• 📚 Tutorials
• 💻 Software
• 📄 Papers
• 🗂️ Datasets
• 📹 Videos
• 🎪 Conferences Workshops and Hackathons

The tutorials section includes a collection of hands-on guides, jupyter notebooks, and step-by-step instructions for applying Bayesian optimization techniques to various materials science and chemistry problems.

• Acquisition functions in Bayesian Optimization
• Gaussian processes (1/3) - From scratch

• Bayesian machine learning notebooks
• Bayesian-Optimization Notebook Examples

This section features a curated list of software packages and libraries that implement Bayesian optimization algorithms. These tools provide a wide range of functionalities and are designed to be easily integrated into materials science and chemistry workflows.

• Adaptive Experimentation Platform (Ax) is a user-friendly, modular, and actively developed general-purpose Bayesian optimization platform with support for simple and advanced optimization tasks such as noisy, multi-objective, multi-task, multi-fidelity, batch, high-dimensional, linearly constrained, nonlinearly constrained, mixed continuous/discrete/categorical, and contextual Bayesian optimization.
• BoTorch is the backbone that makes up the Ax platform and allows for greater customization and specialized algorithms such as risk-averse Bayesian optimization and constraint active search.
• Dragonfly is an open source python library for scalable Bayesian optimization with multi-objective and multi-fidelity support.
• RayTune offers experiment execution and hyperparameter tuning at any scale with many supported search algorithms and trial schedulers under a common interface.
• Aspuru-Guzik Group
  • Atlas is a Python package that offers Bayesian optimization tailored towards real-world experimental science problems: mixed parameters, multi-objective, noisy, constrained, multi-fidelity, and meta-learning optimization along with search space expansion/contraction. [WIP]
  • Chimera is a hierarchy-based multi-objective optimization scalarizing function.
  • Gryffin enables Bayesian optimization of continuous and categorical variables with support for physicochemical descriptors and batch optimization.
  • Gemini is a scalable multi-fidelity Bayesian optimization technique and is supported by Gryffin.
  • Golem is an algorithm that helps identify optimal solutions that are robust to input uncertainty (i.e., robust optimization).
  • Phoenics is a linear-scaling Bayesian optimization algorithm with support for batch and periodic parameter optimization.
  • Anubis is a Bayesian optimization algorithm that models unknown feasibility constraints and incorporates it into the acquisition function
• BoFire is a Bayesian Optimization Framework Intended for Real Experiments (under development) with support for advanced optimization tasks such as mixed variables, multiple objectives, and generic constraints.
• NIMS-OS is a Python package (+GUI) for workflow orchestration and multi-objective optimization software that supports BLOX, PDC, random exploration, and a multi-objective variant of PHYSBO.
• SLAMD - A web app leveraging data-driven design for cementitious materials via a digital lab twin, complemented by a (materials-agnostic) AI optimization feature. Features UI to interactively explore design spaces. The web app uses Python and Javascript.
• Summit is a set of tools for optimizing chemical processes with a wide variety of design of experiments (DoE) and adaptive design methods along with benchmarks.
• GPax is a Python package for physics-based Gaussian processes (GPs) built on top of NumPyro and JAX that take advantage of prior physical knowledge and different data modalities for active learning and Bayesian optimization. It supports "deep kernel learning", structured probabilistic mean functions, hypothesis learning workflows, multitask, multifidelity, heteroscedastic, and vector BO and emphasizes user friendliness.
• Bayesian Back End (BayBE) is a open-source toolbox by Merck KGaA for Bayesian optimization, featuring custom encodings, chemical knowledge integration, hybrid spaces, transfer learning, simulation tools, and robust, serializable code for real-world experimental campaigns.
• Bayesian-Optimization Pure Python implementation of bayesian global optimization with gaussian processes.
• GPFLow GPflow is a package for building Gaussian process models in python, using TensorFlow.

• Matterhorn Studios A platform for hosting Bayesian optimization algorithms
• ChemOS
• The Citrine Platform
• Omni Advanced Agent A chat based platform incorporating Bayesian Optmization tools
• IBM Accelerated Discovery
  • RXN for Chemistry
  • Simulation Toolkit For Scientific Discovery (ST4SD)
  • Generative Toolkit for Scientific Discovery
  • Deep Search (semantic extraction from documents)

The datasets section provides links to publicly available datasets that can be used for benchmarking and evaluating Bayesian optimization algorithms in the context of materials science and chemistry. These datasets should span various domains, such as catalysis, battery materials, and polymers.

• Foundry-ML A collection of "ML-ready" datasets, with many possibly useful for Bayesian optimization

This section consists of a list of seminal and state-of-the-art research papers that showcase the application of Bayesian optimization in materials science and chemistry.

• Jones, D.R., Schonlau, M. & Welch, W.J. Efficient Global Optimization of Expensive Black-Box Functions. Journal of Global Optimization 13, 455–492 (1998). link open link

• Coding gaussian process regressors FROM SCRATCH in python Implement a Gaussian process regressor with squared exponential kernel in Python using numpy only and code several interactive plots to visualize it.

This section highlights upcoming and past conferences, workshops, and symposia that focus on the intersection of Bayesian optimization, materials science, and chemistry. These events provide a platform for researchers and practitioners to share their latest findings, exchange ideas, and foster collaborations.

Awesome Design of Experiments (DOE)

Contributions of any kind welcome, just follow the guidelines!

Thanks goes to these contributors!

Thank you also to Sterling Baird, who started collecting many of the software links included here.