Nonlinear Program Large Scale Options Constructor
Constructs options for limited-memory, Newton and sparse nonlinear methods.
Definition
Namespace: Numerics.NET.Optimization
Assembly: Numerics.NET (in Numerics.NET.dll) Version: 10.9.0
C#
Assembly: Numerics.NET (in Numerics.NET.dll) Version: 10.9.0
public NonlinearProgramLargeScaleOptions(
int memory = 10,
NewtonCgPreconditioner preconditioner = NewtonCgPreconditioner.Automatic,
bool useNewtonLineSearch = false,
SparseNlpCurvature sparseCurvature = SparseNlpCurvature.Automatic,
SparseNlpLinearSolver sparseLinearSolver = SparseNlpLinearSolver.Adaptive,
bool enableFeasibilityRestoration = true,
bool enableSecondOrderCorrection = true
)Parameters
- memory Int32 (Optional)
- The number of retained BFGS steps. The default is 10.
- preconditioner NewtonCgPreconditioner (Optional)
- The Newton conjugate-gradient preconditioner.
- useNewtonLineSearch Boolean (Optional)
- Whether unbounded Newton steps use a backtracking line search.
- sparseCurvature SparseNlpCurvature (Optional)
- How sparse Lagrangian curvature is obtained.
- sparseLinearSolver SparseNlpLinearSolver (Optional)
- The factorization policy for sparse steps.
- enableFeasibilityRestoration Boolean (Optional)
- Whether sparse optimization can try a short recovery phase after a failed ordinary step.
- enableSecondOrderCorrection Boolean (Optional)
- Whether sparse optimization can correct steps rejected because constraints curve away from their linear approximation.
Exceptions
| Argument | Memory is not positive or an enum value is undefined. |