Human-in-the-loop Bayesian Optimization

In chemistry and materials science, Bayesian optimization is often used to guide experimental workflows. However, often there is a person running the algorithm, carrying out the suggested experiment, and then passing the data back to the algorithm. This can introduce some complexity from an implementation point of view, so here's a really simple way to use a Bayesian optimization platform (ax.dev/) in an interactive fashion. Simply add the new data to a Python dictionary at the beginning of a notebook and run the notebook again. The notebook shown is available at colab.research....