RSE training in algorithms for exascale simulations
Lead Research Organisation:
UNIVERSITY OF EXETER
Department Name: Mathematics
Abstract
The exascale computing landscape in the UK is at an exciting stage,
with funding being allocated to novel architectures, new software
frameworks and innovative algorithms. Through training RSEs we have an
opportunity to embed the progress made in these areas into the core of
academic research and industrial applications, positioning the UK as
an international leader in exascale simulations. To grasp this
opportunity is essential that RSEs are trained in algorithms so
that they can take an active part in research in this area. In order
to make informed and creative design choices when writing and
optimising software, RSEs need to have core knowledge of algorithms so
that they can confidently innovate and avoid the common pitfalls that
academics and industrial partners are already aware of through their
research and experience. If this core knowledge is not passed on to
RSEs and shared throughout the RSE community, advances made through
the ExCALIBUR programme research projects risk failing to achieve
crucial impact in academic and industrial applications.
As described in section 7.3 of the RSE Knowledge Integration Landscape
Review, it is crucial that RSEs have the potential to be actively
involved in research. This is one of the key attractions of the job
for skilled postgraduate students and is essential for retaining
skilled RSEs in the role. The Landscape Review acknowledges that
design of new algorithms is a research field in itself and requires
`strong domain specific knowledge'. We propose to provide training in
this area, alongside opportunities for knowledge exchange and
networking between academic researchers, postgraduate students, RSEs
and industrial partners.
We propose to run two three-day workshops and a Summer School to
provide training in state of the art algorithms and core knowledge of
the underlying foundational mathematical and numerical analysis on
which they are based. The materials developed in advance of, and
during, these events will be curated and shared online to either be
used as stand alone material for individual training or to form the
basis of future summer schools.
with funding being allocated to novel architectures, new software
frameworks and innovative algorithms. Through training RSEs we have an
opportunity to embed the progress made in these areas into the core of
academic research and industrial applications, positioning the UK as
an international leader in exascale simulations. To grasp this
opportunity is essential that RSEs are trained in algorithms so
that they can take an active part in research in this area. In order
to make informed and creative design choices when writing and
optimising software, RSEs need to have core knowledge of algorithms so
that they can confidently innovate and avoid the common pitfalls that
academics and industrial partners are already aware of through their
research and experience. If this core knowledge is not passed on to
RSEs and shared throughout the RSE community, advances made through
the ExCALIBUR programme research projects risk failing to achieve
crucial impact in academic and industrial applications.
As described in section 7.3 of the RSE Knowledge Integration Landscape
Review, it is crucial that RSEs have the potential to be actively
involved in research. This is one of the key attractions of the job
for skilled postgraduate students and is essential for retaining
skilled RSEs in the role. The Landscape Review acknowledges that
design of new algorithms is a research field in itself and requires
`strong domain specific knowledge'. We propose to provide training in
this area, alongside opportunities for knowledge exchange and
networking between academic researchers, postgraduate students, RSEs
and industrial partners.
We propose to run two three-day workshops and a Summer School to
provide training in state of the art algorithms and core knowledge of
the underlying foundational mathematical and numerical analysis on
which they are based. The materials developed in advance of, and
during, these events will be curated and shared online to either be
used as stand alone material for individual training or to form the
basis of future summer schools.
Organisations
| Description | This grant provided funding for workshops and training in algorithms for exascale computing. One key finding was that there was a lot of enthusiasm for both delivering and receiving this training. In particular, participants appreciated the opportunity to work through exercises with experts on hand to help and they benefitted from networking with their peers. Another key finding relates to the running of this type of training. It was hard for RSEs (the intended audience) to attend due to constraints on their own time and funds. Making material available online for asynchronous learning would help with this but is very time consuming for those delivering it and does not provide the same networking opportunities. |
| Exploitation Route | Other people running similar training sessions would benefit from considering the obstacles and successes we experienced. Dr Shipton has already used her experience in her role on the advisory board for the UNIVERSE-HPC project. |
| Sectors | Digital/Communication/Information Technologies (including Software) Education |
| URL | https://github.com/jshipton/ExascaleAlgorithmsPDEs |
| Description | Summer School: Algorithms for Exascale Computing |
| Form Of Engagement Activity | Participation in an activity, workshop or similar |
| Part Of Official Scheme? | No |
| Geographic Reach | National |
| Primary Audience | Postgraduate students |
| Results and Impact | A week long Summer School was held at the University of Exeter, delivering training in algorithms for exascale computing, including numerical discretisation, finite element methods, computational linear algebra and parallel-in-time methods. There were keynote talks by PIs of related ExCALIBUR projects and time for discussion and networking. The audience was a mix of PhD students interested in RSE careers, current RSEs and ECRs in related fields. |
| Year(s) Of Engagement Activity | 2024 |
| URL | https://sites.google.com/view/algorithmsforexascale/ |
| Description | Workshop Exascale Computing Challenges: Data Driven Algorithms |
| Form Of Engagement Activity | Participation in an activity, workshop or similar |
| Part Of Official Scheme? | No |
| Geographic Reach | National |
| Primary Audience | Professional Practitioners |
| Results and Impact | Workshop to provide training in data driven algorithms for exascale computing for Research Software Engineers in academia and industry. |
| Year(s) Of Engagement Activity | 2024 |
| URL | https://www.eventbrite.co.uk/e/excalibur-workshop-data-driven-algorithms-tickets-780794767167 |
| Description | Workshop Exascale Computing Challenges: Parallel-in-Time Algorithms |
| Form Of Engagement Activity | Participation in an activity, workshop or similar |
| Part Of Official Scheme? | No |
| Geographic Reach | National |
| Primary Audience | Professional Practitioners |
| Results and Impact | Workshop to provide training in parallel time stepping algorithms for Research Software Engineers in academia and industry. |
| Year(s) Of Engagement Activity | 2023 |
| URL | https://www.eventbrite.co.uk/e/exascale-computing-challenges-parallel-in-time-algorithms-registratio... |