jobshoplab.env package

Subpackages

Submodules

jobshoplab.env.env module

class DependencyBuilder[source]

Bases: object

A builder class for managing dependencies in the JobShopLab environment.

This class handles initialization and configuration of various components needed for the JobShopLab environment, including logging, random seeds, and factory objects for observations, rewards, rendering, and state simulation.

_config

Configuration object containing environment settings

_logger

Logger instance for this class

loglevel(loglevel

int | str | None) -> int | str: Gets loglevel from config or arguments

logger(loglevel

int | str) -> tuple[any, int | str]: Creates and returns a logger instance

seed(seed

int | None) -> int | None: Sets random seeds for reproducibility

compiler(compiler

Compiler | None) -> Compiler: Returns compiler instance, creates new one if None provided

lower_bound(instance_config

InstanceConfig) -> int: Calculates lower bound for given instance

max_allowed_time(instance_config

InstanceConfig) -> int: Gets maximum allowed time for instance

num_operations(instance_config

InstanceConfig) -> int: Returns total number of operations in instance

observation_factory(observation_factory

any, log_level: int | str, instance: InstanceConfig) -> any: Builds observation factory instance

reward_factory(reward_factory

any, log_level: int | str, instance: InstanceConfig, max_allowed_time: int) -> any: Builds reward factory instance

render_backend(render_backend

any, log_level: int | str, instance: InstanceConfig) -> any: Builds render backend instance

action_factory(action_factory

any, log_level: int | str, instance: InstanceConfig) -> any: Builds action_factory instance

state_simulator(middleware

any, log_level: int | str, instance: InstanceConfig, action_factory: any, observation_factory: any) -> any: Builds state simulator instance

Private Methods:
_to_lowercase(string: str) -> str:

Converts string to lowercase

_config_args_getter(kwd_list: list[str]) -> dict:

Gets configuration arguments from keyword list

_get_args(log_level: str | int | None, instance: InstanceConfig, additional_args_kwd: list[str], additional_args: dict) -> dict:

Builds arguments dictionary for factory creation

_get_instance_from_config(conf_obj: str) -> str:

Gets instance name from configuration

_build_factory(module: any, env_arg: any, config_name: str, additional_args: dict, log_level: int | str, instance: InstanceConfig) -> any:

Generic factory builder method

__init__(config, loglevel=None)[source]
Parameters:
config : __SPHINX_IMMATERIAL_TYPE_VAR__V_Config | None

loglevel : int | str | None

Return type:

None

config(_config)[source]
Parameters:
_config : __SPHINX_IMMATERIAL_TYPE_VAR__V_Config | None

Return type:

__SPHINX_IMMATERIAL_TYPE_VAR__V_Config

loglevel(loglevel)[source]
Parameters:
loglevel : int | str | None

Return type:

int | str

logger(loglevel)[source]
Parameters:
loglevel : int | str

Return type:

Logger

seed(seed)[source]
Parameters:
seed : int | None

Return type:

int | None

compiler(compiler)[source]
Parameters:
compiler : Compiler | None

Return type:

Compiler

lower_bound(instance_config)[source]
Parameters:
instance_config : InstanceConfig

Return type:

int

max_allowed_time(instance_config)[source]
Parameters:
instance_config : InstanceConfig

Return type:

int

num_operations(instance_config)[source]
Parameters:
instance_config : InstanceConfig

Return type:

int

observation_factory(observation_factory, log_level, instance)[source]
Parameters:
observation_factory : any

log_level : int | str

instance : InstanceConfig

Return type:

any

reward_factory(reward_factory, log_level, instance, max_allowed_time)[source]
Parameters:
reward_factory : any

log_level : int | str

instance : InstanceConfig

max_allowed_time : int

Return type:

any

render_backend(render_backend, log_level, instance)[source]
Parameters:
render_backend : any

log_level : int | str

instance : InstanceConfig

Return type:

any

action_factory(action_factory, log_level, instance)[source]
Parameters:
action_factory : any

log_level : int | str

instance : InstanceConfig

Return type:

any

state_simulator(middleware, log_level, instance, action_factory, observation_factory)[source]
Parameters:
middleware : any

log_level : int | str

instance : InstanceConfig

action_factory : any

observation_factory : any

Return type:

any

class JobShopLabEnv[source]

Bases: Env

JobShopLabEnv is an environment for job shop scheduling problems using the OpenAI Gym interface.

