ALchemist: Active Learning Toolkit for Chemical and Materials Research
ALchemist is a modular Python toolkit that brings active learning and Bayesian optimization to experimental design in chemical and materials research. It is designed for scientists and engineers who want to efficiently explore or optimize high-dimensional variable spaces—using intuitive graphical interfaces, programmatic APIs, or autonomous optimization workflows.
NLR Software Record: SWR-25-102
Full user guide and documentation:
https://nrel.github.io/ALchemist/
Key Features:
- Flexible variable space definition: Real, integer, and categorical variables with bounds or discrete values
- Probabilistic surrogate modeling: Gaussian process regression via BoTorch or scikit-learn backends
- Advanced acquisition strategies: Efficient sampling using qEI, qPI, qUCB, and qNegIntegratedPosteriorVariance
- Modern web interface: React-based UI with FastAPI backend for seamless active learning workflows
- Desktop GUI: CustomTkinter desktop application for offline optimization
- Session management: Save/load optimization sessions with audit logs for reproducibility
- Multiple interfaces: No-code GUI, Python Session API, or REST API for different use cases
- Autonomous optimization: Human-out-of-the-loop operation for real-time process control
- Experiment tracking: CSV logging, reproducible random seeds, and comprehensive audit trails
- Extensibility: Abstract interfaces for models and acquisition functions enable future backend and workflow expansion Architecture:
ALchemist is built on a clean, modular architecture:
- Core Session API: Headless Bayesian optimization engine (
alchemist_core) that powers all interfaces - Desktop Application: CustomTkinter GUI using the Core Session API, designed for human-in-the-loop and offline optimization
- REST API: FastAPI server providing a thin wrapper around the Core Session API for remote access
- Web Application: React UI consuming the REST API, supporting both interactive and autonomous optimization workflows
Session files (JSON format) are fully interoperable between desktop and web applications, enabling seamless workflow transitions.
- Interactive Optimization: Use desktop or web GUI for manual experiment design and human-in-the-loop optimization
- Programmatic Workflows: Import the Session API in Python scripts or Jupyter notebooks for batch processing
- Autonomous Optimization: Use the REST API to integrate ALchemist with automated laboratory equipment for real-time process control
- Remote Monitoring: Web dashboard provides read-only monitoring mode when ALchemist is being remote-controlled
For detailed application examples, see Use Cases in the documentation.
Requirements: Python 3.11 or higher
Recommended (Optional): We recommend using Anaconda to manage Python environments:
conda create -n alchemist-env python=3.11
conda activate alchemist-envBasic Installation:
pip install alchemist-nrelFrom GitHub:
Note: This installs the latest unreleased version. The web application is not pre-built with this method because static build files are not included in the repository.
pip install git+https://github.com/NREL/ALchemist.gitFor advanced installation options, Docker deployment, and development setup, see the Advanced Installation Guide in the documentation.
Web Application:
alchemist-webOpens at http://localhost:8000/app
Desktop Application:
alchemistFor detailed usage instructions, see Getting Started in the documentation.
ALchemist is under active development at NLR as part of the DataHub project within the ChemCatBio consortium.
If you encounter any issues or have questions, please open an issue on GitHub or contact [email protected].
For the latest known issues and troubleshooting tips, see the Issues & Troubleshooting Log.
We appreciate your feedback and bug reports to help improve ALchemist!
This project is licensed under the BSD 3-Clause License. See the LICENSE file for details.