Department of Mathematics and Computer Science

Imported from Grants Tool 4722431

The Square Kilometre Array Observatory (SKAO) is a next-generation radio astronomy facility, involving partners around the globe, that will lead to groundbreaking new insights in astrophysics and cosmology. Established on March 12, 2019 the SKAO is the second inter-governmental organisation dedicated to astronomy in the world. It will be operated over three sites: the Global Headquarters in the UK, the mid-frequency array in South Africa (SKA-mid), and the low-frequency array in Australia (SKA-low). 

The two telescopes under construction, SKA-Mid and SKA-Low, will combine the signals received from thousands of small antennae spread over a distance of several thousand kilometres to simulate a single giant radio telescope capable of extremely high sensitivity and angular resolution, using a technique called aperture synthesis. Some of the sub-arrays of the SKA will also have a very large field-of-view (FOV), making it possible to survey very large areas of the sky at once!

Switzerland has become the eighth country to join the intergovernmental nations that will collaborate in building the Square Kilometre Array Observatory (SKAO), to be built in Australia and South Africa. Swiss involvement is organized through a strong consortium of research institutions, called SKACH, including, Fachhochschule Nordwestschweiz (FHNW), Universität Zürich (UZH), Eidgenössische Technische Hochschule Zürich (ETHZ), École Polytechnique Fédérale de Lausanne (EPFL), Zürcher Hochschule für Angewandte Wissenschaften (ZHAW), Universität Basel (UniBas), Université de Genève (UniGE), Haute École spécialisée de Suisse Occidentale (HES-SO), Centro Svizzero di Calcolo Scientifico (CSCS).

As part of SKACH, the aim of our group is to extend the SPH-EXA simulation framework to include proper cosmological physics to reach trillion particle simulations on hybrid Tier-0 computing architectures. To this end we aim at coupling relevant physics modules with our SPH framework enabling the possibility of addressing both long-standing and cutting-edge problems via beyond state state-of-the-art simulations at extreme scales in the fields of Cosmology and Astrophysics. Such simulations include the formation, growth, and mergers of supermassive black holes in the early universe which would greatly impact the scientific community (for instance, the 2020 Nobel Prize in Physics has been awarded for pioneering research on super-massive black holes). Moreover, the ability to simulate planet formation with high-resolution models will play an important role in consolidating Switzerland’s position as a leader in experimental physics and observational astronomy. Additional targets will be related to explosive scenarios such as core-collapse and Type Ia supernovas, in which Switzerland has also maintained a long record of international renown. These simulations would be possible with a Tier-0-ready SPH code and would have a large impact on projects such as the current NCCR PlanetS funded by the SNF.

Domain-independent dynamic programming (DIDP) is a recently proposed framework for modeling combinatorial optimization problems using dynamic programming and solving them with a general purpose solver. Initial results show strong performance using a heuristic state-based search solver, thus demonstrating the success of problem solving approaches developed in Artificial Intelligence (AI) on problems traditionally studied in Operations Research (OR). This project seeks to develop the theory and practice of DIDP-based problem solving by characterizing DIDP from computational and knowledge representation perspectives; by understanding its relationship to the problem solving paradigms embodied in constraint programming, satisfiability, and AI planning; and by developing novel domain-independent heuristics for DIDP through the adaptation and extension of techniques from AI planning, state abstraction in OR, and decision diagram-based optimization.

Automated planning, also known as factored state-space search, is a fundamental problem in AI with numerous applications in computer science and elsewhere. There are three dominant algorithmic paradigms: heuristic search, symbolic search, and SAT planning. While important theoretical results exist to explain and improve the performance of some of these algorithms, almost nothing is known about the connections between them.

We build a unified theory of factored state-space search that identifies a common reasoning core encompassing all three algorithmic paradigms that allows us to directly relate these algorithms to each other in theory and build hybridized algorithms combining their individual strengths in practice. Identifying such a core also allows us to better understand the trade-off involved between representational expressiveness and efficiency of reasoning and design better algorithms to address this trade-off.