A tour d'horizon of constrained continuous optimization problems for benchmarking

1 minute read

Coming soon

This is a draft blog post using MathJax to render math. I plan to add several blog posts in the future but first I need to publish papers to level up my Scholar profile.

For example, I could talk about what I did during my PhD, that is to solve black-box optimizations problems of the form

\begin{align} \underset{x\in\mathcal{R}^n}{\text{minimize}} \quad &f(x) \\ \text{subject to} \quad & g_k(x) \le 0 \quad , \end{align}

where $f$ and $g_k$ are all functions of $\mathcal{R}^n\rightarrow\mathcal{R}$ for $k=1,\ldots,m$.

We proposed a methodology to construct test problems, and instantiated a concrete collection of these problems, to assess the performance of optimization algorithms.

This is implemented in the Comparing Continuous Optimizers (COCO) library under the bbob_constrained test suite. Check also this documentation.