Connor Bottrell
Title:"The importance of realism in classifying galaxy interactions using convolutional neural networks"
Abstract:
Characterizing galaxy mergers is challenging. But recently, major advances in the technology and availability of computing resources have revived efforts in this area of research. Machine learning methods offer crucial advantages over traditional methods of characterizing galaxy mergers: consistency, expediency and the ability to exploit intricate connections between data and targets. However, based on observations alone, many of the biases that these new methods propose to solve still end up embedded in the models. Hydrodynamical simulations offer a significant advantage: truth!
Title: Simon Guichandut - "History and importance of stellar winds"
Abstract:
The fact that our own sun is constantly outflowing with matter at supersonic speeds might be baffling to today's layman, but it also was to physicists 60 years ago when Eugene Parker proposed his model. In this retrospective talk, I will talk about Parker's famous wind and describe how additional physics modify its outcome.