ultra::parameters Struct Reference

Classes
struct	slp_parameters
struct	population_parameters
struct	evolution_parameters
struct	de_parameters
struct	refinement_parameters
	Parameters controlling numerical optimisation of GP individuals. More...
struct	cache_parameters
struct	team_parameters

Public Member Functions
parameters &	fill_undefined_from (const parameters &)
parameters &	init ()
bool	needs_init () const noexcept
bool	is_valid (bool) const

Public Attributes
struct ultra::parameters::slp_parameters	slp
struct ultra::parameters::population_parameters	population
struct ultra::parameters::evolution_parameters	evolution
alps::parameters	alps
struct ultra::parameters::de_parameters	de
struct ultra::parameters::refinement_parameters	refinement
struct ultra::parameters::cache_parameters	cache
struct ultra::parameters::team_parameters	team
std::size_t	min_samples_per_eval_thread {0}

Member Function Documentation

fill_undefined_from()

parameters & ultra::parameters::fill_undefined_from

(

const parameters &

defaults

)

Fills undefined parameters with values from another parameter set.

Parameters

[in]

defaults

source of fallback values

Returns

a reference to this parameter set

init()

parameters & ultra::parameters::init

(

)

Initialises the undefined parameters with "common" values.

Returns

a reference to the "filled" struct parameters

Usually:

• the undefined parameters are tuned before the start of the search (search::run calls search::tune_parameters) when there are enough data at hand;
• the user doesn't have to fiddle with them (except after careful consideration).

This function is mainly convenient for debugging purpose. The chosen values are reasonable but most likely far from ideal.

See also

search::tune_parameters

is_valid()

bool ultra::parameters::is_valid ( bool force_defined ) const

nodiscard

Parameters

[in]

force_defined

all the undefined / auto-tuned parameters have to be in a "well defined" state for the function to pass the test

Returns

true if the object passes the internal consistency check

Member Data Documentation

min_samples_per_eval_thread

std::size_t ultra::parameters::min_samples_per_eval_thread {0}

Minimum number of dataset examples per thread during parallel evaluation.

Evaluators supporting parallel dataset evaluation divide workloads only when each worker thread can be assigned at least this number of samples.

Note

0 means auto-tune (use the evaluator's default threshold).

---

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

• parameters.h
• parameters.cc