In the last decade there has been an explosion in artificial intelligence research in which artificial neural networks, emulating biological brains, are used to solve problems ranging from obstacle avoidance in self-driving cars to playing complex strategy games. This has been driven by mathematical advances and powerful new computer hardware which has allowed large 'deep networks' to be trained on huge amounts of data. For example, after training a deep network on 'ImageNet' - which consists of over 14 million manually annotated images - it can accurately identify the content of images. However, while these deep networks have been shown to learn similar patterns of connections to those found in the parts of our brains responsible for early visual processing, they differ from real brains in several important ways, especially in how the individual neurons communicate. Neurons in real brains exchange information using relatively infrequent electrical pulses known as 'spikes', whereas, in typical artificial neural network models, the spikes are abstracted away and values representing the 'rates' at which spikes would be emitted are continuously exchanged instead. However, neuroscientists believe that large amounts of information is transmitted in the precise times at which spikes are produced. Artificial 'spiking neural networks' can harness these properties, making them useful in applications which are challenging for current models such as real-world robotics and processing data with a temporal component, such as video. However, spiking neural networks can only be used effectively if suitable computer hardware and software is available. While there is existing software for simulating spiking neural networks, it has mostly been designed for studying real brains, rather than building AI systems. In this project, I am going to build a new software package which bridges this gap. It will use abstractions and processes familiar to machine learning researchers, but with techniques developed for brain simulation, allowing exciting new SNN models to be used by AI researchers. We will also explore how spiking models can be used with a special new type of sensors which directly outputs spikes rather than a stream of images. In the first phase of the project, I will focus on using Graphics Processing Units to accelerate spiking neuron networks. These devices were originally developed to speed up 3D games but have evolved into general purpose devices, widely used to accelerate scientific and AI applications. However, while these devices have become incredibly powerful and are well-suited to processing lots of data simultaneously, they are less suited to 'live' applications such as when video must be processed as fast as possible. In these situations, Field Programmable Gate Arrays - devices where the hardware itself can be re-programmed - can be significantly faster and are already being used behind the scenes in data centres. In this project, by incorporating support for FPGAs into our new software, we will make these devices more accessible to AI researchers and unlock new possibilities of using biologically-inspired spiking neural networks to learn in real-time. As well as working on these new research strands, I will also dedicate time during my fellowship to advocate for research software engineering as a valuable component of academic institutions, both via knowledge exchange and research funding. In the shorter term, I will work to develop a community of researchers involved in writing software at Sussex by organising an informal monthly 'surgery' as well as delivering specialised training on programming Graphics Processing Units and more fundamental computational and programming training for new PhD students. Finally, I will develop internship and career development opportunities for undergraduate students, to gain experience in research software engineering.

Materials discovery feeds scientific and technological progress. Quantum materials host collective phenomena that defy a semi-classical description, for example because they arise from strong correlations or involve topological order. The diversity of these collective phenomena, their reach into practicable temperature regions and their tunability enable new technologies. Foremost among them is superconductivity, a macroscopic quantum phenomenon with multiple applications ranging from powerful magnets used in MRI scanners, fusion reactors and particle accelerators to lightweight motors and generators, low-noise rf filters, low-power electronics, and quantum devices used in sensing or computing. In most superconductors, the required electronic interactions are produced by dynamic lattice distortions. Alternatively, these interactions can be caused by more complex quantum processes similar to those which give rise to magnetism. Such unconventional 'superconductivity without phonons' is associated with a rich range of properties, some of which are highly desirable, such as resilience to high magnetic fields, current densities or temperatures. In this project, we investigate the drivers of the unusual superconducting and normal states in four material families, building on our recent breakthroughs and discoveries: (i) iron-germanide superconductors YFe2Ge2 and LuFe2Ge2, (ii) moderate heavy fermion compounds CeNi2Ge2 and CePd2Si2, (iii) the high pressure Kondo lattice superconductor CeSb2, (iv) quasiperiodic host-guest structures such as high pressure Bi, Sb and Ba. YFe2Ge2 in family (i) and CeNi2Ge2 in family (ii) form close to the border of magnetism at low temperature but just on the paramagnetic side, whereas their isoelectronic sister materials LuFe2Ge2 and CePd2Si2 order magnetically. High pressure CeSb2 (iii) displays robust superconductivity at magnetic fields that appear too high to allow spin singlet Cooper pairs. The quasiperiodic materials (iv) can host a low frequency sliding mode which dramatically affects normal state properties and causes unusually strong electron-phonon coupling. Because these materials differ in many details but also share common phenomenology, new insights will arise from studying them in one coherent programme. Fuelled by the clean, high quality samples that our recent crystal growth advances have produced, the programme leverages strong input from multiple project partners. These augment our local high field, high pressure measurements with specialised spectroscopic, thermodynamic and transport techniques. Prominent theory support will examine experimental findings to answer key research questions concerning (a) the role of soft modes, whether vibrational, magnetic or otherwise, (b) the origin of non-Fermi liquid signatures in transport and the notion of Planckian dissipation in correlated metals, (c) the nature and tunability of superconducting pairing interactions, and (d) the nature and gap structure of the superconducting state itself. These are hard but timely questions: 40 years after the discovery of the first unconventional superconductor, CeCu2Si2, the nature of its superconducting state is again under intense scrutiny, and the first oxide superconductor to be found outside the copper-oxide family, Sr2RuO4, is likewise hotly debated. The new superconductors listed above significantly widen the range of clean materials in which these fundamental questions can be studied effectively. The resulting insights help guide the search for further new unconventional superconductors in the vast space of materials, and studying these new materials in turn produces new insights and more precise guiding principles. There is scope and need for improving the success rate of these searches by leveraging computer modelling, which will gather momentum as the programme unfolds, eventually leading the way to functional quantum materials with practically useful properties.

