impact
Impact at the Maxwell Institute
The creation of Impact outside academia is a key priority of the Maxwell Institute, and examples of recent and current impact successes are listed below. To facilitate Impact there has been investment in Business Development Executives, consultancy units and support for individual researchers. The industry and agency partners contributing to the MAC-MIGS CDT are also enabling the Maxwell Institute to foster Impact. The main areas in which the Maxwell Institute is aiming to generate Impact are Finance, Energy, Pensions and Insurance, Software, Data Analytics, and Protecting the Natural Environment, but there will always remain scope for individual initiatives.
Global food solutions via optimal diet formulations
Research Team: Julian Hall, Ken McKinnon, Andreas Grothey, Ivet Galabova
Research Area(s): Optimization, Operational Research
Impact on food formulation: High performance optimization software is essential to the efficient formulation of food, particularly in the (farm) animal and pet food markets, but also in important areas of human food production.
Beneficiaries: The optimization solvers have been largely developed for Format Solutions, the world’s leading supplier of software for food formulation.
Significance and Reach: The reach is global. Significance is demonstrated by the dependence of Format Solutions on the Edinburgh technology when selling its software – in particular to a manufacturer producing more than half of the world’s pet food.
Understanding the Impact of Covid-19 on Life Insurance Portfolios
Research Lead: Andrew Cairns
Research Area: Actuarial Mathematics
Early in the 2020 pandemic, Andrew Cairns was approached by the US-based multinational, Prudential Retirement to develop models that would help understand better the impact of Covid-19 on different sections of society. How much socio-economic inequality is there, and how much does mortality vary with age? How will it impact on life insurance liabilities for the survivors? The results of this research proved to be of great interest to the insurance community generally with numerous invited presentations at industry conferences, specialist teams in international re-insurers and the French Institute of Actuaries. Cairns’s research helped to maintain momentum in the longevity risk transfer market through the highly uncertain early stages of the pandemic.
STACK: the world-leading open-source online assessment system for mathematics and STEM
Research Lead: Chris Sangwin
Research Area(s): Technologically Enhanced Learning
STACK generate random questions and provides objective online assessment to support students and teachers, focused on mathematics and science subjects. Students can answer with algebraic expressions and answers are graded based on mathematical properties. STACK also allows a student to rearrange mathematical diagrams, and interact with sophisticated drag and drop questions. Multi-part questions support learning of complex tasks and the development of higher level skills.
STACK is an open source project, with a full authoring interface for teachers. Large question banks are available of tried and ested materials STACK is widely used internationally at large universities and is available in many languages. A vibrant international community of users develop and share material banks and knowledge of effective practice, helping teachers create their own high quality courses efficiently. There is a rapidly growing pan-African user group. STACK has been continuously developed since 2003 and Edinburgh is the home of the STACK project.
The evaluation and interpretation of evidence in forensic science casework
Research Team: Colin Aitken, Amy Wilson
Research Area(s): Statistics
Our judicial system increasingly relies on the quantification of the value of evidence presented in court. As a result, advanced statistical methods have a strong impact on the administration of justice. The research team has applied Bayesian statistics to develop methodology for this quantification of evidence and has proposed and implemented procedures for the evaluation of forensic evidence from (i) multivariate hierarchical data and (ii) autocorrelated data (exemplified with work on drugs on banknotes). The procedures and methods developed are routinely used in forensic laboratories worldwide for casework, are recommended in international guidelines for forensic scientists, and have supported the accreditation of a UK laboratory. Research outputs have been cited in expert witness reports in court cases worldwide. Beneficiaries include both forensic scientists and the justice system.
Helping to control the spread of Mycoplasma bovis in New Zealand
Research Team: Gavin Gibson and George Streftaris
Research Area: Statistics
Methods for the simultaneous estimation of epidemic dynamics and pathogen evolution developed by Gibson, Streftaris and PhD student Lau were incorporated into a user-friendly computer package BORIS by scientists from the University of Melbourne. From July 2018, the BORIS package has been used as an analytic tool within the New Zealand Ministry of Primary Industries (MPI) eradication programme for Mycoplama Bovis, a bacterial disease that affects dairy and beef cattle, estimated to cost NZD 886,000,000. Specifically, BORIS has been used to identify potential times and sources for observed infections so that risk factors for transmission, or potential failures in biosecurity can be identified. The eradication programme has been effective with only 4 premises out of more than 20,000 having active disease in July 2020.
