Geoffrey De Smet | Devoxx

Geoffrey De Smet
Geoffrey De Smet Twitter

From Red Hat

Geoffrey De Smet is the founder and lead of OptaPlanner (, the leading open source A.I. constraint solver in Java.


bigd Big Data, Machine Learning, AI & Analytics

Constraint solving A.I. algorithms in OptaPlanner


Some problems are (NP-)hard: employee shift rostering, vehicle routing problems, cloud optimization, skill based task assigning or even just building a fair tennis club schedule. Simple algorithms can’t solve these problems anywhere near optimally. But if we used advanced algorithms, we get better solutions that increase profit, reduce our ecological footprint and improve customer satisfaction.

In this session, I will introduce constraint optimization, demo a few use cases, use weighted hard and soft constraints to formalize business goals, walk through example code in Java of our open source constraint solver OptaPlanner, cover real-time continuous planning and teach algorithms such as Exhaustive Search (Brute Force), Construction Heuristics (First Fit) and Local Search (Tabu Search, Simulated Annealing).

lang Programing Languages

System.out.println(0.01 + 0.09 != 0.10); // Prints true


Not all data is the same. Some input data has surprising side-effects. Take the number 9007199254740993 for example. It is the smallest number that exists as a long, but not as a double. And that can cause havoc. There are many types of weird data, such as escape characters, the Turkish i, certain datetimes, etc.

In this session, I’ll explain those cases, how to cover them in your unit tests efficiently and how to fix your code.