Trait Algorithm

pub trait Algorithm<AlgorithmOptions: Serialize + DeserializeOwned>: Display {
    // Required methods
    fn initialise(&mut self) -> Result<(), OError>;
    fn evolve(&mut self) -> Result<(), OError>;
    fn generation(&self) -> u32;
    fn number_of_function_evaluations(&self) -> u32;
    fn name(&self) -> String;
    fn start_time(&self) -> &Instant;
    fn stopping_condition(&self) -> &StoppingCondition;
    fn population(&self) -> &Population;
    fn problem(&self) -> Arc<Problem>;
    fn export_history(&self) -> Option<&ExportHistory>;
    fn algorithm_options(&self) -> AlgorithmOptions;

    // Provided methods
    fn additional_export_data(&self) -> Option<HashMap<String, DataValue>> { ... }
    fn elapsed(&self) -> [u64; 3] { ... }
    fn elapsed_as_string(&self) -> String { ... }
    fn do_parallel_evaluation(
        individuals: &mut [Individual],
        nfe: &mut u32,
    ) -> Result<(), OError> { ... }
    fn do_evaluation(
        individuals: &mut [Individual],
        nfe: &mut u32,
    ) -> Result<(), OError> { ... }
    fn evaluate_individual(idx: usize, i: &mut Individual) -> Result<(), OError> { ... }
    fn count_unevaluated(individuals: &[Individual]) -> u32 { ... }
    fn run(&mut self) -> Result<(), OError> { ... }
    fn is_stopping_condition_met(
        &self,
        condition: &StoppingCondition,
    ) -> Result<bool, OError> { ... }
    fn get_results(&self) -> AlgorithmExport { ... }
    fn save_to_json(
        &self,
        destination: &PathBuf,
        file_prefix: Option<&str>,
    ) -> Result<(), OError> { ... }
    fn read_json_file(
        file: &PathBuf,
    ) -> Result<AlgorithmSerialisedExport<AlgorithmOptions>, OError> { ... }
    fn read_json_files(
        folder: &PathBuf,
    ) -> Result<Vec<AlgorithmSerialisedExport<AlgorithmOptions>>, OError> { ... }
    fn seed_population_from_file(
        problem: Arc<Problem>,
        name: &str,
        expected_individuals: usize,
        file: &PathBuf,
    ) -> Result<Population, OError> { ... }
}

The trait to use to implement an algorithm.

Required Methods

fn initialise(&mut self) -> Result<(), OError>

Initialise the algorithm.

return: Result<(), OError>

fn evolve(&mut self) -> Result<(), OError>

Evolve the population.

return: Result<(), OError>

fn generation(&self) -> u32

Return the current step of the algorithm evolution.

return: u32.

fn number_of_function_evaluations(&self) -> u32

Return the number of function evaluations. This is the number of times the algorithm evaluates an individual’s objectives and constraints using Algorithm::evaluate_individual. If no new solutions/individuals are chosen by an algorithm, this counter will not increase, as past solutions are already evaluated.

return: u32.

fn name(&self) -> String

Return the algorithm name.

return: String.

fn start_time(&self) -> &Instant

Get the time when the algorithm started.

return: &Instant.

fn stopping_condition(&self) -> &StoppingCondition

Return the stopping condition.

return: &StoppingConditionType.

fn population(&self) -> &Population

Return the evolved population.

return: &Population.

fn problem(&self) -> Arc<Problem>

Return the problem.

return: Arc<Problem>.

fn export_history(&self) -> Option<&ExportHistory>

Return the history export configuration, if provided by the algorithm.

return: Option<&ExportHistory>.

fn algorithm_options(&self) -> AlgorithmOptions

Provided Methods

fn additional_export_data(&self) -> Option<HashMap<String, DataValue>>

Export additional data stored by the algorithm.

return: Option<HashMap<String, DataValue>>

fn elapsed(&self) -> [u64; 3]

Get the elapsed hours, minutes and seconds since the start of the algorithm.

return: [u64; 3]. An array with the number of elapsed hours, minutes and seconds.

fn elapsed_as_string(&self) -> String

Format the elapsed time as string.

return: String.

fn do_parallel_evaluation( individuals: &mut [Individual], nfe: &mut u32, ) -> Result<(), OError>

Evaluate the objectives and constraints for unevaluated individuals in the population. This updates the individual data only, runs the evaluation function in threads and increase the nfe counter by the number of evaluated individuals. This returns an error if the evaluation function fails or the evaluation function does not provide a value for a problem constraints or objectives for one individual.

