ax.metrics¶

Branin¶

class ax.metrics.branin.BraninMetric(name, param_names, noise_sd=0.0, lower_is_better=None)[source]¶

Bases: ax.metrics.noisy_function.NoisyFunctionMetric

f(x)[source]¶

The deterministic function that produces the metric outcomes.

Return type:	float

class ax.metrics.branin.NegativeBraninMetric(name, param_names, noise_sd=0.0, lower_is_better=None)[source]¶

Bases: ax.metrics.branin.BraninMetric

f(x)[source]¶

The deterministic function that produces the metric outcomes.

Return type:	float

Factorial¶

class ax.metrics.factorial.FactorialMetric(name, coefficients, batch_size=10000, noise_var=0.0)[source]¶

Bases: ax.core.metric.Metric

Metric for testing factorial designs assuming a main effects only logit model.

clone()[source]¶

Create a copy of this Metric.

Return type:	FactorialMetric

fetch_trial_data(trial, **kwargs)[source]¶

Fetch data for one trial.

Return type:	Data

ax.metrics.factorial.evaluation_function(parameterization, coefficients, weight=1.0, batch_size=10000, noise_var=0.0)[source]¶

Return type:	Tuple[float, float]

Hartmann6¶

class ax.metrics.hartmann6.Hartmann6Metric(name, param_names, noise_sd=0.0, lower_is_better=None)[source]¶

Bases: ax.metrics.noisy_function.NoisyFunctionMetric

f(x)[source]¶

The deterministic function that produces the metric outcomes.

Return type:	float

Noisy Functions¶

class ax.metrics.noisy_function.NoisyFunctionMetric(name, param_names, noise_sd=0.0, lower_is_better=None)[source]¶

Bases: ax.core.metric.Metric

A metric defined by a generic deterministic function, with normal noise with mean 0 and mean_sd scale added to the result.

clone()[source]¶

Create a copy of this Metric.

Return type:	NoisyFunctionMetric

f(x)[source]¶

The deterministic function that produces the metric outcomes.

Return type:	float

fetch_trial_data(trial, noisy=True, **kwargs)[source]¶

Fetch data for one trial.

Return type:	Data