Dynamic Centrality
Research Lead: Des Higham
Research Area(s): Applied & Computational Mathematics
Identifying “highly influential” individuals in peer-to-peer networks is critical in sectors as varied as advertising media and security. Research by Grindrod and Higham – focused on mathematical modelling and analysis of evolving networks – has developed first-principles discrete time dynamical models for digital communications between people, putting forward a new theoretical framework for describing and analysing time-varying connectivity. This work has led to algorithmic approaches to identifying strong influencers; proposes a tool that identifies key players in complex evolving networks; and opens up the possibility of real-time monitoring and prediction. The digital marketing agency Bloom (acquired by Jaywing Plc in 2016 ) used these ideas to launch ‘Whisper’: a commercially available, real time social planning software product that has been used in campaigns for high-profile clients such as Sky, Yorkshire Tea and KitKat.
Strengthening asset management by building diverse portfolios
Research Team: Sotirios Sabanis, Jacek Gondzio, Sergio García Quiles, Joerg Kalcsics
Research Area(s): Mathematical Finance, Optimization, Probability & Stochastic Analysis
Research was carried out in collaboration with Aberdeen Standard Investments (ASI), the investment arm of one of the largest asset managers in the UK and in Europe, namely Standard Life Aberdeen, for the design and implementation of new diversification algorithms, based on which multi-asset portfolios are created with better performance under adverse market conditions.
Direct beneficiaries of this work include ASI with the launch of a new product (fund) and their clients. Indirectly, increasing the resilience of UK’s asset management sector carries clear benefits both for the UK society and its economy.
HiGHS – high performance software for linear optimization
Research Team: Julian Hall, Ivet Galabova, Filippo Zanetti, Ben Champion, Yanyu Zhou
Research Area(s): Optimization, Operational Research
HiGHS is the world’s best open-source software for linear optimization, so its impact is both broad and deep. Building on the award-winning research work of Hall, HiGHS currently solves linear programming (LP) problems via the simplex, interior point and first order methods, convex quadratic programming problems using an active set method, and mixed-integer programming (MIP) problems via branch-and-cut. These are the fundamental models for optimal decision-making. HiGHS provides the LP and MIP solvers in commercial software products such as MATLAB, and open-source interfaces such as SciPy. It is the default open-source solver for major commercial operational research modelling products such as AMPL and GAMS, as well as open-source modelling languages such as JuMP. The Python interface to HiGHS has over 200,000 downloads per month.
The impact of HiGHS can be measured in various ways. Commercial companies save on the license fees charged by vendors of closed-source optimization software without compromising solver performance, or develop products where the use of licensed software is impractical. Those who cannot afford commercial optimization software licenses, such as smaller companies, NGOs and those working in developing countries, are able to solve operational research problems that would have been intractable with lesser open-source solvers. The value of HiGHS in the world of open energy modelling has been recognised by recent philanthropic donations to the project of over $800k.
Electricity security of supply and climate resilience
Research Team: Chris Dent, Stan Zachary and Amy Wilson
Research Area(s): Optimization, Operational Research, Statistics
Maxwell Institute researchers work with the UK energy industry in these areas. There is a long standing relationship with the National Energy System Operator on development of methodology for assessing the risk of future electricity supply shortfalls (commonly known as ‘the lights going out’), and the amount of supply capacity required to keep this risk to an appropriate level. Past work has included how to compare contributions of new technologies (e.g. renewables and storage) on a fair basis with conventional generating capacity (gas, nuclear etc), and how to manage uncertainty background against which capacity procurement decisions are taken. Current work includes the reliability contribution of interconnection to other systems, use of weather and climate data in assessments, and the methodology for the new long term reliability study looking to the 2030s.
The Institute also works on climate resilience projects with companies such as UK Power Networks and National Grid Electricity Transmission, studying the potential impacts of extreme heat on electricity networks. Chris Dent sits on the Science Assurance Group for the UK Climate Change Risk Assessment Technical Report. He contributes regularly to international industry activities, such as IEEE Standards relating to the GB reliability work, and Energy System Integration Group reports which are highly influential in setting technical agendas internationally.
Mathematical Modelling Protects the UK Red Squirrel Population
Research Lead: Andy White
Research Area: Mathematical Biology
Mathematical modelling was used to direct policy and practice to conserve red squirrels – a protected species in the UK – from squirrelpox, carried by invasive grey squirrels. The research demonstrated that the red squirrel conservation policy used up to 2015 was not sufficient to contain squirrelpox in the UK and was the trigger for a radical change in policy to protect red squirrels in UK priority areas, directly informing the ‘Scottish Strategy For Red Squirrel Conservation (2015)’ and the Red Squirrel Survival Trust policy to protect red squirrels on Anglesey. The research also evaluated red squirrel population viability under different forest management scenarios, underpinning commissioned reports for the Forestry Commission that have had a direct impact on current and future forest management practice in the UK. The techniques are being extended to develop disease control strategies in a wide range of ecological systems, including assessing the spread of tuberculosis and African swine fever in wild boar across Europe and understanding the risk of vorroa mite infestation on the viability of honey bee colonies.
