2022 Sustainability and Computational Science
This one-day workshop was held on 24 November 2022, fika to fika, in room MA:4. It could also be attended via Zoom.
The idea of the workshop was to discuss and think about how sustainability and computational science interact. What is a framework for sustainability that is useful for computational scientists to make use of? What is the energy usage of computations and how can we change it? How is computational science used to answer questions about sustainability?
9:45: Fika outside of the lecture hall.
10:15: Introduction by Philipp Birken, Lund
10:20: Introduction to COMPUTE, Sonja Aits, Lund
10:30: Kristina Jönsson (Lund): The UN Agenda 2030 for sustainable development: a holistic perspective
Abstract: The adoption of the 2030 Agenda and its 17 Sustainable Development Goals in 2015 came with high expectations on how to cater for a sustainable future. The universal ambition and holistic approach were praised but implementing the agenda has proven difficult for a range of reasons. There are not only synergies but also tensions between the goals, transformation is challenged by path dependency, uneven resource distribution, power structures, and so on. This raises a number of questions: What can we expect from the 2030 Agenda? Are there different ways to approach the agenda? How can we work with the overall framework and with the different goals? The idea is to put the 2030 Agenda in perspective but also to discuss how it can be used as a framework for advancing our own research.
11:15: Heiner Linke (Lund): Network-based biocomputation using molecular motors: a paradigm for highly energy-efficient computing
Abstract: Network-based biocomputation (NBC) is an alternative, parallel computation approach that can potentially solve technologically important, combinatorial problems with much lower energy consumption than electronic processors. In NBC, a combinatorial problem is encoded into a physical, nanofabricated network. The problem is solved by biological agents (such as cytoskeletal filaments driven by molecular motors) that explore all possible pathways through the network in a massively parallel and highly energy-efficient manner. I will introduce the approach and report the state-of-the-art, which includes proof-of-concept solutions to instances of hard combinatorial problems such as Exact Cover and Satisfiability (SAT). I will also briefly discuss energy limits for computation in finite time, pointing out that massively parallel computation offers a fundamental energy advantage over traditional serial computation.
13:00: Gabriele Paciucci (nVidia Corporation): Partnering with NVIDIA on Sustainability
Abstract: Energy crisis hits science: supercomputing centres struggle with surging gas and electricity prices. Accelerated computing is energy efficient. It is also possible to optimize even further energy consumption, de-emphasizing Time-To-Solution and operating a throughput-oriented architecture at Max-Q. Conventional applications can find a more sustainable path embracing ARM and the recently announced NVIDIA GRACE Superchip.
13:45: Gerhard Wellein (Erlangen): Power, Energy and HPC
Abstract: Demand for High-Performance Computing and Data Center Services is continuously increasing, driving power demands and energy consumption of computing centers. This Development has a high economic and ecological impact. I will discuss countermeasures taken in HPC from the perspective of the HPC center and the user. Specifically, I will report on efforts to integrate a multi-PetaFlop cluster into existent Infrastructure and discuss the problem of energy efficient execution settings using an analytic energy model.
14:30: Sonja Aits (Lund): Natural language processing in environmental and health research
Abstract: Our planet’s ecosystems and human society are threatened by the interlinked climate and biodiversity crises, which put humans, animals, plants and even microorganisms at risk. In addition, we are facing a string of global health crises, e.g. COVID-19, antibiotics resistance and diseases linked to pollution. To help achieve a sustainable world and stop or even revert the ongoing crises, researchers and policy makers need to stay on top of the existing knowledge. This is not easy, however, as small pieces of information are scattered over millions of scientific articles, governmental and non-governmental reports, patents, databases, datasets and even news and social media posts. We have thus reached a level of information overload that makes processing by human researchers impossible. Natural language processing, a form of artificial intelligence, can help alleviate this problem and assist with tasks such as information extraction, text summarization, topic modelling and sentiment analysis. Consequently, there is a wide range of applications for this group of technologies in environmental and health research.
15:00: Fika outside of the lecture hall.