Algorithms, Live: Differential Evolution

Part 7 of my interactive optimisation series went up today: differential evolution. It takes two of its own population members, uses the gap between them as the step, and so the search contracts as the points converge without anything in the code arranging it.

Four landscapes - you can click the map to dump the population in one spot and watch it fan out.

https://www.kenreid.co.uk/blog/differential-evolution-live.html

Anyone else build toys to understand a thing?