idinn 0.2.0.post1

Inventory dynamics–informed neural networks and other related controllers for solving single-sourcing and dual-sourcing problems.

idinn: Inventory-Dynamics Control with Neural Networks

idinn implements inventory dynamics–informed neural network and other related controllers for solving single-sourcing and dual-sourcing problems. Neural network controllers and inventory dynamics are implemented into customizable objects using PyTorch as backend to enable users to find the optimal controllers for the user-specified inventory systems.

Demo

For a quick demo, you can run our Streamlit app. The app allows you to interactively train and evaluate neural controllers for user-specified dual-sourcing systems. Alternatively, you may use our notebook in Colab.

Installation

The package can be installed from PyPI. To do that, run

pip install idinn

Or, if you want to inspect the source code and edit locally, run

git clone https://gitlab.com/ComputationalScience/idinn.git
cd idinn
pip install -e .

Example Usage

import torch
from idinn.sourcing_model import SingleSourcingModel
from idinn.single_controller import SingleSourcingNeuralController
from idinn.demand import UniformDemand

# Initialize the sourcing model and the neural controller
sourcing_model = SingleSourcingModel(
    lead_time=0,
    holding_cost=5,
    shortage_cost=495,
    batch_size=32,
    init_inventory=10,
    demand_generator=UniformDemand(low=1, high=4),
)
controller = SingleSourcingNeuralController(
    hidden_layers=[2],
    activation=torch.nn.CELU(alpha=1)
)
# Train the neural controller
controller.fit(
    sourcing_model=sourcing_model,
    sourcing_periods=50,
    validation_sourcing_periods=1000,
    epochs=5000,
    seed=1,
)
# Simulate and plot the results
controller.plot(sourcing_model=sourcing_model, sourcing_periods=100)
# Calculate the optimal order quantity for applications
controller.predict(current_inventory=10)

Documentation

See the official documentation for more information.

Contribution

We welcome contributions of all kinds! To get started, please see our contribution guide.

Papers using idinn

• Böttcher, Lucas, Thomas Asikis, and Ioannis Fragkos. "Control of dual-sourcing inventory systems using recurrent neural networks." INFORMS Journal on Computing 35.6 (2023): 1308-1328.
• Li, Jiawei, Thomas Asikis, Ioannis Fragkos, and Böttcher, Lucas. "idinn: A Python package for inventory-dynamics control with neural networks." Journal of Open Source Software 10.112 (2025): 8508.

@article{bottcher2023control,
  title={Control of dual-sourcing inventory systems using recurrent neural networks},
  author={B{\"o}ttcher, Lucas and Asikis, Thomas and Fragkos, Ioannis},
  journal={INFORMS Journal on Computing},
  volume={35},
  number={6},
  pages={1308--1328},
  year={2023}
}

@article{li2025, 
title = {idinn: A {P}ython package for inventory-dynamics control with neural networks}, 
author = {Li, Jiawei and Asikis, Thomas and Fragkos, Ioannis and B{\"o}ttcher, Lucas}, 
journal = {Journal of Open Source Software},
volume = {10}, 
number = {112}, 
pages = {8508}, 
year = {2025}
}

Contributors

• Jiawei Li
• Thomas Asikis
• Ioannis Fragkos
• Lucas Böttcher