The environment simulates a job shop scheduling problem where jobs need to be processed on machines in a specific order. The goal is to minimize the makespan (total completion time) while respecting all constraints.

config (Config | None): Configuration object containing environment parameters. Defaults to None. seed (int | None): Random seed for reproducibility. Defaults to None. compiler (Compiler | None): Instance compiler to parse problem instances. Defaults to None. observation_factory (observations.ObservationFactory | None): ActionFactory for creating observations. Defaults to None. reward_factory (rewards.RewardFactory | None): ActionFactory for calculating rewards. Defaults to None. middleware (middleware_collection.Middleware | None): Middleware for state transitions. Defaults to None. action_factory (action_factory_collection.ActionFactory | None): ActionFactory for actions. Defaults to None. render_backend (Callable | None): Function for rendering the environment. Defaults to None. loglevel (int | str | None): Logging level. Defaults to None.

logger (Logger): Logger instance for the environment config (Config): Configuration object loglevel (int|str): Current logging level seed (int): Random seed used instance (InstanceConfig): Problem instance configuration lower_bound (int): Lower bound on makespan max_allowed_time (int): Maximum allowed timesteps num_operations (int): Total number of operations reward_factory (rewards.RewardFactory): ActionFactory for calculating rewards render_backend (Callable): Function for rendering state_simulator (middleware_collection.Middleware): Middleware for state transitions init_state (StateMachineResult): Initial state observation_space (gym.Space): Space of possible observations action_space (gym.Space): Space of possible actions state (StateMachineResult): Current state history (tuple[StateMachineResult]): History of states truncated (bool): Whether episode was truncated terminated (bool): Whether episode terminated naturally done (bool): Whether episode is done (terminated or truncated) current_observation (dict): Current observation

step(action): Takes an action and returns next observation, reward, done flags and info reset(seed): Resets environment to initial state with optional new seed render(): Renders current state of environment _is_done(): Checks if episode should terminate naturally _is_truncated(): Checks if episode should be truncated _get_info(): Returns info dict about current state _gen_render_metadata(): Generates metadata for rendering

observation (dict): Current observation of environment state reward (float): Reward from last action terminated (bool): Whether episode terminated naturally truncated (bool): Whether episode was truncated info (dict): Additional information about current state

__init__(config=None, seed=None, compiler=None, observation_factory=None, reward_factory=None, middleware=None, action_factory=None, render_backend=None, loglevel=None)[source]

Initialize JobShopLabEnv environment.

This class represents a Job Shop scheduling environment for reinforcement learning. It inherits from the base environment class and sets up the required components for simulation, observation generation, reward calculation and visualization.

Parameters:
config : Config, optional

Configuration object containing environment parameters. Defaults to None.

seed : int, optional

Random seed for reproducibility. Defaults to None.

compiler : Compiler, optional

Compiler instance for processing job shop problems. Defaults to None.

observation_factory : ObservationFactory, optional

ActionFactory class for creating observations. Defaults to None.

reward_factory : RewardFactory, optional

ActionFactory class for calculating rewards. Defaults to None.

middleware : Middleware, optional

Middleware component for pre/post processing. Defaults to None.

action_factory : ActionFactory, optional

ActionFactory component for action processing. Defaults to None.

render_backend : Callable, optional

Backend function for visualization. Defaults to None.

loglevel : int | str, optional

Logging level for the environment. Defaults to None.

init_args

Dictionary storing initialization arguments

Type:

dict

current_observation

Latest observation from the environment

Type:

object

step(action)[source]

Take a step in the environment.

Parameters:
action

The action to take.

Returns:

The observation after taking the action. reward: The reward after taking the action. terminated (bool): True if the episode is terminated, False otherwise. truncated (bool): True if the episode is truncated, False otherwise. info: Additional information.

Return type:

observation

reset(seed=None)[source]

Reset the environment to its initial state.

This method initializes or reinitializes all the environment components including: - Configuration and logging setup - Instance compilation - Static instance-based values - Observation and reward factories - State machine and action_factory - Initial state and observation

Parameters:
seed : int | None

Random seed to use for this reset. Defaults to None.

Returns:

A tuple containing:

  • Initial observation [dict space]

  • Empty info dictionary

Return type:

tuple[np.ndarray, dict]

get_config()[source]

Get the current environment configuration.

Returns:

The configuration object for this environment

Return type:

Config

render(mode='normal')[source]

Render the current state of the environment.

Visualizes the current state using one of several available rendering backends.

Parameters:
mode : str

The rendering mode to use - “normal”: Default rendering to console/terminal - “dashboard”: Web-based dashboard visualization - “simulation”: (Not implemented) Simulation-based visualization

Returns:

The rendering is displayed but not returned

Return type:

None

jobshoplab.env.env_wrapper module

Module contents