Andrew Shao, PhD: Senior Software Engineer, Master Technologist, HPE
Scott Bachman, PhD: Research Scientist, National Center for Atmospheric Research; Visiting Research Scholar, HPE
Large, complex computer simulations of the physical world, once only available to national laboratories, are now becoming more common-place within the academic community thanks to the broader availability of high performance computing resources. Contemporaneously, the advent of open-source software, data science libraries, and machine-learning frameworks mean that some aspect of software engineering is now required of domain scientists. In this presentation, Andrew Shao and Scott Bachman will discuss how these advances have made collaborations with software engineers, data scientists, and domain scientists a necessity for advancing scientific progress.
They will also describe the benefits (and caveats) to the scientific community from the open source software culture and how machine-learning and traditional scientific simulations are setting the stage for a new type of modelling. As an example of how science advances in this new environment, they close with their story of how they used HPE’s SmartSim, an open-source solution for integrating ML into simulations, along with the open-source ocean climate model MOM6, to perform the first realistic global simulations of the world’s oceans with an embedded machine-learning model.