Skip to content

Commit

Permalink
defined multi-period env
Browse files Browse the repository at this point in the history
  • Loading branch information
@majoma7
majoma7 committed Aug 15, 2024
1 parent 1b5967d commit 01e942a
Show file tree
Hide file tree
Showing 14 changed files with 1,798 additions and 890 deletions.
39 changes: 36 additions & 3 deletions ddopnew/_modidx.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -578,12 +578,44 @@
'ddopnew/envs/inventory/base.py'),
'ddopnew.envs.inventory.base.BaseInventoryEnv.set_observation_space': ( '21_envs_inventory/base_inventory_env.html#baseinventoryenv.set_observation_space',
'ddopnew/envs/inventory/base.py')},
'ddopnew.envs.inventory.inventory_utils': { 'ddopnew.envs.inventory.inventory_utils.OrderPipeline': ( '21_envs_inventory/inventory_utils.html#orderpipeline',
'ddopnew/envs/inventory/inventory_utils.py'),
'ddopnew.envs.inventory.inventory_utils.OrderPipeline.__init__': ( '21_envs_inventory/inventory_utils.html#orderpipeline.__init__',
'ddopnew/envs/inventory/inventory_utils.py'),
'ddopnew.envs.inventory.inventory_utils.OrderPipeline.check_max_min_mean_lt': ( '21_envs_inventory/inventory_utils.html#orderpipeline.check_max_min_mean_lt',
'ddopnew/envs/inventory/inventory_utils.py'),
'ddopnew.envs.inventory.inventory_utils.OrderPipeline.check_stochasticity': ( '21_envs_inventory/inventory_utils.html#orderpipeline.check_stochasticity',
'ddopnew/envs/inventory/inventory_utils.py'),
'ddopnew.envs.inventory.inventory_utils.OrderPipeline.draw_lead_times': ( '21_envs_inventory/inventory_utils.html#orderpipeline.draw_lead_times',
'ddopnew/envs/inventory/inventory_utils.py'),
'ddopnew.envs.inventory.inventory_utils.OrderPipeline.get_orders_arriving': ( '21_envs_inventory/inventory_utils.html#orderpipeline.get_orders_arriving',
'ddopnew/envs/inventory/inventory_utils.py'),
'ddopnew.envs.inventory.inventory_utils.OrderPipeline.get_pipeline': ( '21_envs_inventory/inventory_utils.html#orderpipeline.get_pipeline',
'ddopnew/envs/inventory/inventory_utils.py'),
'ddopnew.envs.inventory.inventory_utils.OrderPipeline.reset': ( '21_envs_inventory/inventory_utils.html#orderpipeline.reset',
'ddopnew/envs/inventory/inventory_utils.py'),
'ddopnew.envs.inventory.inventory_utils.OrderPipeline.set_param': ( '21_envs_inventory/inventory_utils.html#orderpipeline.set_param',
'ddopnew/envs/inventory/inventory_utils.py'),
'ddopnew.envs.inventory.inventory_utils.OrderPipeline.shape': ( '21_envs_inventory/inventory_utils.html#orderpipeline.shape',
'ddopnew/envs/inventory/inventory_utils.py'),
'ddopnew.envs.inventory.inventory_utils.OrderPipeline.step': ( '21_envs_inventory/inventory_utils.html#orderpipeline.step',
'ddopnew/envs/inventory/inventory_utils.py')},
'ddopnew.envs.inventory.multi_period': { 'ddopnew.envs.inventory.multi_period.MultiPeriodEnv': ( '21_envs_inventory/multi_period_envs.html#multiperiodenv',
