Machine learning methods are now used in the simulation of the building blocks of matter: from the electronic- to the molecular-level structure. These tools have boosted well-known computational methods such as density functional theory or molecular dynamics simulation, and are expected to lead to new physical insights which, in turn, can enable the engineering of exotic new materials.
About the event
In this event, we will hear from Professor Eun-Ah Kim and Assistant Professor Michele Ceriotti.
Professor Eun-Ah Kim - Machine leaning quantum emergence
Decades of efforts in improving computing power and experimental instrumentation were driven by our desire to better understand the complex problem of quantum emergence. The resulting "data revolution" presents new challenges. I will discuss how these challenges can be embraced and turned into opportunities through machine learning. The scientific questions in the field of electronic quantum matter require fundamentally new approaches to data science for two reasons. Firstly, quantum mechanics restricts our access to information and secondly, inference from data should be subject to fundamental laws of physics. Hence machine learning quantum emergence requires collective wisdom of data science and condensed matter physics. I will review rapidly developing efforts by the community in using machine learning to solve problems and gain new insight. I will then present my group’s results on the machine-learning-based analysis of complex experimental data on quantum matter.
Assistant Professor Michele Ceriotti - Atomic-scale simulations of matter with machine learning
Machine learning has become an indispensable tool in the simulation of matter at the atomic scale, replacing or supporting electronic structure calculations that are in turn increasingly more accurate and predictive. I will present a brief overview of the key concepts that have emerged as the guiding principles in the effective application of ML to this class of problems. Additionally, I will provide several examples of applications that have been made possible by the synergies between data-driven and physics-based modelling. I will summarise what I see as the most pressing challenges and promising research directions in the field.
Who should attend?
This event is of interest to physicists from all fields who are using or intending to use AI in their research. It is equally relevant to AI scientists who are developing new machine learning tools and who would want to know what types of problems physicists are trying to solve. The talks and the follow-up discussion will be kept non-technical so that the content is accessible to non-specialists. At the end there will be an opportunity to ask questions.
About the event series
Nature Reviews Physics and The Alan Turing Institute are tapping into international and local expertise to try to identify the different classes of problems that machine learning can help physicists solve and the key issues that physicists and machine learning scientists should be aware of. In doing so we are hoping to create a framework to facilitate understanding and the exchange of ideas. Read more about the series and register for future events.
Chair - Professor Gábor Csányi
16:00 - 16:05 - Event introduction
16:05 - 16:25 - 20 minute presentation
16:25 - 16:45 - 20 minute presentation
16:45 - 17:05 - 20 minute Q&A session
17:05 - 17:10 - Event summary
If you have any further questions about the event, please contact Zaynab Ismail (Research Project Manager) at [email protected].