pymgrid.utils.space.ModuleSpace#

class pymgrid.utils.space.ModuleSpace(unnormalized_low, unnormalized_high, shape=None, dtype=<class 'numpy.float64'>, seed=None)[source]#

A space with both normalized and unnormalized versions.

Both normalized and unormalized versions are of type gym.spaces.Box.

This object also handles conversion of values from normalized to unnormalized and vice-versa.

Parameters#

unnormalized_lowfloat or array-like: Lower bound of the space.
unnormalized_highfloat or array-like: Upper bound of the space.
shapetuple or None, default None: Shape of the space. If None, shape is inferred from unnormalized_low and unnormalized_high.
dtypenp.dtype, default np.float64: dtype of the action space.
seedint or None, default None: Random seed.

Warning

This parameter will be ignored if earlier versions of gym are installed.

Methods

`contains`(x)	Check if x is a valid member of the space.
`denormalize`(val)
`from_jsonable`(sample_n)	Convert a JSONable data type to a batch of samples from this space.
`normalize`(val)
`sample`([normalized])	Randomly sample an element of this space.
`seed`([seed])	Seed the PRNG of this space and possibly the PRNGs of subspaces.
`to_jsonable`(sample_n)	Convert a batch of samples from this space to a JSONable data type.

Attributes

`is_np_flattenable`	Checks whether this space can be flattened to a `spaces.Box`.
`normalized`
`np_random`	Lazily seed the PRNG since this is expensive and only needed if sampling from this space.
`shape`	Return the shape of the space as an immutable property.
`unnormalized`