Struct Learner 

pub struct Learner {
    pub weights: [f64; 5],
    pub epoch: u64,
    pub draws: u64,
    pub shortfall: f64,
}

A training run’s state: what a checkpoint holds.

Fields

weights: [f64; 5]

The policy’s weights.

epoch: u64

Completed epochs.

draws: u64

How many training draws have been taken from TRAIN_SEED: the generator’s position.

shortfall: f64

The sum of the completed epochs’ shortfalls (environment::evaluate).

Implementations

impl Learner

pub const fn new() -> Self

A fresh run at INITIAL_WEIGHTS.

pub fn value(&self) -> f64

The value to report: the mean shortfall over the completed epochs, or the current policy’s shortfall before the first.

pub fn iterate(&mut self, hyper: Hyper)

One ARS iteration at hyper.

pub fn train_epoch(&mut self, hyper: Hyper) -> f64

One epoch: ITERATIONS ARS iterations at hyper, then the evaluation. Returns the run’s value so far (Self::value).

pub fn checkpoint(&self) -> String

The checkpoint text: ars2, the epoch, the draw position, the shortfall sum’s bits and each weight’s bits, in hexadecimal and separated by :, so a restored run continues bit for bit.

pub fn from_checkpoint(text: &str) -> Option<Self>

Parses Self::checkpoint; None for any other text, a non-finite weight, a shortfall sum outside [0, 500 × epoch] (each epoch’s shortfall is in [0, 500]), an epoch beyond 2³², or a draw position that is not the epoch’s.

Trait Implementations

impl Clone for Learner

fn clone(&self) -> Learner

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Copy for Learner

impl Debug for Learner

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Default for Learner

fn default() -> Self

Returns the “default value” for a type.

impl PartialEq for Learner

fn eq(&self, other: &Learner) -> bool

Equality operator ==.

fn ne(&self, other: &Rhs) -> bool

Inequality operator !=.

impl StructuralPartialEq for Learner