Mulitilayered Abstractions for Partial Differential Equations

This EPSRC-funded research project (reference no EP/I00677X/1) aims to deliver new capabilities and new levels of performance for a large class of computational simulations based on partial differential equations.

The project is tightly coupled with other projects, in particular:

• Sustainable domain-specific software generation tools for extremely parallel particle-based simulations” (EPSRC, EP/I006761/1)
• PhD studentship - NERC DTG NE/G523512/1
• Strategic Investment in Low Carbon Engine Technology SILOET (Workpackage 3.7) (Rolls Royce plc and TSB)
• Application Performance Optimisation and Scalability (FP7/277481)

• OP2
• UFL

The project is a collaboration between:

• Paul H J Kelly (Dept of Computing, Imperial College London). Paul leads the Software Performance Engineering research group and is co-director of Imperial's Centre for Computational Methods in Science and Engineering
• David Ham (Dept of Computing, Imperial College London and Grantham Institute for Climate Change, Imperial College London)
• Mike Giles (Professor of Scientific Computing, Oxford University, Mathematical and Computational Finance Group, Oxford-Man Institute of Quantitative Finance, Oxford e-Research Centre)
• Michael Fagan (Dept of Engineering, University of Hull). Prof Fagan is Director of Hull's Centre for Medical Engineering and Technology and leads the Medical and Biological Engineering Research Group.

Associated researchers:

    Nicolas Loriant, Department of Computing, Imperial College London
    Florian Rathgeber, Department of Computing, Imperial College London
    Graham Markall, Department of Computing, Imperial College London
    Lawrence Mitchell,Edinburgh Parallel Computing Centre
    Gihan Mudalige,Oxford University
    István Reguly, Oxford University
    Francis Russell, Department of Computing, Imperial College London
    Georgios Rokos, Department of Computing, Imperial College London
    Dr. Carlo Bertolli, Department of Computing, Imperial College London
    Nathan Chong, Department of Computing, Imperial College London

Software:

• OP2 (see the Oxford University OP2 webpages here)
• PyOP2 (github, here)
• Integration into the FEniCS Project
• Integration into the FLUIDITY Project
• (separately funded through the SILOET programme above) OP2 for Fortran for the Rolls Royce HYDRA CFD code

Publications:

• Francis P. Russell, P.H.J. Kelly, Optimized Code Generation for Finite Element Local Assembly Using Symbolic Manipulation, under review
• G. R. Markall, A. Slemmer, D. A .Ham, P. H. J. Kelly, C. D. Cantwell, and S. J. Sherwin, Finite element assembly strategies on multi- and many-core architectures. International Journal for Numerical Methods in Fluids, published online: 19 Jan 2012, doi: 10.1002/fld.3648
• C. Bertolli, A. Betts, N. Loriant, G. R. Mudalige, D. Radford, D. Ham, M. B. Giles, and P. H. J. Kelly Compiler Optimizations for Industrial Unstructured Mesh CFD Applications on GPUs, Languages and Compilers for Parallel Computing (LCPC'12), September 2012.
• Carlo Bertolli, Adam Betts, Paul H. J. Kelly, Gihan R. Mudalige, Mike B. Giles: Mesh independent loop fusion for unstructured mesh applications. Conf. Computing Frontiers 2012: 43-52
• Mike B. Giles, Gihan R. Mudalige, Z. Sharif, Graham R. Markall, Paul H. J. Kelly: Performance Analysis and Optimization of the OP2 Framework on Many-Core Architectures. Comput. J. 55(2): 168-180 (2012)
• G. J. Gorman, James Southern, P. E. Farrell, M. D. Piggott, Georgios Rokos, Paul H. J. Kelly: Hybrid OpenMP/MPI Anisotropic Mesh Smoothing. Procedia CS 9: 1513-1522 (2012)
• Georgios Rokos, Gerard Gorman, Paul H. J. Kelly: Accelerating Anisotropic Mesh Adaptivity on nVIDIA's CUDA Using Texture Interpolation. Euro-Par (2) 2011: 387-398
• Carlo Bertolli, Adam Betts, Gihan R. Mudalige, Mike B. Giles, Paul H. J. Kelly: Design and Performance of the OP2 Library for Unstructured Mesh Applications. Euro-Par Workshops (1) 2011: 191-200
• Mike B. Giles, Gihan R. Mudalige, Z. Sharif, Graham R. Markall, Paul H. J. Kelly: Performance analysis of the OP2 framework on many-core architectures. SIGMETRICS Performance Evaluation Review 38(4): 9-15 (2011)
• Francis P. Russell, Michael R. Mellor, Paul H. J. Kelly and Olav Beckmann, DESOLA: an Active Linear Algebra Library Using Delayed Evaluation and Runtime Code Generation. Science of Computer Programming 76(4), 2011