Public Member Functions

Optimize (StateIn &)
Restore the state of a Function object. Optimize (const Ref< KeyVal > &)
The KeyVal constructor reads the following information: void save_data_state (StateOut &)
Save the base classes (with save_data_state) and the members in the same order that the StateIn CTOR initializes them. virtual int optimize ()
Do the optimization. void set_checkpoint ()
Set up for checkpointing. void set_checkpoint_file (const char *)
void set_function (const Ref< Function > &)
Set the function to be optimized. void set_max_iterations (int)
Set the iteration limit. virtual void init ()
Initialize the optimizer. virtual int update ()=0
Take a step. virtual void apply_transform (const Ref< NonlinearTransform > &)
Ref< Function > function () const
Returns information about the Function being optimized. Ref< SCMatrixKit > matrixkit () const
RefSCDimension dimension () const

Protected Attributes

int max_iterations_
int n_iterations_
int ckpt_
int print_timings_
double max_stepsize_
char * ckpt_file
Ref< Function > function_
Ref< Convergence > conv_

Additional Inherited Members

sc::Optimize::Optimize (const Ref< KeyVal > &)

The KeyVal constructor reads the following information:

checkpoint

If true, the optimization will be checkpointed. The default is false.

checkpoint_file

The name of the checkpoint file. The name defaults to opt_ckpt.dat.

max_iterations

The maximum number of interations. The default is 10.

max_stepsize

The maximum stepsize. The default is 0.6.

function

A Function object. There is no default.

convergence

This can be either a floating point number or a Convergence object. If it is a floating point number then it is the convergence criterion. See the description Convergence class for the default.

virtual int sc::Optimize::optimize () [virtual]

Do the optimization. Returns nonzero if the optimization is complete.

void sc::Optimize::save_data_state (StateOut &) [virtual]

Save the base classes (with save_data_state) and the members in the same order that the StateIn CTOR initializes them. This must be implemented by the derived class if the class has data.

Reimplemented from sc::SavableState.

Reimplemented in sc::SteepestDescentOpt, sc::QNewtonOpt, and sc::LineOpt.

virtual int sc::Optimize::update () [pure virtual]

Take a step. Returns 1 if the optimization has converged, otherwise 0.

Implemented in sc::SteepestDescentOpt, sc::QNewtonOpt, sc::Backtrack, sc::NewtonOpt, sc::MCSearch, sc::GDIISOpt, and sc::EFCOpt.