Tutorials

Example optimization problems

The following tutorials explore various optimization problems using Trieste.

• Noise-free optimization with Expected Improvement
• Batch Bayesian Optimization with Batch Expected Improvement, Local Penalization, Kriging Believer and GIBBON
• Batch-sequential optimization with Thompson sampling
• Inequality constraints
• Explicit Constraints
• EGO with a failure region
• Multi-objective optimization with Expected HyperVolume Improvement
• Using deep Gaussian processes with GPflux for Bayesian optimization.
• Bayesian optimization with deep ensembles
• Active Learning
• Active Learning for Gaussian Process Classification Model
• Bayesian active learning of failure or feasibility regions
• Trieste meets OpenAI Gym
• Scalable Thompson Sampling using Sparse Gaussian Process Models
• Batch HSRI Tutorial
• Multifidelity Modelling with Autoregressive Model
• High-dimensional Bayesian optimization with Random EMbedding Bayesian Optimization (REMBO).

Frequently asked questions

The following tutorials (or sections thereof) explain how to use and extend specific Trieste functionality.

• How do I set up a basic Bayesian optimization routine?

• How do I set up a batch Bayesian optimization routine?

• How do I make a custom acquisition function?

• How do I recover a failed optimization loop?

• How do I track and visualize an optimization loop in realtime using TensorBoard?

• What are the key Python types used in Trieste and how can they be extended?

• Does Trieste have interface for external control of the optimization loop, also known as Ask-Tell interface?

• How do I perform data transformations required for training the model?

• How do I use Trieste in asynchronous objective evaluation mode?

  • Example of using greedy batch acquisition functions and Python multiprocessing module.

  • Example of using non-greedy batch acquisition functions and Ray.

Run the tutorials interactively

The above tutorials are built from Jupytext notebooks in the notebooks directory of the trieste repository. These notebooks can also be run interactively. To do so, install the library from sources, along with additional notebook dependencies, with (in the repository root)

$ pip install . -r notebooks/requirements.txt

then run

$ jupyter-notebook notebooks

Alternatively, you copy and paste the tutorials into fresh notebooks and avoid installing the library from source. To ensure you have the required plotting dependencies, simply run:

$ pip install trieste[plotting]