Dr. Bojan Nikolic, Principal Research Associate at the Cavendish Laboratory, Cambridge
Abstract: I show that a software framework intended primarily for training of neural networks, PyTorch, is easily applied to a general function minimisation problem in science. The qualities of PyTorch of ease-of-use and very high efficiency are found to be applicable in this domain and lead to two orders of magnitude improvement in time-to-solution with very small software engineering effort. This result demonstrates that re-purposing of machine learning software can lead faster scientific results even when using traditional scientific approaches (i.e., exact models motivated by hypotheses). Paper Link: https://arxiv.org/abs/1805.07439
Bio: Dr. Bojan Nikolic is a Principal Research Associate at the Cavendish Laboratory, Cambridge. Bojan's interests span astronomy, instrumentation, computing and software. He has played a major role in construction/commissioning of three prominent recent radio astronomy facilities: the GBT 100-m telescope, the ALMA 66-element array and the forthcoming Square Kilometre Array. His current astronomical research is focused on the formation of the earliest stars and black holes and the "Epoch of Reionisation" that is triggered by these objects. Bojan has also worked in industry as a senior software engineer and been a speaker at prominent computing conferences. His software is run daily in production at observatories and commercial organisations around the world.