BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:FTorch: facilitating hybrid ML-numerical modelling in scientific c
 omputing - Joe Wallwork and Jack Atkinson\, RSEs - ICCS\, University of Ca
 mbridge
DTSTART:20260226T130000Z
DTEND:20260226T140000Z
UID:TALK244552@talks.cam.ac.uk
CONTACT:Jack Atkinson
DESCRIPTION:In the last decade\, machine learning (ML) and deep learning (
 DL) techniques have revolutionised many fields within science\, industry\,
  and beyond. Researchers across domains are increasingly seeking to combin
 e ML with numerical modelling to advance research. This typically brings a
 bout the challenge of programming language interoperation. PyTorch (Paszke
  et al.\, 2019) is a popular framework for designing and training ML/DL mo
 dels whilst Fortran remains a language of choice for many high-performance
  computing (HPC) scientific models. \n\nThe FTorch library provides an eas
 y-to-use\, performant\, cross-platform method for coupling the two\, allow
 ing users to call PyTorch models from Fortran. This talk will introduce th
 e challenges and our approach to solving them. We will look at building an
  efficient but friendly piece of software for researchers\, touch on recen
 t developments including hardware acceleration and automatic differentiati
 on capabilities\, and take a look at some recent applications of the softw
 are in hybrid modelling.\n \nReference: Atkinson et al.\, (2025). FTorch: 
 a library for coupling PyTorch models to Fortran. Journal of Open Source S
 oftware\, 10(107)\, 7602<href=https://doi.org/10.21105/joss.07602>https://
 doi.org/10.21105/joss.07602</href>
LOCATION:Room C\, West Hub
END:VEVENT
END:VCALENDAR