$extrastylesheet
Dakota
Version 6.2
|
Wrapper class for the NL2SOL nonlinear least squares library. More...
Public Member Functions | |
NL2SOLLeastSq (ProblemDescDB &problem_db, Model &model) | |
standard constructor | |
NL2SOLLeastSq (Model &model) | |
alternate constructor | |
~NL2SOLLeastSq () | |
destructor | |
void | minimize_residuals () |
Used within the least squares branch for minimizing the sum of squares residuals. Redefines the run virtual function for the least squares branch. | |
Static Private Member Functions | |
static void | calcr (int *np, int *pp, Real *x, int *nfp, Real *r, int *ui, void *ur, Vf vf) |
evaluator function for residual vector | |
static void | calcj (int *np, int *pp, Real *x, int *nfp, Real *J, int *ui, void *ur, Vf vf) |
evaluator function for residual Jacobian | |
Private Attributes | |
int | auxprt |
auxilary printing bits (see Dakota Ref Manual): sum of < 1 = x0prt (print initial guess) < 2 = solprt (print final solution) < 4 = statpr (print solution statistics) < 8 = parprt (print nondefault parameters) < 16 = dradpr (print bound constraint drops/adds) < debug/verbose/normal use default = 31 (everything), < quiet uses 3, silent uses 0. | |
int | outlev |
frequency of output summary lines in number of iterations < (debug/verbose/normal/quiet use default = 1, silent uses 0) | |
Real | dltfdj |
finite-diff step size for computing Jacobian approximation < (fd_gradient_step_size ) | |
Real | delta0 |
finite-diff step size for gradient differences for H < (a component of some covariance approximations, if desired) < (fd_hessian_step_size ) | |
Real | dltfdc |
finite-diff step size for function differences for H < (fd_hessian_step_size ) | |
int | mxfcal |
function-evaluation limit (max_function_evaluations ) | |
int | mxiter |
iteration limit (max_iterations ) | |
Real | rfctol |
relative fn convergence tolerance (convergence_tolerance ) | |
Real | afctol |
absolute fn convergence tolerance (absolute_conv_tol ) | |
Real | xctol |
x-convergence tolerance (x_conv_tol ) | |
Real | sctol |
singular convergence tolerance (singular_conv_tol ) | |
Real | lmaxs |
radius for singular-convergence test (singular_radius ) | |
Real | xftol |
false-convergence tolerance (false_conv_tol ) | |
int | covreq |
kind of covariance required (covariance ): < 1 or -1 ==> sigma^2 H^-1 J^T J H^-1 < 2 or -2 ==> sigma^2 H^-1 < 3 or -3 ==> sigma^2 (J^T J)^-1 < 1 or 2 ==> use gradient diffs to estimate H < -1 or -2 ==> use function diffs to estimate H < default = 0 (no covariance) | |
int | rdreq |
whether to compute the regression diagnostic vector < (regression_diagnostics ) | |
Real | fprec |
expected response function precision (function_precision ) | |
Real | lmax0 |
initial trust-region radius (initial_trust_radius ) | |
Static Private Attributes | |
static NL2SOLLeastSq * | nl2solInstance |
pointer to the active object instance used within the static evaluator functions |
Wrapper class for the NL2SOL nonlinear least squares library.
The NL2SOLLeastSq class provides a wrapper for NL2SOL (TOMS Algorithm 573), in the updated form of Port Library routines dn[fg][b ] from Bell Labs; see http://www.netlib.org/port/readme. The Fortran from Port has been turned into C by f2c. NL2SOL uses a function pointer approach for which passed functions must be either global functions or static member functions.
void minimize_residuals | ( | ) | [virtual] |
Used within the least squares branch for minimizing the sum of squares residuals. Redefines the run virtual function for the least squares branch.
Details on the following subscript values appear in "Usage Summary for Selected Optimization Routines" by David M. Gay, Computing Science Technical Report No. 153, AT&T Bell Laboratories, 1990. http://netlib.bell-labs.com/cm/cs/cstr/153.ps.gz
Implements LeastSq.
References NL2SOLLeastSq::afctol, NL2SOLLeastSq::auxprt, Iterator::bestResponseArray, Iterator::bestVariablesArray, Minimizer::boundConstraintFlag, NL2SOLLeastSq::calcj(), NL2SOLLeastSq::calcr(), Minimizer::calibrationDataFlag, Model::continuous_lower_bounds(), Model::continuous_upper_bounds(), Model::continuous_variables(), Dakota::copy_data(), NL2SOLLeastSq::covreq, NL2SOLLeastSq::delta0, NL2SOLLeastSq::dltfdc, NL2SOLLeastSq::dltfdj, Minimizer::expData, NL2SOLLeastSq::fprec, Model::gradient_type(), ExperimentData::interpolate_flag(), Iterator::iteratedModel, NL2SOLLeastSq::lmax0, NL2SOLLeastSq::lmaxs, NL2SOLLeastSq::mxfcal, NL2SOLLeastSq::mxiter, NL2SOLLeastSq::nl2solInstance, Minimizer::numContinuousVars, LeastSq::numLeastSqTerms, Minimizer::numUserPrimaryFns, NL2SOLLeastSq::outlev, NL2SOLLeastSq::rdreq, NL2SOLLeastSq::rfctol, NL2SOLLeastSq::sctol, Minimizer::speculativeFlag, Minimizer::vendorNumericalGradFlag, NL2SOLLeastSq::xctol, and NL2SOLLeastSq::xftol.