The speed of a wave moving through a material is set by the refractive index; something immutable we might look up in a table and perhaps promptly forget. But imagine having the power to change it at will. What could we do? It would allow a single object to have different functions: a chassis that becomes transparent at the flick of a switch, or a room that can be made instantly private, turning thin walls into sound absorbers. Yet these ideas are just the beginning of the story. If we can rapidly switch the wave speed, then completely new effects emerge. For example, changing the refractive index abruptly causes a wave to "reflect in time" - a paradoxical temporal analogue of the ordinary reflection we see and hear every day (e.g. the echo from a wall), but one that can cause the wave to gain energy. Other new effects arise if we can also change the refractive index differently at each point in space. With this control it becomes possible - for instance - to make a stationary object look like it is moving. Unlike true motion there is no restriction on this speed, and we can even mimic objects moving faster than light! Our research will develop new materials where the refractive index can be changed in time, exploring switchable functionality and the plethora of new wave effects that emerge when the material properties are varied rapidly. This is not always an easy thing to do and to avoid potential obstacles to our research we take a "wave agnostic" view, where we - in parallel - explore the effects of a time varying wave speed for airborne acoustic waves, mechanical vibrations, radio frequency waves, terahertz waves, and in optics. To illustrate the huge advantage of this approach, consider the time scales involved: "rapid" means the change must be imposed more quickly than the wave oscillates. For audible sound this is milliseconds, for visible light femtoseconds. We should use very different techniques in these two cases! In optics, special materials are subject to ultra-fast, high-intensity fields, while in acoustics we use electronically controlled transducers. Through considering different wave regimes we can implement a time varying wave speed by the most promising means, avoiding the limitations of any individual technique. Our program of research is split into four, first developing experiments to demonstrate rapid switching of acoustic, elastic, and electromagnetic wave speeds in time, and the theory required to design them. The second part pushes this work to the next stage, developing materials where the wave speed varies in both space and time, allowing us to e.g. mimic motion. Having developed these experimental and theoretical capabilities, the final two parts of the project explore new wave effects in these materials, specifically wave amplification and unusual materials where the wave can only propagate in one direction. While our research is a fundamental study into wave physics in time-varying materials, we predict multiple applications of this technology. Future communications (6G) is perhaps the simplest. This will need an enormous number of separately powered antennas to precisely direct beams of electromagnetic waves. But if we can rapidly change the reflective properties of a surface next to a single antenna, we can make it alone perform the function of these many different antennas, reducing energy requirements and complexity! Wave-based computing is a second example: like every physical process, the scattering of a wave from a material is equivalent to a computation. Although electromagnetic waves perform this computation very quickly - at the speed of light! - to use it as a "computer" we need to program it. The material properties are fixed, so the wave always scatters in the same way. If we can switch the material properties, we can program it and create a new class of high-speed computational devices based on wave-scattering.

In this Turing Artificial Intelligence Acceleration Fellowship, I will focus on artificial intelligence for medical treatments and therapies. I take the view that AI is a question on how to realise artificial systems that solve practical problems currently requiring human intelligence to solve, such as those solved by clinicians, nurses and therapists. Critical care is high risk and highly invasive environment caring for the sickest patients at greatest risk of death. Patients within this environment are highly monitored, enabling sudden changes in physiology to be attended to immediately. In addition, this monitoring requires a heavier staffing ratio (often 1:1 nursing; 1:8 medical) and variances in human factors and non-technical pressures (e.g. staffing, skill-mix, finances) leads to critical care delivery being disparate. AI in healthcare is a hard problem as, due to the diversity and variability of human nature, systems have to cope with unexpected circumstances when solving perceptual, reasoning or planning problems. Crucially, AI has two facets: Understanding from data, and Agency. While rapid strides have been made on learning from data, e.g. how to make medical diagnosis more precise and faster than human experts, there is little work on how to carry on after the diagnosis, e.g. which therapy and treatment to conduct. The latter requires agency and has seen fewer applications as it is a harder problem to solve. My clinical partners and I want to develop the required AI algorithms that can learn and distil the best plan of action to treat a specific patient, from the expert knowledge of clinicians. We will focus on an area of AI called RL that has been successful in enabling robots and self-driving cars to learn a form of autonomous agency. We want to transform these methods into the healthcare domain. This will require the development of new RL algorithms, able to efficiently understand the state of a patient from noisy and ambiguous hospital data. The system will not only learn to recommend interventions such as prescribing drugs and changing dosages as needed per patient but to make these recommendations in a manner that is meaningful to the clinical decision-makers and helps them make the best final decision on a course of action. The methods developed as part of this project can be used in different applications beyond healthcare. Many sectors within industry, such as aerospace, or energy, deal with similar bottlenecks. These are highly regulated environments, with great need for decisions making support, but a scarcity of highly skilled human experts. With sufficient data, our methods can be applied to these sectors as well, to distil the required human expertise and best practices from top experts, and use them to drive decision making all over the sector, for increased efficiency and safety.

Introducing porosity onto an aerofoil has been shown to have a significant influence on the boundary layer and provide significant reductions in its noise radiation. This proposal describes a multi-disciplinary research project aimed at understanding and exploiting the interactions between porous aerofoils and the boundary layers developing over them for the purpose of optimising noise reductions without compromising aerodynamic performance. The use of adaptive manufacturing technology will be investigated for providing the optimum porosity at different operating conditions.

