Gymnasium-Compatible Inventory Management Environments & Benchmarks (or-gym-inventory)

This repository provides or_gym_inventory, an installable Python package containing implementations of classic inventory management environments. These environments are adapted from the original OR-Gym library and updated for compatibility with the Gymnasium API (the successor to OpenAI Gym).

The package also includes comprehensive benchmarking scripts (located in the examples/ directory of the source repository) to compare various heuristic, optimization-inspired, and Reinforcement Learning (RL) policies on these environments.

Environments Included:

1. Newsvendor (or_gym_inventory.newsvendor): Multi-period newsvendor problem with lead times and stochastic Poisson demand (based on Zipkin (2000, 2008), and Balaji et al. 2019, https://arxiv.org/abs/1911.10641). Class: NewsvendorEnv.
2. Inventory Management (or_gym_inventory.inventory_management): Multi-period, multi-echelon inventory system for a single product (based on Paul Glasserman and Sridhar Tayur (1995) and D. Hubbs (2020)). Includes InvManagementBacklogEnv and InvManagementLostSalesEnv.
3. Network Inventory Management (or_gym_inventory.network_management): Multi-period, multi-node inventory system with a network structure (factories, distributors, retailers, markets). Includes NetInvMgmtBacklogEnv and NetInvMgmtLostSalesEnv (based on Paul Glasserman and Sridhar Tayur (1995) and Perez et al. (2021)).

Features

• Installable Package: Easily install using pip (pip install or_gym_inventory).
• Gymnasium Compatible: Environments adhere to the modern Gymnasium API standard (reset returns obs, info, step returns obs, reward, terminated, truncated, info).
• Three Core Environments: Covers single-item, multi-echelon, and network inventory problems.
• Backlog & Lost Sales Variants: Specific environment classes (*BacklogEnv, *LostSalesEnv) implement these dynamics.
• Comprehensive Benchmarking Examples: Includes dedicated scripts (examples/benchmark_*.py in the source repo) comparing various agents:
  • Baselines: Random Agent.
  • Heuristics: Relevant heuristics adapted for each environment type (e.g., Order-Up-To, Classic Newsvendor, (s,S) for Newsvendor; Base Stock, Constant Order for multi-echelon/network).
  • Stable Baselines3 Agents: PPO, SAC, TD3, A2C, DDPG examples.
  • Ray RLlib Agents: PPO, SAC examples.
• Detailed Reporting (from Benchmarks): Benchmark examples generate:
  • Summary tables (CSV).
  • Raw results per episode (CSV).
  • Detailed step-by-step data (optional, JSON Lines).
  • Comparison plots (PNG).

Installation

You can install the core package using pip:

pip install or_gym_inventory