A Dumb Introduction to z3 using Rust (asibahi.github.io)

A newbie-friendly walkthrough shows how to use the Z3 SMT solver from Rust (via its Rust bindings) to express and solve constraint and optimization problems. The author walks through tiny, concrete examples — solving x+4=7, a two-variable system that is Unsat for Int but satisfiable with Real (giving rational solutions 17/3 and 34/3), enumerating multiple integer solutions for x*x=4, and listing integer lattice points on a circle — to demonstrate the typical workflow: declare solver variables (Int or Real "Sorts"), assert constraints, call solver.check(), and extract values with get_model() or the convenient solutions iterator. The piece also highlights that the bindings translate Rust expressions to SMT-LIB2, and explains jargon: "Sort" = type, "constants" can be free (variables) or interpreted (literals), and fresh_const auto-generates unique solver names. The post then shows optimization with Z3's Optimize object via a coin-change example (minimize total coin count) and surfaces an important pitfall: unbounded Ints allow negative coins, producing a nonsensical optimum until you assert non-negativity. Practical implications for the AI/ML community are clear: SMT solvers like Z3 are powerful and expressive for scheduling, combinatorial constraints, and small-scale optimization; Rust bindings are ergonomic but require careful encoding (choose sorts, add bounds, and be mindful of infinite solution spaces) to get correct, useful results.