nvidia/cuopt-examples

This repository is a collection of example notebooks for NVIDIA cuOpt, a GPU-accelerated tool for solving optimization problems such as route planning, linear programming, and mixed integer programming. The examples show how to use cuOpt through its Python API, its server API, and in combination with other open-source optimization packages. The examples are organized as Jupyter notebooks, which are interactive documents that mix explanatory text, code, and output in one place. You can run them in a browser after starting a Jupyter server. The easiest way to get started is to pull the official cuOpt Docker container image and run a single command that launches the notebook interface. The repository provides separate commands for two versions of the underlying GPU software, CUDA 12 and CUDA 13. The notebooks have been tested on Google Colab, NVIDIA's own cloud environment called Launchable, and standard local Jupyter setups. Specific system requirements for each example are listed in that example's own README file. A GPU with appropriate drivers is required in all cases. One featured example demonstrates route optimization for autonomous mobile robots moving materials inside a factory. The problem involves assigning pickup and delivery tasks to multiple robots while respecting vehicle capacity and time constraints. The repository is licensed under Apache 2.0 and accepts contributions. Tutorial videos linked from the NVIDIA documentation site accompany many of the examples for people who prefer watching a walkthrough before reading code.