Building AI Solutions with Google OR-Tools
summary
Shows how to model and solve real problems with Google OR-Tools, from Sudoku pruning a 10^77 search space down to about 10^21 using all-different constraint propagation to linear programming for a pottery example with 24 oz clay and 16 oz glaze maximizing 3x+9y revenue (bounds 16 small or 6 large, 71 feasible points), then mixed integer scheduling encoded as a 3D boolean matrix with about 120 boolean variables to enforce speaker and room constraints, plus links to working GitHub examples.
AI-written summary · about
chapters (24)
- 0:00Intro
- 1:22What do I mean by "Artificial Intelligence"?
- 3:43Types of Al Models
- 6:12Constraint-Based Optimization
- 7:09Classes of Problems
- 7:52Feasibility vs. Optimization
- 11:45Constraint: A Required Condition
- 13:19Types of constraints
- 15:04Objective: A Goal for the Solution
- 16:54Constraint Programming: Sudoku
- 24:03Linear Example: Pete's Pottery Paradise
- 25:37Pottery Production - Solution Space
- 28:35Pottery Production - The Polytope
- 33:07Defining an Optimization Model
- 33:59Decision Variables
- 37:53Execute the Model
- 40:00Mixed-Integer Example: Scheduling
- 40:39LP Model - Variables
- 41:21LP Model - Constraints
- 43:40MIP Model - Variables
- 45:04MIP Model - Constraints
- 50:03MIP Model - Objective Example
- 50:54Summary
- 52:02Resources
related talks
29mLet our optima combine!
29mLet our optima combine!
51mCombinatorial Optimisation Problems – from Drug Discovery to Space
44mMathematical Modeling with Kotlin
30mBuild on-demand logistics apps with Python, OR-Tools, and DecisionOps
45mLLMs can't optimize schedules, but AI can!
24mThe Rise of Reasoning Models
54mAmazing Algorithms for Solving Problems in Software