Gastvortrag: Martin Ziegler: Computer Science of Continuous Data

Dienstag, 21.09.2021, 10.30 Uhr

Ort: Online Session

Vortragender: Martin Ziegler

Abstract:

Since its early days, Computer Science has enjoyed the support and guidance of Logic, from Theory via Engineering to Practice: recall Alan Turing’s 1936 publication preceding nowdays ubiquitous digital computers or Alonzo Church’s Lambda Calculus having led to functional programming languages or axiomatized structures in Model Theory corresponding to specification of Abstract Data Types or Hoare Logic for formal program verification -- concerning the processing of discrete data, such as graphs or integers or strings.
Continuous data on the other hand commonly arises in Engineering and Science (natura non facit saltus) in the form of temperatures and fields.  It mathematically includes real numbers, smooth functions, bounded operators, or compact subsets of an abstract metric space. Processing such continuous data has arguably been lacking the foundational support from Logic in Computer Science that the discrete case is enjoying, and keeps forcing users to resort to discretizations such as IEEE 754 that violate associateive and distributive laws, break symmetries, introduce and propagate rounding and cancellation errors in addition to an involved (and incomplete) axiomatization including NaNs and denormalized numbers.We present an overview of the motivation, history, state of the art and future vision of Computer Science for Continuous Data.