Solver for a specific problem class. More...
#include <roboptim/core/solver.hh>
Public Types | |
typedef Problem< F, C > | problem_t |
Solver problem type. More... | |
typedef F::vector_t | vector_t |
Import vector type from cost function. More... | |
typedef std::map< std::string, Parameter > | parameters_t |
Map of parameters. More... | |
typedef SolverState< problem_t > | solverState_t |
State of the solver. More... | |
typedef boost::function< void(const problem_t &problem, solverState_t &state)> | callback_t |
Per-iteration callback type. More... | |
![]() | |
enum | solutions { SOLVER_NO_SOLUTION, SOLVER_VALUE, SOLVER_VALUE_WARNINGS, SOLVER_ERROR } |
Define the kind of solution which has been found. More... | |
typedef boost::variant< NoSolution, Result, ResultWithWarnings, SolverError > | result_t |
Result type. More... | |
Public Member Functions | |
Solver (const problem_t &problem) | |
Instantiate a solver from a problem. More... | |
template<typename F_ , typename C_ > | |
Solver (const Problem< F_, C_ > &problem) | |
Instantiate a solver from a problem in a different problem class. More... | |
virtual | ~Solver () |
const problem_t & | problem () const |
Retrieve the problem. More... | |
virtual void | setIterationCallback (callback_t) |
Set the per-iteration callback. More... | |
virtual std::ostream & | print (std::ostream &) const |
Display the solver on the specified output stream. More... | |
Parameters | |
const parameters_t & | parameters () const |
parameters_t & | parameters () |
template<typename T > | |
const T & | getParameter (const std::string &key) const |
Plugin name | |
const std::string & | pluginName () const |
std::string & | pluginName () |
![]() | |
void | reset () |
Force to restart the optimization. More... | |
virtual void | solve ()=0 |
Solve the problem. More... | |
const result_t & | minimum () |
Returns the function minimum This solves the problem automatically, if it has not yet been solved. More... | |
solutions | minimumType () |
Determine real minimum type. More... | |
template<typename T > | |
const T & | getMinimum () |
Get real result. More... | |
GenericSolver () | |
GenericSolver (const GenericSolver &) | |
virtual | ~GenericSolver () |
Protected Attributes | |
const problem_t | problem_ |
Problem that will be solved. More... | |
parameters_t | parameters_ |
Solver parameters (run-time configuration). More... | |
std::string | plugin_name_ |
Plugin name. More... | |
![]() | |
result_t | result_ |
Optimization result. More... | |
Static Protected Attributes | |
static log4cxx::LoggerPtr | logger |
Pointer to function logger (see log4cxx documentation). More... | |
![]() | |
static log4cxx::LoggerPtr | logger |
Pointer to function logger (see log4cxx documentation). More... | |
Solver for a specific problem class.
This class is parametrized by two types: the cost function type and the constraints type.
Solver classes are immutable, the problem can not be changed after the class instantiation.
F | cost function type |
C | constraints functions type |
typedef boost::function<void (const problem_t& problem, solverState_t& state)> roboptim::Solver< F, C >::callback_t |
Per-iteration callback type.
Callback parameters:
typedef std::map<std::string, Parameter> roboptim::Solver< F, C >::parameters_t |
Map of parameters.
typedef Problem<F, C> roboptim::Solver< F, C >::problem_t |
Solver problem type.
The solver can solve problems of this type. If another kind of problem is given, a conversion will be required.
typedef SolverState<problem_t> roboptim::Solver< F, C >::solverState_t |
State of the solver.
typedef F::vector_t roboptim::Solver< F, C >::vector_t |
Import vector type from cost function.
|
explicit |
Instantiate a solver from a problem.
problem | problem that should be solved |
|
explicit |
Instantiate a solver from a problem in a different problem class.
This constructor is called when the problem cost function or/and constraints type does not match solver's types.
This is only possible if the problem provides too much information compared to the solver requirements: if the problem contains twice derivable function and the solver requires only derivable function, it will work however the opposite will fail. Problem compatibility is known at compile-time, so the failure will be at compile-time.
F_ | original cost function type |
C_ | original constraints functions type |
problem | problem that should be solved |
|
virtual |
const T & roboptim::Solver< F, C >::getParameter | ( | const std::string & | key | ) | const |
const Solver< F, C >::parameters_t & roboptim::Solver< F, C >::parameters | ( | ) | const |
Solver< F, C >::parameters_t & roboptim::Solver< F, C >::parameters | ( | ) |
const std::string & roboptim::Solver< F, C >::pluginName | ( | ) | const |
std::string & roboptim::Solver< F, C >::pluginName | ( | ) |
|
virtual |
Display the solver on the specified output stream.
o | output stream used for display |
Reimplemented from roboptim::GenericSolver.
References roboptim::decindent(), roboptim::iendl(), roboptim::incindent(), and result_.
const Solver< F, C >::problem_t & roboptim::Solver< F, C >::problem | ( | ) | const |
Retrieve the problem.
|
inlinevirtual |
Set the per-iteration callback.
The per-iteration callback is a callback called each time one iteration of the optimization process is finished.
Not all the solvers support such a callback so this method may throw a std::runtime_error to let you know this feature is unsupported.
std::runtime_error |
Reimplemented in roboptim::GenericDummySolverLastState< F >.
|
staticprotected |
Pointer to function logger (see log4cxx documentation).
|
protected |
Solver parameters (run-time configuration).
|
protected |
Plugin name.
|
protected |
Problem that will be solved.