This repository contains my code for working on an optimal placement problem using Bayesian Optimization methods. I have used Facebook's newly released library Ax and the Botorch framework for my purpose. I have also used the Matlab Engine for python since some of the code needs functions written in Matlab and a constratined optimization problem needs to be solved in the objective function. This includes a custom code for fmincon for a python interface since matlab engine has problems accepting function pointers.
Ax or Adaptive Experimentation is a great library to start with Bayesian Optimizations and supports powerful MC based Acquisition function based methods. I have worked with the stndard Gaussian Process + Expectation Improvement framework so far but this is by no means Ax's limit. A great place to start with Ax would be ax.dev/tutorials/ . This has integrability with Facebook's Botorch (https://botorch.org/docs/botorch_and_ax) a Bayesian Optimization framework integrated with pytorch and conventional pytthon. The math behind what was offered , which took me some time to parse is explained here (https://botorch.org/docs/botorch_and_ax)