`evorl.ec.optimizers.openes`¶

Module Contents¶

Classes¶

`OpenES`	OpenAI ES.
`OpenESNoiseTable`	OpenAI ES with noise table.
`OpenESNoiseTableState`	State of the OpenES with noise table.
`OpenESState`	State of the OpenES.

Functions¶

`compute_centered_ranks`	Get centered ranks in [-0.5, 0.5].
`compute_ranks`	Get ranks in [0, len(x)-1].

API¶

class evorl.ec.optimizers.openes.OpenES[source]¶

Bases: evorl.ec.optimizers.ec_optimizer.EvoOptimizer

OpenAI ES.

ask(state: evorl.ec.optimizers.ec_optimizer.ECState) → tuple[chex.ArrayTree, evorl.ec.optimizers.ec_optimizer.ECState][source]¶: Generate new candidate solutions.

fitness_shaping_fn: collections.abc.Callable[[chex.Array], chex.Array]¶: ‘pytree_field(…)’

init(mean: evorl.types.Params, key: chex.PRNGKey) → evorl.ec.optimizers.ec_optimizer.ECState[source]¶

lr_schedule: evorl.ec.optimizers.utils.ExponentialScheduleSpec¶: None

mirror_sampling: bool¶: True

noise_std_schedule: evorl.ec.optimizers.utils.ExponentialScheduleSpec¶: None

optimizer: optax.GradientTransformation¶: ‘pytree_field(…)’

optimizer_name: str¶: ‘adam’

pop_size: int¶: None

tell(state: evorl.ec.optimizers.ec_optimizer.ECState, fitnesses: chex.Array) → tuple[evorl.types.PyTreeDict, evorl.ec.optimizers.openes.OpenESState][source]¶: Update the optimizer state based on the fitnesses of the candidate solutions.

weight_decay: float | None¶: None

class evorl.ec.optimizers.openes.OpenESNoiseTable[source]¶

Bases: evorl.ec.optimizers.ec_optimizer.EvoOptimizer

OpenAI ES with noise table.

ask(state: evorl.ec.optimizers.ec_optimizer.ECState) → tuple[chex.ArrayTree, evorl.ec.optimizers.ec_optimizer.ECState][source]¶: Generate new candidate solutions.

fitness_shaping_fn: collections.abc.Callable[[chex.Array], chex.Array]¶: ‘pytree_field(…)’

init(mean: evorl.types.Params, key: chex.PRNGKey) → evorl.ec.optimizers.ec_optimizer.ECState[source]¶

lr_schedule: evorl.ec.optimizers.utils.ExponentialScheduleSpec¶: None

mirror_sampling: bool¶: True

noise_std_schedule: evorl.ec.optimizers.utils.ExponentialScheduleSpec¶: None

noise_table_size: int¶: None

optimizer: optax.GradientTransformation¶: ‘pytree_field(…)’

optimizer_name: str¶: ‘adam’

pop_size: int¶: None

tell(state: evorl.ec.optimizers.ec_optimizer.ECState, fitnesses: chex.Array) → tuple[evorl.types.PyTreeDict, evorl.ec.optimizers.openes.OpenESState][source]¶: Update the optimizer state based on the fitnesses of the candidate solutions.

weight_decay: float | None¶: None

class evorl.ec.optimizers.openes.OpenESNoiseTableState[source]¶

Bases: evorl.types.PyTreeData

State of the OpenES with noise table.

key: chex.PRNGKey¶: None

mean: chex.ArrayTree¶: None

noise: None | chex.ArrayTree¶: None

noise_std: chex.Array¶: None

noise_table: chex.ArrayTree¶: None

opt_state: optax.OptState¶: None

class evorl.ec.optimizers.openes.OpenESState[source]¶

Bases: evorl.types.PyTreeData

State of the OpenES.

key: chex.PRNGKey¶: None

mean: chex.ArrayTree¶: None

noise: None | chex.ArrayTree¶: None

noise_std: chex.Array¶: None

opt_state: optax.OptState¶: None

evorl.ec.optimizers.openes.compute_centered_ranks(x)[source]¶: Get centered ranks in [-0.5, 0.5].

evorl.ec.optimizers.openes.compute_ranks(x)[source]¶

Get ranks in [0, len(x)-1].

This is different from scipy.stats.rankdata, which returns ranks in [1, len(x)].

evorl.ec.optimizers.openes¶

Module Contents¶

Classes¶

Functions¶

API¶

`evorl.ec.optimizers.openes`¶