Risk is the potential of experiencing a loss when a system does not operate as expected due to uncertainties. Its assessment requires the quantification of both the system failure potential and the multi-faceted failure consequences, which affect further systems. Modern industries (including the engineering and financial sectors) require increasingly large and complex models to quantify risks that are not confined to single disciplines but cross into possibly several other areas. Disasters such as hurricane Katrina, the Fukushima nuclear incident and the global financial crisis show how failures in technical and management systems cause consequences and further failures in technological, environmental, financial, and social systems, which are all inter-related. This requires a comprehensive multi-disciplinary understanding of all aspects of uncertainty and risk and measures for risk management, reduction, control and mitigation as well as skills in applying the necessary mathematical, modelling and computational tools for risk oriented decision-making. This complexity has to be considered in very early planning stages, for example, for the realisation of green energy or nuclear power concepts and systems, where benefits and risks have to be considered from various angles. The involved parties include engineering and energy companies, banks, insurance and re-insurance companies, state and local governments, environmental agencies, the society both locally and globally, construction companies, service and maintenance industries, emergency services, etc. The CDT is focussed on training a new generation of highly-skilled graduates in this particular area of engineering, mathematics and the environmental sciences based at the Liverpool Institute for Risk and Uncertainty. New challenges will be addressed using emerging probabilistic technologies together with generalised uncertainty models, simulation techniques, algorithms and large-scale computing power. Skills required will be centred in the application of mathematics in areas of engineering, economics, financial mathematics, and psychology/social science, to reflect the complexity and inter-relationship of real world systems. The CDT addresses these needs with multi-disciplinary training and skills development on a common mathematical platform with associated computational tools tailored to user requirements. The centre reflects this concept with three major components: (1) Development and enhancement of mathematical and computational skills; (2) Customisation and implementation of models, tools and techniques according to user requirements; and (3) Industrial and overseas university placements to ensure industrial and academic impact of the research. This will develop graduates with solid mathematical skills applied on a systems level, who can translate numerical results into languages of engineering and other disciplines to influence end-users including policy makers. Existing technologies for the quantification and management of uncertainties and risks have yet to achieve their significant potential benefit for industry. Industrial implementation is presently held back because of a lack of multidisciplinary training and application. The Centre addresses this problem directly to realise a significant step forward, producing a culture change in quantification and management of risk and uncertainty technically as well as educationally through the cohort approach to PGR training.

