ultra::src::search< P > Class Template Reference

#include <search.h>

Public Types
using	individual_t = P
using	fitness_t = double
using	after_generation_callback_t
using	on_training_new_best_callback_t
using	class_evaluator_t = gaussian_evaluator<P>
using	reg_evaluator_t = rmae_evaluator<P>

Public Member Functions
	search (problem &, metric_flags=metric_flags::accuracy)
search_stats< P, fitness_t >	run (unsigned=1, const model_measurements< fitness_t > &={})
std::unique_ptr< basic_oracle >	predictor (const P &) const
search &	after_generation (after_generation_callback_t)
search &	logger (search_log &)
search &	messages (bool) noexcept
search &	on_training_new_best (on_training_new_best_callback_t)
template<class F>
search &	refinement (F &&)
search &	stop_source (std::stop_source)
search &	tag (const std::string &)
template<ValidationStrategy, class... Args>
search &	validation_strategy (Args &&...)
template<class F>
search< P > &	refinement (F &&f)
template<ValidationStrategy V, class... Args>
search< P > &	validation_strategy (Args &&...args)
template<class F>
search< P > &	refinement (F &&f)
template<ValidationStrategy V, class... Args>
search< P > &	validation_strategy (Args &&...args)

Detailed Description

template<Individual P>
class ultra::src::search< P >

Thin facade over the concrete classification or regression search engine.

The engine kind is selected at construction time from the problem type. Configuration calls are forwarded directly to the selected engine.

Examples

sonar.cc, symbolic_regression.cc, symbolic_regression01.cc, and symbolic_regression02.cc.

Member Typedef Documentation

after_generation_callback_t

template<Individual P>

using ultra::src::search< P >::after_generation_callback_t

Initial value:

 
    ultra::after_generation_callback_t<P, fitness_t>

on_training_new_best_callback_t

template<Individual P>

using ultra::src::search< P >::on_training_new_best_callback_t

Initial value:

 
    ultra::on_new_best_callback_t<P, fitness_t>

Member Function Documentation

after_generation()

template<Individual P>

search< P > & search::after_generation

(

after_generation_callback_t

f

)

Sets a callback function executed at the end of every generation.

Parameters

[in]

f

callback function

Returns

a reference to this object (method chaining / fluent interface)

Examples

symbolic_regression02.cc.

logger()

template<Individual P>

search< P > & search::logger

(

search_log &

sl

)

Sets the search/evolution logger.

Parameters

[in]

sl

logger

Returns

a reference to this object (method chaining / fluent interface)

Remarks

Logger must be set before calling run. By default, data logging is disabled.

on_training_new_best()

template<Individual P>

search< P > & search::on_training_new_best

(

on_training_new_best_callback_t

f

)

Hook invoked whenever the evolution discovers a new training-set best.

Parameters

[in]

f

callback function

Returns

a reference to this object (method chaining / fluent interface)

predictor()

template<Individual P>

std::unique_ptr< basic_oracle > search::predictor ( const P & prg ) const

nodiscard

Creates a runtime-polymorphic predictor associated with a given individual.

Parameters

[in]

prg

individual used as building block for the predictor

Returns

a dynamically allocated predictor or nullptr if the active evaluator does not provide one

This function provides a uniform interface over different evaluator types. If the selected evaluator exposes a predictor() method, its result is wrapped into a std::unique_ptr<basic_oracle>. Otherwise nullptr is returned.

Note

Predictor availability is detected at compile time and mapped to a runtime polymorphic interface.

refinement() [1/2]

template<Individual P>

template<class F>

search< P > & ultra::src::search< P >::refinement

(

F &&

f

)

Sets a refinement callback used to locally improve individuals.

Parameters

[in]

f

callable object implementing the refinement logic

Returns

a reference to *this object (method chaining / fluent interface)

The refinement callback is invoked during the evolutionary process to perform local optimisation (e.g. numerical tuning of parameters) on candidate solutions.

The callable must have a signature compatible with at least one of the following forms:

std::optional<fitness_t>(
  P &, const evaluator_proxy<class_evaluator_t> &,
  const parameters::refinement_parameters &)
 
std::optional<fitness_t>(
  P &, const evaluator_proxy<reg_evaluator_t> &,
  const parameters::refinement_parameters &)

A generic callable can be used to support both cases:

[](P &ind, const auto &eva,
   const parameters::refinement_parameters &params)
{
  // eva is the evaluator proxy used by the search engine
  // (includes caching and other internal mechanisms)
  return std::optional<fitness_t>();
}

The evaluator passed to the callback is the same object used internally by the evolutionary engine (i.e. an evaluator_proxy), ensuring that refinement benefits from cached evaluations and consistent semantics.

Note

The concrete evaluator type depends on the problem (classification or regression) and is selected automatically at runtime. An evaluator-specific callable that does not match the selected evaluator disables refinement.

refinement() [2/2]

template<Individual P>

template<class F>

search< P > & ultra::src::search< P >::refinement

(

F &&

f

)

Sets a refinement callback used to locally improve individuals.

Parameters

[in]

f

callable object implementing the refinement logic

Returns

a reference to *this object (method chaining / fluent interface)

The refinement callback is invoked during the evolutionary process to perform local optimisation (e.g. numerical tuning of parameters) on candidate solutions.

The callable must have a signature compatible with at least one of the following forms:

std::optional<fitness_t>(
  P &, const evaluator_proxy<class_evaluator_t> &,
  const parameters::refinement_parameters &)
 
std::optional<fitness_t>(
  P &, const evaluator_proxy<reg_evaluator_t> &,
  const parameters::refinement_parameters &)

A generic callable can be used to support both cases:

[](P &ind, const auto &eva,
   const parameters::refinement_parameters &params)
{
  // eva is the evaluator proxy used by the search engine
  // (includes caching and other internal mechanisms)
  return std::optional<fitness_t>();
}

The evaluator passed to the callback is the same object used internally by the evolutionary engine (i.e. an evaluator_proxy), ensuring that refinement benefits from cached evaluations and consistent semantics.

Note

The concrete evaluator type depends on the problem (classification or regression) and is selected automatically at runtime. An evaluator-specific callable that does not match the selected evaluator disables refinement.

stop_source()

template<Individual P>

search< P > & search::stop_source

(

std::stop_source

ss

)

Set a stop source for performing cooperative task shutdown.

Parameters

[in]

ss

stop source to issue a stop request

Returns

a reference to this object (method chaining / fluent interface)

tag()

template<Individual P>

search< P > & search::tag

(

const std::string &

t

)

Sets the identification tag for this object.

Parameters

[in]

t

identification tag

Returns

a reference to this object (method chaining / fluent interface)

The tag is used to identify this object when multiple searches are performed in parallel.

validation_strategy() [1/2]

template<Individual P>

template<ValidationStrategy V, class... Args>

search< P > & ultra::src::search< P >::validation_strategy

(

Args &&...

args

)

Builds and sets the active validation strategy.

Parameters

[in]

args

parameters for the validation strategy

Returns

reference to the search class (used for method chaining)

validation_strategy() [2/2]

template<Individual P>

template<ValidationStrategy V, class... Args>

search< P > & ultra::src::search< P >::validation_strategy

(

Args &&...

args

)

Builds and sets the active validation strategy.

Parameters

[in]

args

parameters for the validation strategy

Returns

reference to the search class (used for method chaining)

---

The documentation for this class was generated from the following files:

• gp/src/search.h
• gp/src/search.tcc