Struct TpeSamplerBuilder

Source

pub struct TpeSamplerBuilder { /* private fields */ }

Expand description

Builder for configuring a TpeSampler.

This builder allows fluent configuration of TPE hyperparameters.

§Examples

use optimizer::TpeSamplerBuilder;

let sampler = TpeSamplerBuilder::new()
    .gamma(0.15)
    .n_startup_trials(20)
    .n_ei_candidates(32)
    .seed(42)
    .build();

Implementations§

Source §

impl TpeSamplerBuilder

Source

pub fn new() -> Self

Creates a new builder with default settings.

Default settings:

gamma: 0.25 (top 25% of trials are considered “good”)
n_startup_trials: 10 (random sampling for first 10 trials)
n_ei_candidates: 24 (evaluate 24 candidates per sample)
kde_bandwidth: None (uses Scott’s rule for automatic bandwidth)
seed: None (use OS-provided entropy)

Source

pub fn gamma(self, gamma: f64) -> Self

Sets the gamma quantile for splitting trials into good/bad groups.

A gamma of 0.25 means the top 25% of trials (by objective value) are considered “good” and used to build the l(x) distribution.

§Arguments

gamma - Quantile value, must be in (0.0, 1.0).

§Panics

Panics if gamma is not in (0.0, 1.0).

§Examples

use optimizer::TpeSamplerBuilder;

let sampler = TpeSamplerBuilder::new()
    .gamma(0.10)  // Use top 10% as "good" trials
    .build();

Source

pub fn n_startup_trials(self, n: usize) -> Self

Sets the number of startup trials before TPE sampling begins.

During the startup phase, the sampler uses uniform random sampling to gather initial data. Once n_startup_trials have completed, TPE-based sampling begins.

§Arguments

n - Number of random trials before TPE kicks in.

§Examples

use optimizer::TpeSamplerBuilder;

let sampler = TpeSamplerBuilder::new()
    .n_startup_trials(20)  // Random sample first 20 trials
    .build();

Source

pub fn n_ei_candidates(self, n: usize) -> Self

Sets the number of EI (Expected Improvement) candidates to evaluate.

When sampling a new point, TPE generates this many candidates from the l(x) distribution and selects the one with the highest l(x)/g(x) ratio.

§Arguments

n - Number of candidates to evaluate per sample.

§Examples

use optimizer::TpeSamplerBuilder;

let sampler = TpeSamplerBuilder::new()
    .n_ei_candidates(48)  // Evaluate more candidates
    .build();

Source

pub fn kde_bandwidth(self, bandwidth: f64) -> Self

Sets a fixed bandwidth for the kernel density estimator.

By default, TPE uses Scott’s rule to automatically select the bandwidth based on the sample data. Use this method to override with a fixed value.

Smaller bandwidths give more localized, peaky distributions. Larger bandwidths give smoother, more spread-out distributions.

§Arguments

bandwidth - The fixed bandwidth (standard deviation) for Gaussian kernels.

§Panics

Panics if bandwidth is not positive.

§Examples

use optimizer::TpeSamplerBuilder;

let sampler = TpeSamplerBuilder::new()
    .kde_bandwidth(0.5)  // Fixed bandwidth of 0.5
    .build();

Source

pub fn seed(self, seed: u64) -> Self

Sets a seed for reproducible sampling.

§Arguments

seed - Seed value for the random number generator.

§Examples

use optimizer::TpeSamplerBuilder;

let sampler = TpeSamplerBuilder::new()
    .seed(42)  // Reproducible results
    .build();

Source

pub fn build(self) -> TpeSampler

Builds the configured TpeSampler.

§Examples

use optimizer::TpeSamplerBuilder;

let sampler = TpeSamplerBuilder::new()
    .gamma(0.15)
    .n_startup_trials(20)
    .n_ei_candidates(32)
    .seed(42)
    .build();