Reaction–diffusion failures

Three distinct ways the Gray-Scott setup blew up this week, all of which turned out to be the same lesson wearing different masks: watch your attribute ranges.

First failure — everything went uniform grey. The feed and kill rates were balanced so perfectly that the reaction never got going. Second — values exploded to NaN because the timestep was too large for the diffusion coefficients. Third — the pattern was there but invisible, sitting in a 0.49–0.51 range I hadn't remapped for display.

None of these were bugs in the solver. They were the numbers telling me exactly what was wrong, if I'd bothered to print them sooner. Dropping a quick min/max readout on the concentration attributes turned an afternoon of guessing into a five-minute fix.