Struct MfReport 

pub struct MfReport {
    pub label: String,
    pub units_spent: u64,
    pub trials: Vec<MfTrial>,
    pub incumbent_true: f64,
    pub curve: Vec<(u64, f64)>,
}

What one search run achieved, in resource units.

Fields

label: String

A short label for the searcher (dehb, random, tpe, …).

units_spent: u64

Total fidelity units spent — the sum over trials of the budget each ran to. This is the axis every comparison here is matched on.

trials: Vec<MfTrial>

Every trial, in trial order.

incumbent_true: f64

The true value of the incumbent — the best configuration this run evaluated at the full budget. f64::INFINITY if it never reached the top fidelity.

curve: Vec<(u64, f64)>

(cumulative units, incumbent true value) after every trial, in trial order. The curve both claims are read off.

Implementations

impl MfReport

pub fn units_to_reach(&self, target: f64) -> Option<u64>

The fewest units at which the incumbent was already at least as good as target — None if the run never reached it.

The budget-efficiency answer: “how much compute did this searcher need to match that quality”.

pub fn evaluations(&self) -> usize

How many trials ran.

pub fn evaluations_at(&self, fidelity: u64) -> usize

How many trials ran at exactly fidelity.

pub fn units_at(&self, fidelity: u64) -> u64

The units spent on trials that ran at exactly fidelity.

Trait Implementations

impl Clone for MfReport

fn clone(&self) -> MfReport

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for MfReport

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

Formats the value using the given formatter.

impl PartialEq for MfReport

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

Equality operator ==.

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

Inequality operator !=.

impl StructuralPartialEq for MfReport