madupite.createTransitionProbabilityTensor#
- madupite.createTransitionProbabilityTensor(*, comm: int = 0, name: str = '', numStates: int, numActions: int, func: collections.abc.Callable[[int, int], tuple[collections.abc.Sequence[float], collections.abc.Sequence[int]]], preallocation: madupite._madupite_impl.MatrixPreallocation = <madupite._madupite_impl.MatrixPreallocation object at 0x7fcc2cc10ed0>) madupite._madupite_impl.Matrix#
Creates a transition probability tensor for a Markov Decision Process (MDP).
This function constructs a madupite.Matrix that represents the probabilities of transitioning from one state to another, given a specific action.
Parameters#
- commint, optional
The communicator identifier for parallel computation.
- namestr
The name of the transition probability tensor, used for identification within the MDP environment.
- numStatesint
The total number of states in the Markov Decision Process.
- numActionsint
The total number of possible actions in the Markov Decision Process.
- funcCallable[[int, int], Tuple[Sequence[float], Sequence[int]]]
A callable function that takes in two arguments: a state index and an action index. It returns a tuple containing:
A list of transition probabilities to other states.
A list of corresponding state indices for these probabilities.
- preallocationmadupite.MatrixPreallocation, optional
A preallocation structure that defines how the matrix memory is allocated. Defaults to an instance of MatrixPreallocation.
Returns#
- madupite.Matrix
A matrix object representing the transition probability tensor.