Skip to content

schola.core.spaces.discrete.MultiDiscreteSpace

Class Definition

class schola.core.spaces.discrete.MultiDiscreteSpace(nvec)

Bases: MultiDiscrete, UnrealSpace

A Space representing a vector of discrete values.

See also:

  • gymnasium.spaces.MultiDiscrete – The gym space object that this class is analogous to
  • proto_spaces.MultiDiscreteSpace – The protobuf representation of this space

Parameters

nvec

Type: List[int]
The number of discrete values in each dimension of the space.

Attributes

nvec

Type: List[int]
The number of discrete values in each dimension of the space.

is_np_flattenable

Checks whether this space can be flattened to aspaces.Box.

np_random

Lazily seed the PRNG since this is expensive and only needed if sampling from this space.

shape

Has stricter type thangym.Space - never None.

proto_space

Alias ofDiscreteSpace

Methods

__init__

__init__(nvec)

Constructor of MultiDiscrete space.

The argument nvec will determine the number of values each categorical variable can take. If start is provided, it will define the minimal values corresponding to each categorical variable.

Parameters:

  • nvec (List[int]) – vector of counts of each categorical variable. This will usually be a list of integers. However, you may also pass a more complicated numpy array if you’d like the space to have several axes
  • dtype – This should be some kind of integer type
  • seed – Optionally, you can use this argument to seed the RNG that is used to sample from the space
  • start – Optionally, the starting value the element of each class will take (defaults to 0)

contains

contains(x)

Return boolean specifying if x is a valid member of this space.

fill_proto

fill_proto(msg, values)

Convert a python representation of point in this space to a protobuf message. Mutates msg with the result.

Parameters:

  • msg (proto_points.FundamentalPoint) – The protobuf message to fill
  • value (Any) – The pythonic representation of the point
  • values (ndarray)

Return type: None

from_jsonable

from_jsonable(sample_n)

Convert a JSONable data type to a batch of samples from this space.

from_proto

@classmethod
from_proto(message)

Create a Space Object from a protobuf representation.

Parameters:

  • message (proto_space) – The protobuf message to convert

Returns: The Space subclass created from the protobuf message

Return type: UnrealSpace

is_empty_definition

@classmethod
is_empty_definition(message)

Returns True iff this space has magnitude 0.

Parameters:

  • message (proto_space) – The protobuf message to check for emptiness

Returns: True iff the space is empty.

Return type: bool

merge

@classmethod
merge(*spaces)

Merge multiple DiscreteSpaces into a single space.

Parameters:

  • *spaces (List[Union[DiscreteSpace, MultiDiscreteSpace]]) – The spaces to merge

Returns: The merged space.

Return type: MultiDiscreteSpace

Raises: TypeError – If any of the spaces are not Discrete or MultiDiscrete.

See also: merge_discrete_like_spaces – Merge multiple Discrete or MultiDiscrete spaces into a single MultiDiscrete space.

process_data

process_data(msg)

Convert a protobuf message corresponding to a point in this space to a pythonic representation.

Parameters:

  • msg (proto_points.FundamentalPoint) – The protobuf message to convert

Returns: The pythonic representation of the point.

Return type: np.ndarray

sample

sample(mask=None)

Generates a single random sample this space.

seed

seed(seed=None)

Seed the PRNG of this space and possibly the PRNGs of subspaces.

to_jsonable

to_jsonable(sample_n)

Convert a batch of samples from this space to a JSONable data type.

to_normalized

to_normalized()

Returns a normalized version of the space. Is a noop if a space subclass does not implement to_normalized.

Returns: The normalized space.

Return type: UnrealSpace