Computational Catalyst Research
Catalysis is a crucial research field for the UK's economy, for example, 90% of all chemical processes are catalysed. Research into improved catalysts will be crucial to keep the UK's competitive edge, and to help achieve a net zero carbon future.
The computational chemistry group in the scientific computing department (SCD-STFC) have been awarded a 4-year EPSRC grant to investigate the emerging field of hybrid catalysis. PI Tom Keal, Chin Yong and You Lu will work closely with leading computational researchers at University College London (UCL) and the University of Bristol and experimental collaborators across the country on the £2.6m project.The six investigators, three researchers and five experimental groups will work closely together to combine advanced computational methods with cutting edge experimental techniques to evolve the design of hybrid catalysts, in association with, and in support of, the UK Catalysis Hub.
The project team will seek computational methods for the treatment of hybrid catalysts, which combine aspects of three traditional categories of catalyst (heterogeneous, homogeneous, and biocatalysis) – taking the best ideas from all to produce new, highly efficient designs.
The project will further advance SCD's state of the art ChemShell1 software package with features to support modelling of hybrid catalysts with unprecedented accuracy.
Join CoSeC researchers at the exascale frontier
Four CoSeC-funded scientists are co-investigators on a successful £3m proposal as part of the “ExCALIBUR" exascale computing inititative. ExCALIBUR2 is a major research programme led by EPSRC and the Met Office to develop high-performance simulation software for the highest-priority fields in UK research.
In the first phase of the ExCALIBUR programme, a number of Design and Development Working Groups (DDWGs) were set up to bring together research communities to develop use cases and software blueprints for exploiting the exascale platforms of the future. These included the Materials and Molecular Modelling group (MMM DDWG), bringing together several materials science consortia supported by CoSeC, including the Materials Chemistry Consortium (HEC-MCC) and the UK Car-Parrinello consortium (UKCP), and the Massively Parallel Particle Hydrodynamics working group (MPPH DDWG), targeting applications in engineering and cosmology.
CoSeC researchers Ian Bush, Alin Elena and Tom Keal (MMM) and Stephen Longshaw (MPPH) are co-investigators in the working groups, taking an active role in the efforts to identify software development requirements, key scientific drivers, underlying benchmarking calculations and promoting knowledge exchange in close collaboration with the academic partners and through engagement exercises with their respective communities.
For the second phase of the project, the MMM and MPPH working groups have joined forces to form the new Particles At eXascale on High Performance Computers (PAX-HPC) consortium, a £3m proposal led by UCL, which has now been funded to pursue new software developments for the exascale across the whole spectrum of particle-based modelling, including all the fields supported by the MMM and MPPH efforts. Key research themes include task-based parallelism, accelerator development and complex exascale workflows.
As part of the next phase we have two vacancies for research software engineers to join the Scientific Computing Department to work on the PAX-HPC project. For more details on the posts and to apply please see the links below:
Post 1 – Multiscale Workflows
Post 2 – Accelerator Development
CoSeC Multi-CCP GPU Hackathon
In the second week of this year CoSeC together with NVIDIA3 organised a GPU hackathon to support porting to modern hybrid architectures. The event targeted software developed and maintained by the UK Collaborative Computational Projects (CCP) and by ISIS Neutron and Muon Source Facility, specifically software that underpins the research of many scientific communities in the UK.
Four codes were considered: Xcompact3D from the UK Turbulence community4, CHAPSims for the CCP Nuclear Thermal-Hydraulic5, DL_POLY for the CCP5++ community6 related to condensed matter physics, and Dissolve for the modelling and refining of scattering data from Neutron and Muon sources.
The event included a mix of lectures delivered by NVIDIA staff, and practical sessions where the above codes were ported to GPU using the newly available NVIDIA V100 nodes on SCD's SCARF supercomputing cluster.
More than 80 people across STFC departments, Universities and NVIDIA attended the lectures, and 20 worked on the hackathon delivering real progress on the road to modernising codes for the UK scientific communities. CoSeC organisers would like to send a big thank you to mentors Filippo Spiga and Paul Graham (both NVIDIA), Jony Castagna (STFC-Hartree Centre), Dimitar Pashov (King's College London) for helping in the hack sessions and also to NVIDIA for supporting the event.
1ChemShell is a computational environment for multiscale computational chemistry developed at Daresbury, which provides efficient methods for modelling catalysis by combining quantum mechanical calculations and classical forcefields (“QM/MM"). The new approaches developed in the project will result in more accurate simulations that scale well to the largest supercomputers.
2ExCALIBUR
5CCP Nuclear Thermal-Hydraulic
6CCP5++ community