Mortality Models for the International Insurance and Pensions Industry
Research Lead: Andrew Cairns
Research Area: Actuarial Mathematics
Pension funds, life insurers and regulators are concerned about the financial consequences of longevity risk: the financial risk that people live longer than anticipated. Novel stochastic models for the assessment of longevity risk (the Cairns-Blake-Dowd – CBD – family) in life insurance and pensions have been widely adopted – in the UK and globally – by insurers, actuarial consultancies, insurance regulators, specialist software providers, and in professional education. The CBD models have played a central role in the transfer of 10’s of billions (GBP) pension liabilities from pension funds to multinational insurers including the £16bn transfer of BT pension liabilities in 2014. Use of the models provides an improved assessment of the financial consequences of longevity risk. This has enhanced the security of both pension funds and insurers and, through good risk management and regulation, has reduced the risk of insolvency.
Modelling Anomalies in Population Data
Research Team: Andrew Cairns and Torsten Kleinow
Research Area: Actuarial Mathematics
The 2011 UK census revealed accuracy issues with the underlying population data that were of considerable concern to the pensions and life insurance sectors. This led to a research project at the Maxwell Institute that developed methods for identifying anomalies in national population and mortality data. The work has impacted on institutions in the UK, US and France and has enabled insurers to reduce prices for the transfer of pension liabilities, saving UK pension funds between £330 million and £1 billion. It has persuaded actuaries to revise the mortality tables used for pricing and reserving, including changes in the methodology underpinning the UK actuaries’ Continuous Mortality Investigation mortality projection tables. The research team is also working with data in developing countries where census counts and death registrations are affected significantly by so-called “digit preference” (e.g. rounding ages to the nearest 0 or 5).
Using machine learning to predict battery performance with Dukosi
Research Team: Gonçalo dos Reis, Miguel Anjos, Paula Fermin, Encarni Medina-Lopez
Research Area(s): Machine Learning
Lithium lon Cells are ubiquitous in Electric Vehicle and Energy Storage solutions but vary in life expectancy due to their source chemistry and usage patterns. Goncalo dos Reis was awarded an IAA Grant to develop and run advanced Machine Learning models to accurately predict the end of Lithium lon Cell’s useful life. An innovative new concept for battery management systems was built – called knee-onset – which gives a very early warning of rapid cell degradation. Mathematical models were developed to identify knee points from capacity degradation curves and to predict when it will occur from a very early stage of the cell’s life. The methods used have significant positive commercial applications for the energy storage sector and are also of considerable academic interest, given the growing attention being devoted to the modelling of cell behaviour. A paper has appeared in ‘Energy and AI’.
WEST Brewery
Research Lead: Ben Goddard
Research Area(s): Complex Fluids
Ben Goddard was awarded an EPSRC Impact Acceleration funding for his project ‘Development of Spectral Element Methods for Complex Fluids Modelling and Control’, carried out in conjunction with Heidi Beers. The project employed a postdoctoral researcher, Rory Mills-Williams, to implement numerical methods devised by Ben Goddard and John Pearson to tackle the optimization of processes in the brewing industry.
Non-stationary models within an automated machine learning framework
Research Team: Sara Wade, Karla Monterrubio-Gomez
Research Area(s): Non-stationary Models, Machine Learning, AI
Wade and Monterrubio-Gomez were awarded an EPSRC IAA Secondment with Secondmind, an AI and machine learning start up. They worked with the company to integrate their research on non-stationary models within an automated machine learning framework; leading to improved forecasting tools for the company’s clients, as well as better automated-decisions through integration within the company’s decision-making platform.
Statistical learning for precision oncology
Research Lead: Timothy Cannings
Research Area(s): Statistics, Machine Learning, Precision Medicine, Cancer Genomics
Timothy Cannings was awarded an EPSRC IAA Secondmnet to Cambridge Cancer Genomics (CCG.ai), an AI startup working on personalized cancer treatment. Throughout 2019, he worked closely with industry experts at CCG.ai in order to help develop and improve their data-driven tool for precision oncology OncOS.
Interest Rate models with Moodys
Research Team: David Siska and Lukasz Szpruch
Research Area(s): Mathematical Finance
David Siska and Lukasz Szpruch were awarded part of the IAA fund as a follow up to a FinTech Aim day session with Moodys Analytics. The project employed a colleague from EPCC to work on applying algorithms developed by Dr. Siska and Dr. Szpruch to interest rate models and data provided by Moodys.
