York: Transforming Research-Oriented Software Engineering

Lead Research Organisation: University of York
Department Name: Physics

Abstract

Research Software Engineering (RSE) is the creation of well-designed, reliable, efficient computer programs to solve research problems. In my Fellowship I will focus on: RSE in materials modelling, a large research field with important industrial applications which affect our everyday lives; plasma and fluid modelling, which has many applications to clean energy-generation as well as industrial & medical processes; and on promoting and developing RSE skills in the research community.

The core software development is focused on CASTEP, a state-of-the-art implementation of density functional theory (DFT) for materials modelling, and BOUT++, a plasma fluid code. These programs are world-leading exemplars for UK RSE. Both codes were designed from the ground up using sound RSE principles, and are free to all UK academics. CASTEP's ease of use has drawn users from across the STEM disciplines; it is used by over 900 academic & industrial groups worldwide and cited over 9200 times in the scientific literature. I will transform CASTEP's ease of use by non-computational scientists, ensuring quick, accurate and reliable simulations, and reduce the time-to-science for all users. I will further enable CASTEP to become the software foundation for new, higher-level computational methods, including multiscale modelling, rare-event sampling & high-throughput materials discovery. This will strengthen UK science right across materials research, and ultimately lead to better materials for everyone.

The developments in BOUT++ will expand its field of applicability to allow both advanced new plasma physics and geometries, and to enable it to solve equations from other fields of science. This will both empower plasma scientists to model sophisticated new designs for fusion reactors, and open BOUT++ up to whole new scientific communities; as an example, UoY will shortly start a pilot project with Nestle to use BOUT++ to model bubbles in chocolate.

This Fellowship also includes development of new software to tackle current research problems, not only in York but also in industry and at the UK's world-leading experimental facilities Diamond Light Source, ISIS Neutron Facility and SuperSTEM. Initial projects include RSE to aid modelling chemical synthesis, predicting core-loss spectra and crystal & magnetic structure prediction, with further projects to be sought and delivered within the Fellowship period. These software services will promote RSE across a diverse range of STEM, increase the effectiveness and impact of a wide variety of research initiatives, and address directly many of EPSRC's Grand Challenges in Physics, Engineering and Chemical Science.

The final component of this Fellowship is to train, support and inspire the next generation of research software engineers. I will develop new training material, to be delivered in York but disseminated online to the wider community; create support groups within York, and link up with neighbouring institutions; and work with national scientific consortia (including HECs and CCPs) to promote and support RSE nationally. The people this supports are the future of this vital field, and will be invaluable to research in the UK, as well as the wider world. It is they who will ensure that the skills and experiences gained by researchers on the core development projects are transferred into the wider community.

In addition to these specific RSE components, I will also raise the profile and recognition of RSE workers and skills in the UK and abroad. I will champion RSE particularly in the materials modelling domain, promoting RSE to academia and industry through high-profile showcases, conferences, workshops and targeted, ongoing collaborations. The strong RSE Group I will build at the University of York will extend RSE provision and skills training to all researchers at York, promoting Research Software Excellence in all disciplines.

Planned Impact

Economic & societal impacts will come from the development and wider uptake of state-of-the-art software that this Fellowship facilitates, the science that its use enables, and the highly-skilled RSEs who will be trained.

The first strand of the Fellowship is core development of CASTEP and BOUT++. CASTEP is an ab initio materials modelling code and a powerful enabler of materials design, whose capabilities and potential for impact is demonstrated by high-profile programmes like the Materials Genome Initiative in the US [1] or by the European Commission [2], and their widespread use in electronic & manufacturing industries [2-3]. A general problem with ab initio codes is the specialist knowledge required for their efficient and accurate use, but this will be removed by PJH in this project, opening CASTEP up to non-computational scientists, enabling them to make predictions to guide experimental design and interpret observations. This is particularly useful for partners at Diamond Light Source, SuperSTEM and ISIS Neutron facility. It will also enable far larger simulations, bringing new capabilities such as "whole-device modelling".

BOUT++ is a world-leading plasma fluid code, but will be transformed in this Fellowship into a general partial differential equation solver, capable of handling multiple physics models. This simultaneously allows simulations of complex geometries, such as advanced diverters, and opens BOUT++ up to applications in new areas of science such as bacterial motion or wound healing.

The potential economic & societal impact is difficult to overstate. There are over 3.5 billion mobile phones and 1 billion PCs worldwide, with total energy consumption projected to triple by 2030, exceeding the present combined residential electricity use of the US and Japan [4]. "Materials design" and whole-device modelling can revolutionise the capabilities and energy efficiency of these devices, and new fusion reactor geometries could provide a breakthrough in clean, renewable energy production.

These developments will be distributed to UK academics free-of-charge. In the case of CASTEP, BIOVIA market it worldwide and provide an easy-to-use CASTEP GUI, plus training & support. Customers include many UK and global companies who use materials modelling in R&D. The new developments will enable many more industrial users, outside the traditional confines of expert "DFT modellers". The direct economic impact of CASTEP's industrial use is evident from the Return on Investment of 300-1000% in the physical and life sciences sectors[5][6], and the 260 patents citing Castep simulations[7].

