Informatics researcher’s thesis earns an outstanding dissertation prize from TU Dresden
Andres Goens, who is currently a researcher in the Consistency, Availability and Persistency via Synthesis research group within Institute for Computing Systems Architecture has been awarded “Cloud & Heat Dissertationspreis” for outstanding doctoral dissertation work at TU Dresden.
Computer hardware keeps increasing in complexity and software design needs to keep up with this. For example, it is difficult to decide how to distribute computation among many different processors in a chip, especially when these are of different kinds.
Andres’ thesis deals a family of methods to address this problem, called (model-based) software synthesis. In these, the main idea is to have a mathematical model of how the different parts of the computation interact with each other so that we can reason about the computation. The main example used in the thesis is called Kahn Process Networks. It describes a way to compute something in different places at different times while ensuring the result is deterministic: If we give it the same inputs, we will get the same outputs, independent of when and where the different parts of the computation executed.
An important question in this context is to decide where (and when) to execute things, to get the most performance or efficiency in the system. The main contribution of this thesis are methods to improve these decisions, by using mathematical structure found in the problem. The thesis also focuses on applications in various domains, primarily embedded systems used for telecommunication or in the automotive industry.
Andrés Goens has been working as Research Associate at the School of Informatics since 2021. He did his PhD at TU Dresden. His research focuses on mathematical structures in design-space exploration for heterogeneous systems, as well as semantics and machine learning for compilers and programming languages.
This story has been developed with the help of Adarsh Doddappagouda Patil from InfGang (the Informatics Science Communications Group).