The epidermis is the largest, most complex epithelial tissue in the human body and its mechanical integrity is vital in protecting the human body from harm. Keratinizing skin disorders are a large group of highly debilitating, difficult-to-manage hereditary skin conditions that present major treatment challenges in the clinic. Collectively these diseases affect ~1 in 2000 people, but because they are individually quite rare very little progress has been made towards developing effective treatments. The quality-of-life impacts for these patients are life-long and can be devastating, thus, the overall healthcare and societal burden is very great. This proposal aims to begin to tackle the major challenge of developing effective, long-term treatments for these conditions. Most keratinizing disorders arise from single nucleotide mutations in one of several critically important structural molecules of the skin, which leads to the development of weakened, thick skin - epidermal fragility and hyperkeratosis. The great challenge in developing treatments for these disorders lies in selectively repairing or silencing the causative mutation. RNA interference (RNAi) is a remarkable natural cellular process that uses a unique class of molecules, called small interfering RNA (siRNA), to specifically and potently control gene activity. Harnessing the therapeutic potential of the RNAi pathway, several siRNAs that specifically target keratinizing skin disorder mutations have been identified. Unfortunately, the physical properties of siRNA molecules, they are very large and carry a negative charge, make epidermal delivery difficult. One of the major goals of this research proposal is to develop a patient-friendly way to delivery siRNA into the skin. We will use a unique in vivo reporter model, that produces a visually-trackable enzyme and facilitates real-time monitoring, to develop clinically-viable siRNA skin delivery methods. Once we have an efficient and effective mode of delivery, we will use it to test whether disease targeting siRNA treatments alleviate disease symptoms in an in vivo keratinizing skin disorder model displaying symptoms similar to the human disease. Our findings will provide the preclinical evidence required to progress siRNA therapeutics into clinical trials and, ultimately, patient prescribed treatments. Because each condition is individually rare and multiple mutation-specific siRNAs will be required to treat each patient population, we feel that it is important to also explore the possibility to developing one common treatment for all of these disorders. New medicines aimed at a common disease feature, like hyperkeratosis, rather than the specific mutant gene, may allow treatment of several keratinizing disorders regardless of the genetic abnormality. Unfortunately, we do not understand why hyperkeratosis develops or how to stop it. Therefore, the other major goal of this research proposal is to define the molecules and identify the biological pathways that cause a single gene mutation to develop into the phenotypic end product of weak, thickened, blistering, painful skin. We will do this by examining unique in vivo keratinizing skin disorder models, that develop hyperkeratosis, using cutting edge molecular profiling techniques to assemble an in-depth inventory of the individual cellular components that are present in normal but not diseased states, and vise versa. This will enable us to identify the molecular mechanisms that trigger hyperkeratosis and begin to understand how hyperkeratosis is regulated. It is our hope that these findings will seed the development of a generic treatment for most, if not all, keratinizing skin disorders.

Of the animal groups for which comprehensive assessments have been made, amphibians rank as the most threatened major taxon, with nearly three times as many amphibian species threatened with extinction as bird species. Disease has been identified as one of the major contributors to amphibian declines and extinctions, with one pathogen singled out as the most dangerous amphibian disease identified to date. Batrachochytrium dendrobatidis, a chytridiomycete fungus, has been implicated in mass mortality events, population declines and species extinctions around the globe. The Amphibian Conservation Summit recently organized by the IUCN recognized that the current state of knowledge regarding this fungus is manifestly inadequate. In response to this research gap, the Summit produced a Declaration calling for immediate research to determine the distribution of the disease, how this distribution was achieved and the consequences of invasion of the pathogen for local amphibian communities. Funded by NERC, we have completed three years of monitoring B. dendrobatidis and chytridiomycosis across Europe and have shown for the first time that infection is widespread, is present in the wild in the UK, and is unfortunately present in the Mallorcan midwife toad, one of the most critically endangered species of amphibian. We now need to add to our current descriptions of the prevalence of infection and test hypotheses generated by these patterns in order to ascertain the processes by which B. dendrobatidis causes populations to collapse. We will do this by using a combination of field surveys and laboratory experiments in order to understand how the dynamics of infection in natural populations lead to extinction. Firstly, we will intensively survey five focal study sites where the disease is present, but is causing different effects in the amphibian populations. We will sample these communities over three years in order to track how infection moves through the various species, and how infection differentially affects larval (tadpole), metamorph and adult stages within a community of amphibians. Using our newly developed environmental molecular assay, we are now able to test the density of infectious stages in the environment. This gives us the ability to directly measure the exposure-levels of amphibians and to assess whether the fungus can persist in the environment in the absence of its host. The idea that there are multiple reservoirs of infection is very important, as extinction is more likely in susceptible species when pathogen 'spillover' occurs from disease reservoirs. We will therefore test these ideas in laboratory and mesocosm systems where we are able to manipulate the density and type of potential reservoirs of B. dendrobatidis. Results from our first NERC grant have shown that there is strong evidence that the international trade in amphibians (specifically Xenopus and North American bullfrogs) is causing multiple introductions of B. dendrobatidis into the UK and mainland Europe. We need to know whether there is variation between these different strains of the pathogen in their ability to cause disease, and to test this idea we will perform challenge experiments in our model species, the common toad Bufo bufo. Once these comparisons have been completed, we will develop mathematical epidemiological models with the explicit goal of predicting the dynamics of disease emergence across several scales. We have strong evidence that temperature is a key determinant of chytrid-driven mortality, and we will develop statistical models to determine the potential effects of global-warming on the distribution of the disease within Europe. Our aim is that these short and long term research goals will eventually enable us to control fungal spread or manage amphibian populations in order to prevent the population declines that are associated with the emergence of this pathogen.