'ddopnew/envs/inventory/multi_period.py'),
'ddopnew.envs.inventory.multi_period.MultiPeriodEnv.__init__': ( '21_envs_inventory/multi_period_envs.html#multiperiodenv.__init__',
'ddopnew/envs/inventory/multi_period.py'),
'ddopnew.envs.inventory.multi_period.MultiPeriodEnv.get_observation': ( '21_envs_inventory/multi_period_envs.html#multiperiodenv.get_observation',
'ddopnew/envs/inventory/multi_period.py'),
'ddopnew.envs.inventory.multi_period.MultiPeriodEnv.reset': ( '21_envs_inventory/multi_period_envs.html#multiperiodenv.reset',
'ddopnew/envs/inventory/multi_period.py'),
'ddopnew.envs.inventory.multi_period.MultiPeriodEnv.set_observation_space': ( '21_envs_inventory/multi_period_envs.html#multiperiodenv.set_observation_space',
'ddopnew/envs/inventory/multi_period.py'),
'ddopnew.envs.inventory.multi_period.MultiPeriodEnv.step_': ( '21_envs_inventory/multi_period_envs.html#multiperiodenv.step_',
'ddopnew/envs/inventory/multi_period.py')},
'ddopnew.envs.inventory.single_period': { 'ddopnew.envs.inventory.single_period.NewsvendorEnv': ( '21_envs_inventory/single_period_envs.html#newsvendorenv',
'ddopnew/envs/inventory/single_period.py'),
'ddopnew.envs.inventory.single_period.NewsvendorEnv.__init__': ( '21_envs_inventory/single_period_envs.html#newsvendorenv.__init__',
'ddopnew/envs/inventory/single_period.py'),
'ddopnew.envs.inventory.single_period.NewsvendorEnv.reset': ( '21_envs_inventory/single_period_envs.html#newsvendorenv.reset',
'ddopnew/envs/inventory/single_period.py'),
'ddopnew.envs.inventory.single_period.NewsvendorEnv.step_': ( '21_envs_inventory/single_period_envs.html#newsvendorenv.step_',
'ddopnew/envs/inventory/single_period.py')},
'ddopnew.experiment_functions': { 'ddopnew.experiment_functions.EarlyStoppingHandler': ( '30_experiment_functions/experiment_functions.html#earlystoppinghandler',
Expand Down Expand Up @@ -680,4 +712,5 @@
'ddopnew.utils.Parameter.shape': ('00_utils/utils.html#parameter.shape', 'ddopnew/utils.py'),
'ddopnew.utils.Parameter.size': ('00_utils/utils.html#parameter.size', 'ddopnew/utils.py'),
'ddopnew.utils.check_parameter_types': ('00_utils/utils.html#check_parameter_types', 'ddopnew/utils.py'),
'ddopnew.utils.merge_dictionaries': ('00_utils/utils.html#merge_dictionaries', 'ddopnew/utils.py')}}}
'ddopnew.utils.merge_dictionaries': ('00_utils/utils.html#merge_dictionaries', 'ddopnew/utils.py'),
'ddopnew.utils.set_param': ('00_utils/utils.html#set_param', 'ddopnew/utils.py')}}}
4 changes: 2 additions & 2 deletions ddopnew/envs/base.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -6,7 +6,7 @@
# %% ../../nbs/20_base_env/10_base_env.ipynb 3
import gymnasium as gym
from abc import ABC, abstractmethod
from typing import Union
from typing import Union, List
import numpy as np

