

Attributes

The table below shows the model attributes supported by CPLEX and Gurobi.

Attribute	CPLEX	Gurobi	Description
ObjCon	✔	✔	Objective constant/offset
NumStart	✔	✔	Number of MIP starts (MIP only)
NumObj	✔	✔	Number of objectives

The table below shows the solution attributes supported by CPLEX and Gurobi.

Attribute	CPLEX	Gurobi	Description
ObjVal	✔	✔	Objective value
ObjBound	✔	✔	Best bound (MIP only)
MIPGap	✔	✔	Relative MIP gap (MIP only)
Runtime	✔	✔	Runtime in seconds
Work	✔	✔	Work or deterministic time
MaxMemUsed		✔	Maximum amount (in GB) of memory used
MemUsed		✔	Current amount (in GB) of memory used
SolCount		✔	Number of solutions (MIP only)
IterCount	✔	✔	Number of simplex iterations
BarIterCount	✔	✔	Number of barrier iterations
NodeCount	✔	✔	Number of branch-and-cut nodes (MIP only)
OpenNodeCount	✔	✔	Number of open branch-and-cut nodes (MIP only)
NodeInt	✔		Node number of the best solution (MIP only)

The table below shows the solution quality attributes supported by CPLEX and Gurobi.

Attribute	CPLEX	Gurobi	Description
MaxVio	✔	✔	Maximum unscaled violation
IntVio	✔	✔	Maximum integrality violation
Kappa	✔	✔	Estimated condition number
KappaExact	✔	✔	Exact condition number
KappaMax	✔		Highest condition number (MIP only)
KappaAttention	✔		Score of numerical difficulties (MIP only)
KappaStable	✔		Percentage of stable simplex bases (MIP only)
KappaSuspicious	✔		Percentage of suspicious simplex bases (MIP only)
KappaUnstable	✔		Percentage of unstable simplex bases (MIP only)
KappaIllposed	✔		Percentage of ill-posed simplex bases (MIP only)

The table below shows the attributes for multi-objective optimization supported by CPLEX and Gurobi. These attribites can be used to retrieve model or solution information for each subproblem solved, as specified by the objNo argument.

Attribute	CPLEX	Gurobi	Description
ObjNPriority	✔	✔	Objective priority
ObjNCon		✔	Objective constant/offset
ObjVal/ObjNVal	✔	✔	Objective value
ObjBound	✔		Best bound (MIP only)
MIPGap	✔		Relative MIP gap (MIP only)
Runtime	✔		Runtime in seconds
Work	✔		Work or deterministic time
Status	✔		Solution status
NodeCount	✔		Number of branch-and-cut nodes (MIP only)
OpenNodeCount	✔		Number of open branch-and-cut nodes (MIP only)
IterCount	✔		Number of simplex iterations
BarIterCount	✔		Number of barrier iterations

For Gurobi also other model and solution attributes are supported. For a complete list; see: Model attributes. Attributes with type ‘string’ are not supported by this function.

The table below shows the attributes that can be called from inside an incumbent, lazy, cuts or heuristic callback procedure, but only for Gurobi.

Attribute	Description
ObjBest	Current best objective
ObjBound	Current best objective bound
NodeCount	Current explored node count
SolCount	Current number of solutions found
Phase	Current phase in the MIP solution
ObjVal	Objective value for new solution

Example

In the example below two multi-objectives are specified, each with its own priority. Therefore two subproblems are solved; one for each objective. Below we retrieve the (relative) MIP gap for both subproblems. We also retrieve the total runtime by the solver.

myGMP := GMP::Instance::Generate( MP );

GMP::Column::SetAsMultiObjective( myGMP, TotalDist, 2, 1.0 );
GMP::Column::SetAsMultiObjective( myGMP, TotalTime, 1, 1.0 );

solSes := GMP::Instance::CreateSolverSession( myGMP 0;

GMP::SolverSession::Execute( solSes );

GMP::Solution::RetrieveFromSolverSession( solSes, 1 );
GMP::Solution::SendToModel( myGMP, 1 );

gap1 := GMP::SolverSession::GetAttributeValue( solSes, "MIPGap", 1 );
gap2 := GMP::SolverSession::GetAttributeValue( solSes, "MIPGap", 2 );

runtime := GMP::SolverSession::GetAttributeValue( solSes, "Runtime" );

See also

The routines GMP::Instance::CreateSolverSession, GMP::Instance::Generate, GMP::SolverSession::Execute, GMP::Column::SetAsMultiObjective and GMP::Instance::GetAttributeValue.