Package ai

Expand source code
from . import environments, simulators, utils, rl, search


__all__ = ["environments", "simulators", "utils", "rl", "search"]
__version__ = "0.0.9"


def _fix_start_method():
    from multiprocessing import set_start_method, get_start_method
    try:
        set_start_method("spawn")
    except RuntimeError:
        if get_start_method().lower() != "spawn":
            raise RuntimeError("Start method is not spawn")


def _fix_pdoc():
    import queue
    import os
    from types import ModuleType

    root = os.path.dirname(__file__)
    modules = queue.Queue()
    for module in __all__:
        modules.put_nowait((module, ))

    while not modules.empty():
        module_name = modules.get_nowait()
        module = eval(".".join(module_name))
        if not isinstance(module, ModuleType):
            continue
        if "__pdoc__" not in dir(module):
            module.__pdoc__ = {}
        if "__all__" not in dir(module):
            module.__all__ = []

        for obj in module.__all__:
            obj = eval(".".join(module_name) + f".{obj}")
            if not isinstance(obj, ModuleType):
                obj.__module__ = "ai." + ".".join(module_name)

        for submodule in os.listdir(os.path.join(root, *module_name)):
            submodule_path = os.path.join(root, *module_name, submodule)
            if submodule.startswith("_"):
                continue
            if os.path.isdir(submodule_path) and "__init__.py" in os.listdir(submodule_path):
                module.__pdoc__[submodule] = submodule in module.__all__
                if submodule in module.__all__:
                    modules.put_nowait(module_name + (submodule, ))
            elif submodule.endswith(".py"):
                submodule = submodule[:-3]
                module.__pdoc__[submodule] = submodule in module.__all__

_fix_start_method()
_fix_pdoc()

Sub-modules

ai.environments

Module containing the abstract definition of an environment, as well as implementations of it.

ai.rl

Module containing implementations of reinforcement learning (rl) algorithms.

ai.search

Search/planning algorithms.

ai.simulators

Module containing the abstract definition of a simulator, as well as several implementations of it.

ai.utils

Module containing general utility methods, used throughout the library.