from ..utils import MDPInfo, Parameter, set_param
Expand Down Expand Up @@ -50,7 +50,7 @@ def __init__(self,

def set_param(self,
name: str, # name of the parameter (will become the attribute name)
input: Parameter | float | np.ndarray, # input value of the parameter
input: Parameter | int | float | np.ndarray | List | None, # input value of the parameter
shape: tuple = (1,), # shape of the parameter
new: bool = False # whether to create a new parameter or update an existing one
) -> None: #
Expand Down
2 changes: 1 addition & 1 deletion ddopnew/envs/inventory/base.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -5,7 +5,7 @@

# %% ../../../nbs/21_envs_inventory/10_base_inventory_env.ipynb 3
from abc import ABC, abstractmethod
from typing import Union, Tuple
from typing import Union, Tuple, List

from ..base import BaseEnvironment
from ...utils import Parameter, MDPInfo
Expand Down
168 changes: 162 additions & 6 deletions ddopnew/envs/inventory/inventory_utils.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -4,13 +4,18 @@
__all__ = ['OrderPipeline']

# %% ../../../nbs/21_envs_inventory/00_inventory_utils.ipynb 3
import logging


from abc import ABC, abstractmethod
from typing import Union, Tuple
from typing import Union, Tuple, List, Literal

from ..base import BaseEnvironment
from ...utils import Parameter, MDPInfo
from ...dataloaders.base import BaseDataLoader
from ...loss_functions import pinball_loss
from ...utils import set_param, Parameter


import gymnasium as gym

Expand All @@ -28,22 +33,173 @@ class OrderPipeline():
def __init__(self,

num_units: int, # number of units (SKUs)
lead_time_mean: list[object] | None = None, # default is empty list

lead_time_mean: Parameter | np.ndarray | List | int | float, # mean lead time
lead_time_stochasticity: Literal["fixed", "gamma", "normal_absolute", "normal_relative"] = "fixed", # "fixed", "gamma", "normal_absolute", "normal_relative"
lead_time_variance: Parameter | np.ndarray | List | int | float | None = None, # variance of the lead time
max_lead_time: list[object] | None = None, # maximum lead time in case of stochastic lead times
min_lead_time: list[object] | None = 1, # minimum lead time in case of stochastic lead times

) -> None:


self.set_param('num_units', num_units, shape=(1,), new=True)
self.set_param('lead_time_mean', lead_time_mean, shape=(self.num_units[0],), new=True)
self.set_param('lead_time_variance', lead_time_variance, shape=(self.num_units[0],), new=True)
self.lead_time_stochasticity = lead_time_stochasticity
self.check_stochasticity(max_lead_time)

if max_lead_time is None:
self.max_lead_time = lead_time_mean
self.set_param('max_lead_time', lead_time_mean, shape=(self.num_units[0],), new=True)
else:
self.max_lead_time = max_lead_time
self.set_param('max_lead_time', max_lead_time, shape=(self.num_units[0],), new=True)
self.set_param('min_lead_time', min_lead_time, shape=(self.num_units[0],), new=True)

self.check_max_min_mean_lt()

self.pipeline = np.zeros((np.max(self.max_lead_time), num_units))
self.lead_time_realized = np.zeros((np.max(self.max_lead_time), num_units))

def get_pipeline(self) -> np.ndarray:
""" Get the current pipeline """

return self.pipeline

def reset(self) -> None:
""" Reset the pipeline """

self.pipeline = np.zeros((np.max(self.max_lead_time), self.num_units[0]))
self.lead_time_realized = np.zeros((np.max(self.max_lead_time), self.num_units[0]))

self.pipeline = np.zeros((np.max(max_lead_time), num_units))

def step(self,
orders: np.ndarray,
) -> np.ndarray:

""" Add orders to the pipeline and return the orders that are arriving """


# print("beginnig pipeline:")
# print(self.pipeline)
# print("beginnig lead_time_realized:")
# print(self.lead_time_realized)


orders_arriving = self.get_orders_arriving().copy()
lead_times = self.draw_lead_times().copy()

# print("orders_arriving:")
# print(orders_arriving)
# print("new orders:")
# print(orders)
# print("new lead_times:")
# print(lead_times)


self.pipeline = np.roll(self.pipeline, -1, axis=0)
self.lead_time_realized = np.roll(self.lead_time_realized, -1, axis=0)
self.pipeline[-1, :] = 0
self.lead_time_realized[-1, :] = 0

self.pipeline[-1, :] = orders.copy()
self.lead_time_realized[-1, :] = lead_times
self.lead_time_realized -= 1
self.lead_time_realized = np.clip(self.lead_time_realized, 0, None)

