Not a big fan of the DEVS formalism for my work. Evaluated simpy years back but dumped it in favor of custom solution (and some other external considerations). I remember seeing some ill advertised libs in scala and java (mostly from academia). I was pretty impressed with JaamSim (although it’s oop and heavy on mutation) talking to folks at WinterSim years back; they were aiming at competing with proprietary software while remaining open source. “Data Science” tends to generally ignore discrete event sim and optimization (bread and butter Ops Research fields) in my experience. They may remain niche/external, or poorly defined terms may end up encompassing both into “Data Science” at some point.
I intend to wrap my incanter fork around kixi.stats in the near future (no idea if upstream would dig that). Great library (also portable cljc).