spaces.LogGridAxis

spaces.LogGridAxis(low, high, n, shape=None)

Axis for systematic logarithmic grid sampling over parameter bounds.

Generates logarithmically spaced values between low and high bounds using deterministic grid sampling. Useful for parameters that span multiple orders of magnitude (e.g., learning rates, regularization coefficients).

Parameters

Name	Type	Description	Default
low	float	Lower bound for sampling (must be positive).	required
high	float	Upper bound for sampling (must be positive).	required
n	int	Number of grid points to generate.	required

Raises

Name	Type	Description
	ValueError	If n <= 0, low >= high, or if low or high are not positive.

Examples

>>> log_grid = LogGridAxis(0.001, 1.0, 5)
>>> values = log_grid.generate_values()
>>> print(values)  # [0.001, 0.00562, 0.0316, 0.178, 1.0] (approximately)

Attributes

Name	Description
size	Number of grid points.

Methods

Name	Description
generate_values	Generate logarithmically spaced grid values.

generate_values

spaces.LogGridAxis.generate_values(key=None)

Generate logarithmically spaced grid values.

Parameters

Name	Type	Description	Default
key	jax.random.PRNGKey	Random key. Ignored for deterministic grid sampling.	None

Returns

Name	Type	Description
	jnp.ndarray	Array of logarithmically spaced values from low to high. If shape is specified, values are broadcast to shape (n,) + shape with identical values across additional dimensions.