Class mmin_constr_gencan (o2scl)¶
-
template<class
param_t
, classfunc_t
, classdfunc_t
= func_t, classhfunc_t
= func_t, classvec_t
= boost::numeric::ublas::vector<double>>
classo2scl
::
mmin_constr_gencan
: public ool_constr_min<param_t, func_t, func_t, func_t, boost::numeric::ublas::vector<double>>¶ Constrained minimization by the “GENCAN” method (OOL)
- Note
Not working yet
Public Functions
-
mmin_constr_gencan
()¶
-
int
alloc
(const size_t n)¶ Allocate memory.
-
int
free
()¶ Free previously allocated memory.
-
int
set
(func_t &fn, dfunc_t &dfn, hfunc_t &hfn, vec_t &init, param_t &par)¶ Set the function, the initial guess, and the parameters.
-
int
restart
()¶ Restart the minimizer.
-
int
iterate
()¶ Perform an iteration.
-
int
is_optimal
()¶ See if we’re finished.
-
const char *
type
()¶ Return string denoting type (“mmin_constr_gencan”)
Public Members
-
double
epsgpen
¶ Tolerance on Euclidean norm of projected gradient (default 1.0e-5)
-
double
epsgpsn
¶ Tolerance on infinite norm of projected gradient (default 1.0e-5)
-
double
fmin
¶ Minimum function value (default \( 10^{-99} \))
If the function value is below this value, then the algorithm assumes that the function is not bounded and exits.
-
double
udelta0
¶ Trust-region radius (default -1.0)
-
double
ucgmia
¶ Maximum interations per variable (default -1.0)
-
double
ucgmib
¶ Extra maximum iterations (default -1.0)
-
int
cg_scre
¶ Conjugate gradient condition type (default 1)
-
double
cg_gpnf
¶ Projected gradient norm (default 1.0e-5)
-
double
cg_epsi
¶ Desc (default 1.0e-1)
-
double
cg_epsf
¶ Desc (default 1.0e-5)
-
double
cg_epsnqmp
¶ Stopping fractional tolerance for conjugate gradient (default 1.0e-4)
-
int
cg_maxitnqmp
¶ Maximum iterations for conjugate gradient (default 5)
-
int
nearlyq
¶ Set to 1 if the function is nearly quadratic (default 0)
-
double
nint
¶ Interpolation constant (default 2.0)
-
double
next
¶ Extrapolation constant (default 2.0)
-
int
mininterp
¶ Minimum interpolation size (default 4)
-
int
maxextrap
¶ Maximum extrapolations in truncated Newton direction (default 100)
-
int
trtype
¶ Type of trust region (default 0)
-
double
eta
¶ Threshold for abandoning current face (default 0.9)
-
double
delmin
¶ Minimum trust region for truncated Newton direction (default 0.1)
-
double
lspgmi
¶ Minimum spectral steplength (default 1.0e-10)
-
double
lspgma
¶ Maximum spectral steplength (default 1.0e10)
-
double
theta
¶ Constant for the angle condition (default 1.0e-6)
-
double
gamma
¶ Constant for Armijo condition (default 1.0e-4)
-
double
beta
¶ Constant for beta condition (default 0.5)
-
double
sigma1
¶ Lower bound to the step length reduction (default 0.1)
-
double
sigma2
¶ Upper bound to the step length reduction (default 0.9)
-
double
epsrel
¶ Relative small number (default 1.0e-7)
-
double
epsabs
¶ Absolute small number (default 1.0e-10)
-
double
infrel
¶ Relative infinite number (default 1.0e20)
-
double
infabs
¶ Absolute infinite number (default 1.0e99)