Struct ratel_bandit::OptimisticAgent

pub struct OptimisticAgent<T> { /* private fields */ }

Agent that follows the Optimistic Algorithm.

Always chooses the arm with the highest confidence bound.

Implementations

impl<T> OptimisticAgent<T>

pub fn new( q_init: Vec<f64>, c: f64, stepper: Box<dyn Stepper> ) -> OptimisticAgent<T>

Initializes a new Optimistic agent.

Trait Implementations

impl<T: ToPrimitive> Agent<T> for OptimisticAgent<T>

fn action(&self) -> usize

The action chosen by the Agent. The agent chooses the action with the highest confidence bound.

fn arms(&self) -> usize

The number of arms in the Bandit the Agent is playing.

fn current_estimate(&self, arm: usize) -> f64

The Agent’s current estimate of the value of a Bandit’s arm.

fn reset(&mut self, q_init: &[f64])

Reset the Agent’s history and give it a new initial guess of the Bandit’s arm values.

fn step(&mut self, arm: usize, reward: T)

Update the Agent’s totals and estimate of a Bandit arm based on a given reward.