$extrastylesheet
Dakota
Version 6.2
|
Base class for OPT++ optimization and least squares methods. More...
Public Member Functions | |
SNLLBase () | |
default constructor | |
SNLLBase (ProblemDescDB &problem_db) | |
standard constructor | |
~SNLLBase () | |
destructor | |
Protected Member Functions | |
void | copy_con_vals_dak_to_optpp (const RealVector &local_fn_vals, RealVector &g, size_t offset) |
convenience function for copying local_fn_vals to g; used by constraint evaluator functions | |
void | copy_con_vals_optpp_to_dak (const RealVector &g, RealVector &local_fn_vals, size_t offset) |
convenience function for copying g to local_fn_vals; used in final solution logging | |
void | copy_con_grad (const RealMatrix &local_fn_grads, RealMatrix &grad_g, size_t offset) |
convenience function for copying local_fn_grads to grad_g; used by constraint evaluator functions | |
void | copy_con_hess (const RealSymMatrixArray &local_fn_hessians, OPTPP::OptppArray< RealSymMatrix > &hess_g, size_t offset) |
convenience function for copying local_fn_hessians to hess_g; used by constraint evaluator functions | |
void | snll_pre_instantiate (bool bound_constr_flag, int num_constr) |
convenience function for setting OPT++ options prior to the method instantiation | |
void | snll_post_instantiate (int num_cv, bool vendor_num_grad_flag, const String &finite_diff_type, const RealVector &fdss, int max_iter, int max_fn_evals, Real conv_tol, Real grad_tol, Real max_step, bool bound_constr_flag, int num_constr, short output_lev, OPTPP::OptimizeClass *the_optimizer, OPTPP::NLP0 *nlf_objective, OPTPP::FDNLF1 *fd_nlf1, OPTPP::FDNLF1 *fd_nlf1_con) |
convenience function for setting OPT++ options after the method instantiation | |
void | snll_initialize_run (OPTPP::NLP0 *nlf_objective, OPTPP::NLP *nlp_constraint, const RealVector &init_pt, bool bound_constr_flag, const RealVector &lower_bnds, const RealVector &upper_bnds, const RealMatrix &lin_ineq_coeffs, const RealVector &lin_ineq_l_bnds, const RealVector &lin_ineq_u_bnds, const RealMatrix &lin_eq_coeffs, const RealVector &lin_eq_targets, const RealVector &nln_ineq_l_bnds, const RealVector &nln_ineq_u_bnds, const RealVector &nln_eq_targets) |
convenience function for OPT++ configuration prior to the method invocation | |
void | snll_post_run (OPTPP::NLP0 *nlf_objective) |
convenience function for setting OPT++ options after the method instantiations | |
Static Protected Member Functions | |
static void | init_fn (int n, RealVector &x) |
An initialization mechanism provided by OPT++ (not currently used). | |
Protected Attributes | |
String | searchMethod |
value_based_line_search, gradient_based_line_search, trust_region, or tr_pds | |
OPTPP::SearchStrategy | searchStrat |
enum: LineSearch, TrustRegion, or TrustPDS | |
OPTPP::MeritFcn | meritFn |
enum: NormFmu, ArgaezTapia, or VanShanno | |
Real | maxStep |
value from max_step specification | |
Real | stepLenToBndry |
value from steplength_to_boundary specification | |
Real | centeringParam |
value from centering_parameter specification | |
bool | constantASVFlag |
flags a user selection of active_set_vector == constant. By mapping this into mode override, reliance on duplicate detection can be avoided. | |
Static Protected Attributes | |
static Minimizer * | optLSqInstance |
pointer to the active base class object instance used within the static evaluator functions in order to avoid the need for static data | |
static bool | modeOverrideFlag |
flags OPT++ mode override (for combining value, gradient, and Hessian requests) | |
static EvalType | lastFnEvalLocn |
an enum used to track whether an nlf evaluator or a constraint evaluator was the last location of a function evaluation | |
static int | lastEvalMode |
copy of mode from constraint evaluators | |
static RealVector | lastEvalVars |
copy of variables from constraint evaluators |
Base class for OPT++ optimization and least squares methods.
The SNLLBase class provides a common base class for SNLLOptimizer and SNLLLeastSq, both of which are wrappers for OPT++, a C++ optimization library from the Computational Sciences and Mathematics Research (CSMR) department at Sandia's Livermore CA site.