This project will, for the first time, connect a detailed scientific understanding of the mechanisms of coatings failure with state-of-the-art machine learning to deliver a design framework for the optimization of protective coatings and nanocomposite materials. It will be game changing for an industry (paint) which is often taken for granted, despite its ubiquity - the screen you are looking at, the color of your car, the protection for the aircraft you fly in, the longevity of bridges, wind turbine masts and other infrastructure. Indeed, almost all materials are made suitable for purpose or given function by the application of coatings. In the UK there are over 10,000 employees involved in manufacturing coatings and the coatings industry directly contributes over £11bn to the economy, supporting UK manufacturing and construction sectors worth around £150bn. The annual costs of corrosion damage in the UK lies in the range of 2-3% of Gross National Product (~£60 bn, 2016) and leads to premature loss of amenity in infrastructure and equipment; hence to environmental damage through accelerated extraction and resource use. Protective organic coatings (i.e. paints) are highly cost effective in limiting early materials damage due to corrosion however these are complex products where the underlying mechanistic links between the formulation and performance are lacking. The increasing need to use environmentally sustainable materials, reduce time-to-market and increase performance requires detailed mechanistic understanding across functions and length scales from the molecular to the macroscopic. With brands such as Dulux, Hammerite and International, AkzoNobel are one of the world's largest manufacturers of protective and decorative coatings and have extensive manufacturing and research operations in the UK. AkzoNobel invests heavily in research, both in its global research hub for performance coatings in the NE of England as well as in UK universities. In particular the company (and its predecessor bodies) has collaborated in polymer science with the University of Sheffield, and in corrosion protection with The University of Manchester, for over 30 years. This prosperity partnership between EPSRC and AkzoNobel/ International Paint with the Universities of Manchester and Sheffield, will enable for the 1st time, a fundamental mechanistic understanding of how the performance of protective organic coatings arises - essentially it will tell us "how paint works". The scope of the program is well beyond the capacity of an individual company, institution or funder and, hence, the collaborative partnership is essential in order to tackle this problem head-on. Success will allow industry to side-step the current trial-and-error approaches and to incorporate digital design (i.e. Industry 4.0) into the development of paints and similar nanocomposite materials resulting in the confidence to utilize sustainable materials, comply with legislative and customer drivers and maintain and extend performance in more extreme environments. Overall the project will deliver understanding and tools that underpin the rapid-to-market development of environmentally sustainable protective organic coatings and nanocomposites by rational design.

Cybersecurity threats are causing damage to business and wider society and, if left unchecked, these threats will continue to grow. Poorly designed software is a significant source of cyber security vulnerabilities. Current software development practice relies heavily on an iterative build-test-fix approach to software correctness and, while testing of software is essential, it is very time-consuming and usually incomplete. A further weakness of the iterative build-test-fix approach is that it often results in design faults being discovered long after they were introduced in the development lifecycle - making them very expensive to fix once discovered. Formal methods are a body of mathematically-based techniques for design and verification of software that are more rigorous and systematic than build-test-fix, leading to better software designs with reduced bugs and vulnerabilities. Our vision is the transformation of security system development from an error-prone, iterative build-test-fix approach to a correctness-by-construction (CxC) approach whereby formal methods guide the design of software in such a way that it satisfies its specification by construction. The impact of this will be to reduce overall development costs, while increasing trustworthiness, of security-critical systems. Systems are designed by humans and used by humans. Formal methods are challenging to use for many software developers we will developed tools that reduce barriers to their deployment. Our tools will support developers to engage with wider stakeholders to elicit and validate requirements. Many secure systems rely on assumptions about the behaviour of trusted and untrusted users but often these assumptions are not clearly understood or defined. Our research will incorporate formal constraints on user data and actions and vulnerabilities in data integrity resulting from user behaviour in modelling and verification. Even if software has been verified correct, it is likely to be running on hardware that is vulnerable to cyber-attack because of poor memory protection. Today's open connected computing platforms allow hardware vulnerabilities to be exploited at scale and capability hardware has been proposed as an approach to reducing hardware vulnerabilities. Capability hardware, such as the CHERI architecture, provides a range of memory protection features, to enforce secure data operations and avoid incorrect or malicious manipulations of data. When using formal methods, we develop software that enforces secure data operations and thus, in principle, additional hardware enforcement is not required. However, securely-developed software is still likely to be executing in a context in which other code may be accidently or maliciously violating data access disciplines which would undermine the securely-developed code. By using capability hardware, we get enforcement of secure data operations on other code, avoiding the need to worry about interference by code over which we have no control. Our project will incorporate capability hardware features in to the formal design approach by developing high level design abstractions that capture properties of data operations appropriate for designing and verifying at higher abstraction levels. Our research will be guided and validated by a range of security-critical industrial case studies with support from our industrial partners (Airbus, Arm, Altran, AWE, Galois, L3Harris, Northrop Grumman, Thales). Key outcomes of HD-Sec will be: . An integrated toolchain to support the CxC approach for design of security systems . Sound high-level abstractions that facilitate exploitation of capability hardware in software design . A functioning prototype application designed using our CxC tools and running on capability hardware