Using Constraint Satisfaction to Optimize Item Selection for Bundles in Minecraft (www.robw.fyi)

A developer used MiniZinc to model and solve the problem of packing Minecraft items into a single bundle slot, demonstrating how constraint satisfaction (CSP) can cleanly express and solve a knapsack-like inventory optimization. Starting from a simple binary selection array (array[1..n] of var 0..1: selected) with the objective solve maximize sum(selected), they added a capacity constraint so the solver only picks items whose combined bundle cost fits the bundle. The post shows incremental modeling: naïve unconstrained selection, adding a capacity constraint to pick the maximal number of slots, and then handling heterogeneous stack sizes (64, 16, unstackable) by integer-scaling to avoid decimals. Technically, the author scales bundle slots by 16 (units = 16*64 = 1024) so costs become fullstack=16, smallstack=64, unstackable=1025 (so it won’t fit). The constraint sum(selected[i] * inventory[i]) <= units plus maximize sum(selected) yields the expected solutions (e.g., selecting half-stacks of dirt and ender pearls to maximize freed slots). The examples run fast in the MiniZinc Playground (≈169–215 ms). For the AI/ML community this is a concise, practical example of using CSP solvers for discrete combinatorial optimization—readily extendable to other storage systems or game-mechanic changes without hand-tuning heuristics.