Z3 API in Python: From Sudoku to N-Queens in Under 20 Lines (ericpony.github.io)

Microsoft Research’s Z3 theorem prover is showcased in a compact Z3Py tutorial that demonstrates how to model and solve constraint problems in Python — from simple algebra to combinatorial puzzles like Sudoku and N-Queens (often expressible in under 20 lines). The write-up emphasizes Z3’s role across verification, testing, security and scientific modeling, and highlights that Z3Py exposes the solver API with Python-friendly constructors (Int, Real, Bool, BitVec), a high-level solve() helper, and the more flexible Solver() with push/pop, check(), and model inspection. It also notes other language bindings (C, .NET, OCaml) and front-ends (Scala, SBV). Technically, Z3Py supports integers, reals (including arbitrary-precision rationals and algebraic irrationals), nonlinear polynomials, Boolean logic (And/Or/Implies/If), mixed-type coercions (ToReal), and bit-vectors for precise machine arithmetic. Users can tune display and solver behavior with set_option (precision, rational_to_decimal, html_mode), traverse and simplify expressions, extract internal s-expressions via sexpr(), and collect solver statistics. Important caveats: some non-polynomial constraints (e.g., 2**x == 3) can return unknown. For AI/ML practitioners, Z3Py offers a compact, expressive toolkit to encode constraints, hybrid discrete/continuous models, and verification tasks directly inside Python workflows.