[FEATURE REQUEST]: example showing a quick way of estimating Pareto fronts

[R-M-Lee]

Motivation

I think an example showing explicitly how to get Pareto fronts from predictive MO strategies would be nice. It's almost what, e.g., tutorials/benchmarks/011-ZDT1.ipynb is doing, but not quite.

Describe the solution you'd like to see implemented in BoFire.

• Define/choose benchmark
• Get initial dataset
• Fit surrogates
• generate lots of random samples in the input space
• Get surrogate predictions for the random samples
• Retain only the non-dominated samples
• Visualize

Describe any alternatives you've considered to the above solution.

The problem can be more elegantly solved with, eg, pymoo. That will be the subject of a separate PR

Is this related to an existing issue in BoFire or another repository? If so please include links to those issues here.

No response

Pull Request

None

[jduerholt]

What do you mean with more elegantly solved with pymoo?

[R-M-Lee]

I mean that instead of generating random samples and then just retaining the non-dominated ones, you can use pymoo to evolve a population of points to cover the pareto front. The pymoo approach will often give (at least in theory) a nicer distribution of points across the pareto front wrt to, for example, the following metrics (although they are not necessarily explicitly optimized)