KafE: From Guess Engineering to Knowledge Engineering

Room 2, all the way upstairs!!!

It’s time for KafE! This time Pieter Collins returns with a follow up talk about his software tool Ariadne, Don’t worry if you missed last years talk (Verification of Cyber-Physical Systems), this is for everyone: from first year bachelors to his fellow colleagues.

Abstract: Although computers are capable of performing trillions of logical and arithmetical operations without making a single error, most computational mathematical software makes so many (tiny) approximations that the results are really only a (very good) guess. For us knowledge engineers, this really isn’t good enough, we want to know for sure our results are correct!

For many years I have been working on a software tool “Ariadne” for proper, rigorous numerical computation with real numbers providing guaranteed bounds on all errors made. The original aim, as discussed in my previous KafE talk “Verification of Cyber-Physical Systems”, was to be able to prove that dynamic control systems, like robots, do what you want them to. But since this problem requires us to rigorously solve all kinds of difficult problems in continuous mathematics, the tool also provides a C++ library and Python scripting environment for general-purpose rigorous numerical computation. In this talk, I will give you a tutorial on the design and use of Ariadne, a tour-de-force of knowledge engineering combining logic, algebra, calculus, numerics, simulation, optimization, data structures and modelling, with some things which you don’t know but should, like interval arithmetic. I will also talk about the development process, which will mostly be explaining all the things that can go wrong if you don’t apply what you learnt in your software engineering course.