Developing the discrete element method for improving battery production
Research Lead: Ben Goddard
Research Area(s): Applied & Computational Mathematics
Lithium batteries are essential in many modern technologies; they are key in electric vehicle power, solar power storage, alarm systems in remote locations, mobility equipment, etc. When linked to renewable energy sources, they can drive the revolution in green transport and enhance the lives of people in developing countries and areas with less well-established power networks. A major bottleneck in such applications is battery performance and lifetime, which is intimately linked to the manufacturing processes, especially of battery electrodes. Working collaboratively with Altair EDEM, we are developing improvements to the Discrete Element Method, allowing the modelling and optimization of battery electrode production. This will ultimately increase the sustainability not only of battery technology, but also of the mathematical and computational methods used to model it.
Statistics for the Justice System
Research Leads: Amy Wilson
Research Area(s): Statistics
The drawing of valid and robust conclusions from observations is crucial to minimise the risk of miscarriages of justice within our judicial system. Wilson at Edinburgh is working on developing statistical methods to support the legal system at different points within the justice system, from policing and forensics through to presenting evidence in the courtroom. She is working with forensic science providers in gait and toxicology to support modelling for the interpretation of evidential data as well as developing Bayesian graphical modelling approaches for decision support when there is a combination of different forms of evidence. She is also working with the Royal Statistical Society to improve training in statistics for lawyers and the police and to develop new guidance to support those that need to work with evidential data and statistical interpretations within the context of the law.
Causal healthcare analytics for Real-World Evidence via Targeted Learning
Research Leads: Sjoerd Beentjes
Research Area(s): Statistics, Data Science
The National Institute for Health and Care Excellence (NICE) produces guidance for the NHS and wider healthcare system based on rigorous assessment of complex real-world evidence (RWE) for new health technologies and treatments. The use of real-world data (RWD), in addition to randomised controlled trials, has become increasingly popular for biomedical researchers, regulatory bodies and pharmaceutical industry. Evaluating reliable RWE and drawing causal conclusions from RWD is challenging due to lack of treatment randomisation, intercurrent events, and informative loss to follow-up. Targeted Learning (TL) offers an ideal step-by-step framework to address these challenges.
TL has been successfully applied with USA industry and Food and Drug Administration. We have a UK software implementation at UoE. We bring together expertise and end-users from NICE, industry (Teva Pharmaceuticals Ltd) and academia (UC Berkeley, School of Public Health) to place this framework into action within the UK healthcare ecosystem for the benefit of UK patients. This research is supported by a Harmonised Impact Accelerator Account jointly funded by EPSRC and MRC.
Blood markers for diagnosis of Myalgic Encephalomyelitis
Research Leads: Chris Ponting (Institute of Genetics & Cancer, Sjoerd Beentjes (Mathematics), Audrey Ryback (IGC), Ava Khamseh (School of Informatics)
Research Area(s): Statistics, Data Science
Myalgic Encephalomyelitis, also known as Chronic Fatigue Syndrome (ME/CFS), is a disease of unmet need which disproportionately affects younger females. Unfortunately, the condition has historically been ascribed to psychosocial, rather than biological, causes. In approximately two thirds of cases the condition is triggered by a viral infection. There are currently no diagnostic criteria for ME/CFS and no curative treatment.
Mortality and health inequalities
Research Leads: Andrew Cairns
Research Area(s): Actuarial Mathematics
Understanding and quantifying mortality and health inequalities is important to many stakeholders including policymakers, the general public, pension schemes and life insurers. Research led by Professor Andrew Cairns uses data science methods to quantify more accurately how big these inequalities are, how these vary depending on age group, and how they vary across the country. A challenge for government is how to spend tax payers money in the most effective way to reduce these inequalities and “level up”. Cairns’s techniques pinpoint which neighbourhoods are most in need of attention. The research team has produced a the open access LIFE App which allows users to visualise how these inequalities vary across different regions of the country.
Synthetic polycrystalline microstructure generation
Research Lead: David Bourne
Research Area(s): Analysis, Applied & Computational Mathematics, Optimization
David Bourne works with engineers from Tata Steel Research & Development to develop fast algorithms for generating realistic synthetic polycrystals for representing the microstructure of steel. These geometric models can be combined with computational homogenization to design and test new alloys. This is an important part of the transition to green steel: in 2020 the steel industry was responsible for about 7%-9% of global C02 emissions. Our mathematical tools include computational geometry, optimisation, and optimal transport theory.
Statistical Consultancy Unit
Research Leads: Michael Allerhand, Gail Robertson
Research Area(s): Statistics
The Statistical Consultancy Unit (School of Maths, University of Edinburgh) have been involved in a range of impactful projects, from ‘Identifying offshore foraging areas used by seabirds’, to ‘Evaluating an instrument to assess delirium in patients in intensive care’, developed at Edinburgh Royal Infirmary.