The software services which form the second strand of this Fellowship will empower researchers to tackle high-impact EPSRC "Grand Challenges" in science. The additional emphasis on national experimental facilities will lead to new modelling capabilities, new user communities, and will maximise the impact of both this work and that of the facilities. The new software will be available freely to all UK academics, along with training material.

The third strand of this project is to train the next generation of RSEs and teach RSE skills to researchers. This will lead to a step-change in the quality of research software produced at York and beyond, and help tackle the "reproducibility crisis" in science. This also addresses directly the "skills gap" in UK industry and enables the impact of this work to continue to grow long after the Fellowship period.

1. G. Ceder & K. Persson, Materials Science, The stuff of dreams, Scientific American, 36-40 (2013).
2. European Commission, "ICT Infrastructures for e-science" Brussels, 2009.
3. http://www.enterprise.cam.ac.uk/news/2013/1/castep-achieves-30-million-sales/
4. Gadgets and Gigawatts - Policies for Energy Efficient Electronics, 2009.
5. M. Swenson et al, IDC white paper, July 2004.
6. A.S. Louie et al, Health Industry Insights, Jan. 2007
7. Google Patent Search (21/8/2017)

Publications

10 25 50
 
Description The work developed in the CASTEP computer program was merged with the main codebase for the 19.1 release, and marketed worldwide by Dassault Systemes. CASTEP is used by around 1000 companies and academic groups worldwide, and is cited in support of over 250 patents.
Sector Aerospace, Defence and Marine,Agriculture, Food and Drink,Chemicals,Digital/Communication/Information Technologies (including Software),Education,Electronics,Energy,Environment,Manufacturing, including Industrial Biotechology,Culture, Heritage, Museums and Collections,Pharmaceuticals and Medical Biotechnology,Security and Diplomacy
Impact Types Economic

 
Description NVIDIA 
Organisation NVIDIA
Country Global 
Sector Private 
PI Contribution Developing a GPU port of the CASTEP code
Collaborator Contribution 2 GPU-based compute cards, compilers, access to NVIDIA's development cluster and expert engineering advice
Impact On-going work
Start Year 2018
 
Title CASTEP 19 
Description CASTEP is a software package for predictive, quantum-mechanical simulations of materials and chemicals. It is based on density functional theory, and can simulate a wide range of materials proprieties including energetics, the structure at the atomic level, vibrational properties and electronic response properties. In particular it has a wide range of spectroscopic features that link directly to experiment, such as infra-red and Raman spectroscopies, NMR, and core level spectra. CASTEP version 19 included many enhancements and optimisations, including improved performance and the ability to compute electron-phonon coupling (important for electronic transport modelling). 
Type Of Technology Software 
Year Produced 2018 
Impact CASTEP is used by around 1000 companies and research groups around the world. The key papers describing CASTEP are cited over 1000/year and CASTEP is cited in support of over 250 patents. CASTEP is available under a free-of-charge licence to UK academia, and marketed commercially worldwide by Dassault Systemes. The enhancements to CASTEP 19 will be particularly useful for industries in the Semiconductor and Energy sectors. 
URL http://www.castep.org
 
Description CASTEP User Training Workshop 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Postgraduate students
Results and Impact Around 60 people attended this week-long workshop, learning how to use the CASTEP computer program to understand, explain and predict the behaviour of materials. By the end of the workshop all attendees had performed advanced simulations such as NMR chemical shifts, EELS or Raman spectra.
Year(s) Of Engagement Activity 2018
URL http://www.castep.org/CASTEP/Workshop2018
 
Description Hermes Summer School 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Postgraduate students
Results and Impact Over 30 students from around the world attended this summer school to learn cutting-edge simulation methods and approaches. The atmosphere was relaxed and collaborative, and there was a lively discussion following the talks which continued throughout the summer school.
Year(s) Of Engagement Activity 2018
URL http://hermessummerschool.org/about-1/
 
Description OMG Science 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach Local
Primary Audience Schools
Results and Impact An Outreach event in a school in Leeds, demonstrating quantum physics to secondary school children. This was organised by the Archbishop of York Youth Trust and concluded with a talk and a lively panel debate on "Faith and Science" (I was a member of the panel).
Year(s) Of Engagement Activity 2018
URL https://www.archbishopofyork.org/news/news-2018/omg-science-abbey-grange-academy
 
Description University Open Days 2018 
Form Of Engagement Activity Participation in an open day or visit at my research institution
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Schools
Results and Impact Thousands of pupils and parents came to the University of York Open Day, and well over 100 were specifically interested in Physics. I devised a computer demonstration of materials modelling, running on an odroid low-power computer, which was on the Physics stall, delivered a talk on "theoretical and computational physics" and showcased some recent research. I also developed a practical demonstration of thermoelectric conversion from a temperature gradient to electrical energy, demonstrating the correctness of the computational predictions.
Year(s) Of Engagement Activity 2018