Struct Simulation

pub struct Simulation<'a, T: SimulatingWorld> { /* private fields */ }

Main struct that handles the learning logic.

Implementations

impl<'a, T: SimulatingWorld> Simulation<'a, T>

pub fn new( population_size: usize, neurons_per_layer: &[usize], randomizer: &'a mut dyn RandomProvider, mutation_probability: Option<f64>, ) -> Result<Simulation<'a, T>, String>

Creates new simulation struct.

mutation_probability is represented as float number from range [0.0, 1.0).

pub fn run_with_parents( &mut self, finish: Finish, parents: [Specimen; 2], ) -> Result<[Specimen; 2], String>

Runs the learning round by using the specified specimen as parents for new generation.

Use this function if you already have some trained networks (represented by Specimen) and would like to use them as a base for further learning

pub fn run(&mut self, finish: Finish) -> Result<[Specimen; 2], String>

Runs the learning round.

Returns two best specimen from the most recent generation. These specimen might be used to resume training by using run_with_parents.

pub fn get_number_of_iterations(&self) -> usize

Returns number of iterations used in recent learning session.