ICCS Research Software Engineers Joe Wallwork and Jack Atkinson ran a workshop in Cambridge that brought together 40 researchers, scientists and software engineers from across Europe to discuss coupling machine learning technologies into large-scale scientific models.
The two-day event at the Maxwell Centre, University of Cambridge, featured eight invited talks, panel discussions and break-out sessions focusing on differentiable models, research-to-operations pathways, and future computer hardware adaptations. The workshop took place thanks to support from Accelerate Programme for Scientific Discovery and the Cambridge Centre for Data-Driven Discovery.
The engineers used funding from Accelerate-C2D3 to develop FTorch – software that provides a technical solution for coupling PyTorch machine learning codes to computational models written in Fortran.
Since its initial development in 2022, FTorch has been adopted globally by research projects including those working with the Community Earth System Model, ICON (the German weather forecasting model), and nuclear fusion research at UKAEA.
Read Joe and Jack's report of the workshop on the Accelerate website