# print("ending pipeline:")
# print(self.pipeline)
# print("ending lead_time_realized:")
# print(self.lead_time_realized)

return orders_arriving

def get_orders_arriving(self) -> np.ndarray:

""" Get the orders that are arriving in the current period """

orders_arriving = np.zeros(self.num_units[0])
for i in range(self.num_units[0]):
# check along the pipeline where the lead time is 0

arriving_indices = np.where(self.lead_time_realized[:, i] == 0)[0]

orders_arriving[i] = np.sum(self.pipeline[arriving_indices, i])

self.pipeline[arriving_indices, i] = 0

return orders_arriving

def draw_lead_times(self) -> np.ndarray:
""" Draw lead times for the orders """

if self.lead_time_stochasticity == "fixed":
lead_times = self.lead_time_mean
elif self.lead_time_stochasticity == "gamma":
lead_times = np.random.gamma(self.lead_time_mean, 1, self.num_units[0])
elif self.lead_time_stochasticity == "normal_absolute":
lead_times = np.random.normal(self.lead_time_mean, self.lead_time_variance, self.num_units[0])
elif self.lead_time_stochasticity == "normal_relative":
lead_times = np.random.normal(self.lead_time_mean, self.lead_time_mean * self.lead_time_variance, self.num_units[0])
else:
raise ValueError("Invalid lead time stochasticity")

lead_times = np.clip(lead_times, self.min_lead_time, self.max_lead_time)
lead_times = np.round(lead_times).astype(int)

return lead_times

def check_stochasticity(self, max_lead_time):
""" Check that params for stochastic lead times are set correctly """

# lead time mean to be set in any case (it will be the determinstic lead time if lead_time_stochasticity is fixed)
if self.lead_time_mean is None:
raise ValueError("Lead time mean is not set")
if self.lead_time_stochasticity == "fixed" or self.lead_time_stochasticity == "gamma":
if self.lead_time_variance is not None:
raise ValueError("Lead time variance must be None for fixed lead times (no variance) or gamma lead times (variance is set by the gamma distribution)")
elif self.lead_time_stochasticity == "normal_absolute" or self.lead_time_stochasticity == "normal_relative":
if self.lead_time_variance is None:
raise ValueError("Lead time variance must be set for normal lead times")
else:
raise ValueError("Invalid lead time stochasticity")

if self.lead_time_stochasticity != "fixed":
if max_lead_time is None:
raise ValueError("Max lead time must be set for stochastic lead times")

def check_max_min_mean_lt(self):
if np.any(self.max_lead_time < self.lead_time_mean):
raise ValueError("Max lead time must be greater than or equal to the lead time mean")
if np.any(self.min_lead_time < 1):
raise ValueError("Min lead time must be at least 1")
if np.any(self.min_lead_time > self.lead_time_mean):
raise ValueError("Min lead time must be less than or equal to the lead time mean")
if np.any(self.max_lead_time < self.min_lead_time):
raise ValueError("Max lead time must be greater than or equal to the min lead time")
if np.any(self.max_lead_time < 1):
raise ValueError("Max lead time must be at least 1")
if np.any(self.lead_time_mean < 1):
raise ValueError("Lead time mean must be at least 1")


def set_param(self,
name: str, # name of the parameter (will become the attribute name)
input: Parameter | int | float | np.ndarray | List, # input value of the parameter
shape: tuple = (1,), # shape of the parameter
new: bool = False # whether to create a new parameter or update an existing one
) -> None: #

"""
Set a parameter for the environment. It converts scalar values to numpy arrays and ensures that
environment parameters are either of the Parameter class of Numpy arrays. If new is set to True,
the function will create a new parameter or update an existing one otherwise. If new is set to
False, the function will raise an error if the parameter does not exist.
"""

set_param(self, name, input, shape, new)

@property
def shape(self) -> Tuple:
""" Get the shape of the pipeline """

return self.pipeline.shape

Loading

0 comments on commit 01e942a

Please sign in to comment.