This site seems to compare various global optimization methods on different test problems (with visualization of “energy landscape”).
Global Optimization Benchmarks
– Global Optimization Benchmarks — Global Optimization Benchmarks 0.1.0 documentation
Also some blog posts about Python global optimization packages.
nlopt, cobyla, minpack, slsqp, pgapack, pikaia
A number of these implement local optimization algorithms, not global ones, although you can increase the chance of finding the global optimum by calling them with multiple initial guesses.
Some time ago I rewrote in modern Fortran the function fminsearch from Matlab, that implements Nelder Mead.
A lot of the popular global optimization algorithms seem to struggle with higher dimension problems (like more than 10 unknowns). If you pair that with a problem that involves some non trivial time to compute a solution (simulations where a single run takes minutes, hours, or days+ to complete), I’ve personally struggled to find anything that seemed significantly better than randomly guessing around the hypothesis “good region”