Arguments

• individuals: The individuals to evaluate.
• nfe: The reference to the number of function evaluation counter.

return Result<(), OError>

fn do_evaluation( individuals: &mut [Individual], nfe: &mut u32, ) -> Result<(), OError>

Evaluate the objectives and constraints for unevaluated individuals in the population. This updates the individual data only, runs the evaluation function in a plain loop and increase the nfe counter by the number of evaluated individuals. This returns an error if the evaluation function fails or the evaluation function does not provide a value for a problem constraints or objectives for one individual. Evaluation may be performed in threads using Self::do_parallel_evaluation.

Arguments

• individuals: The individuals to evaluate.
• nfe: The reference to the number of function evaluation counter.

return Result<usize, OError>.

fn evaluate_individual(idx: usize, i: &mut Individual) -> Result<(), OError>

Evaluate the objectives and constraints for one unevaluated individual. This returns an error if the evaluation function fails or the evaluation function does not provide a value for a problem constraints or objectives.

Arguments

• idx: The individual index.
• individual: The individual to evaluate.

return Result<(), OError>

fn count_unevaluated(individuals: &[Individual]) -> u32

Count the number on unevaluated individuals.

Arguments

• individuals: The individuals to check.

returns: u32

fn run(&mut self) -> Result<(), OError>

Run the algorithm.

return: Result<(), OError>

fn is_stopping_condition_met( &self, condition: &StoppingCondition, ) -> Result<bool, OError>

Check if the given stopping condition is met.

Arguments

• condition: The stopping condition type.

returns: Result<bool, OError>

fn get_results(&self) -> AlgorithmExport

Get the results of the run.

return: AlgorithmExport.

fn save_to_json( &self, destination: &PathBuf, file_prefix: Option<&str>, ) -> Result<(), OError>

Save the algorithm data (individuals’ objective, variables and constraints, the problem, …) to a JSON file. This returns an error if the file cannot be saved.

Arguments

• destination: The path to the JSON file.
• file_prefix: A prefix to prepend at the beginning of the file name. Empty when None.

return Result<(), OError>

fn read_json_file( file: &PathBuf, ) -> Result<AlgorithmSerialisedExport<AlgorithmOptions>, OError>

Read the results previously exported with Self::save_to_json.

Arguments

• file: The path to the JSON file exported from this library.

returns: Result<AlgorithmSerialisedExport<T>, OError>

fn read_json_files( folder: &PathBuf, ) -> Result<Vec<AlgorithmSerialisedExport<AlgorithmOptions>>, OError>

Read the results from files exported during an algorithm evolution. This returns an error if the path does not exist or does not contain valid JSON files.

Arguments

• folder: The folder path to the JSON files.

returns: Result<Vec<AlgorithmSerialisedExport<T>>, OError>

fn seed_population_from_file( problem: Arc<Problem>, name: &str, expected_individuals: usize, file: &PathBuf, ) -> Result<Population, OError>

Seed the population using the values of variables, objectives and constraints exported to a JSON file.

Arguments

• problem: The problem.
• name: The algorithm name.
• expected_individuals: The number of individuals to expect in the file. If this does not match the population size, being used in the algorithm, an error is thrown.
• file: The path to the JSON file exported from this library.

returns: Result<Population, OError>

Dyn Compatibility

This trait is not dyn compatible.

In older versions of Rust, dyn compatibility was called "object safety", so this trait is not object safe.

Implementors

impl Algorithm<NSGA2Arg> for NSGA2

Implementation of Section IIIC of the paper.

impl Algorithm<NSGA3Arg> for NSGA3

Implementation of Section IV of the paper.