Let our optima combine!
summary
Shows a concrete workflow: model a 3×3×3 Hebrew perfect pangram (27 cells) with OR-Tools CP-SAT, channel integer cells to 27×22 Booleans via order encoding, add counting and allowed-assignment constraints, break symmetry by fixing the center word, and maximize a max-min objective (popularity of the least popular word). A Colab runs it end-to-end: feasible solution ~1 min, optimum ~8 min, enumerate all optima ~2 min, yields two optimal solutions and claims one–two orders of magnitude speedup over hand‑crafted search.
AI-written summary · about
chapters (11)
- 0:00Welcome
- 0:43Combinatorial optimization
- 2:08Constraint satisfaction problems
- 4:42OR-Tools CP-SAT solver
- 6:52CP-SAT over other methods
- 9:05Declarative programming with OR-Tools
- 11:21Hebrew 3D pangram challenge
- 16:20Modeling in code
- 18:48Core constraints and optimization
- 21:55Live demo in Google Colab
- 27:35Resources
related talks
29mLet our optima combine!
01h 04mBuilding AI Solutions with Google OR-Tools
17mSynthesizing MILP Constraints for Efficient and Robust Optimization
34mSolving Problems Declaratively
16mJust-in-Time Learning for Bottom-Up Enumerative Synthesis
11mOm! Nom! Nash!
21mCombinatorial Optimization of Train Timetable Fulfilment After Incidents
20mFunction synthesis for maximizing model counting