# Workshop Programme

## for period 27 - 27 March 2013

### Stochastic and Statistical Models at the Interface of Modern Industry and Mathematical Sciences

27 - 27 March 2013

Timetable

 Wednesday 27 March 09:00-09.30 Registration and Coffee 09:30-09:35 Welcome from INI Director, John Toland 09:35-09.40 Opening Remarks by Florian Theil, University of Warwick 09:40-10:20 Muench, A (OCIAM) Math-in-Industry activities in OCIAM (and beyond) Sem 1 In this presentation, we will start from an overview over OCIAMs past and current industrial activities, with an emphasis on the opportunities for younger researchers and the general successful approach to collaboration with industry. A key role for OCIAM are the regular study groups, that were pioneered in Oxford and have now spread around the globe. Actual case studies are also discussed. 10:20-11.00 Davies, P (Thales UK) What Complexity Science offers to Industry Sem 1 This presentation outlines a number of facets of complexity science that is seen as having future potential to a diverse range of industry sectors. Some pointers are given for the complex mathematical modelling skills required by industry, that postgraduate academic engagement can provide. A specific instance is described of applying complexity analysis to the conduct of complex industrial projcts themselves, as disctinct from the underpining technology. 11:00-11:20 Coffee break 11:20-12:00 Mitas, C (RMS) Loss Models of Catastrophes: A Short Introduction and Simple Examples Sem 1 We offer a short introduction into the main ideas and business cases of catastrophe modeling. We also provide a rudimentary description of some statistical tools of interest to cat modeling 12:00-12:30 Guo, Q (Imanova) Positron Emission Tomography Imaging in the Pharmaceutical Industry Sem 1 Human in vivo molecular imaging with positron emission tomography (PET) enables a new kind of 'precision pharmacology', able to address questions central to drug development. Biodistribution studies with drug molecules carrying positron-emitting radioisotopes can test whether a new chemical entity reaches a target tissue compartment (such as the brain) in sufficient amounts to be pharmacologically active. Competition studies, using a radioligand that binds to the target of therapeutic interest with adequate specificity, enable direct assessment of the relationship between drug plasma concentration and target occupancy. Quantitative mathematical techniques run through the whole process, from optimizing adaptive study designs, through image reconstruction, image registration, image segmentation, tracer kinetic analysis, probe development, PK/PD analyses and statistical analysis. We illustrate the application of mathematics to this area of imaging science by focusing on how bio-mathematical techniques can help predict, from in silico and in vitro data, which molecules will possess the right characteristics to produce a signal once labeled and injected into man. 12:30-13:40 Lunch and posters 13:40-14:20 Briggs, K (BT Technology) A tale of two tails Sem 1 In the design and modelling of industrial processes, a common requirement is (loosely expressed) that "the worst case is not too bad", or, more precisely, "the probability of very bad cases occurring is sufficiently small". Mathematically, this means that for some random variable X depending on parameters alpha, Prob[X_alpha>x] must be shown to be less than epsilon for some specified x and epsilon (typically x is large, and epsilon small). I shall describe two practical problems of this type. The first concerns trip-planning on transport systems with random delay, where the requirement is to arrive close to a specific target time, with the probability of arriving very late being very small. The second (which is work in progress), concerns the management of radio spectrum in the case of many small devices all sharing the same channel. Here the requirement is that the probability of seriously bad interference occurring is less than some specified bound. 14:20-15:00 Cassels, R (SELEX ES Ltd) Challenges in Radar ESM Processing Sem 1 Radar Electronic Support Measures (ESM) systems are designed to detect, identify, locate and track radar systems by their emissions. This introduces several instances of problems that do not yield particularly well to classical mathematical analysis techniques. The proliferation of advanced radar systems and the increasingly cluttered spectrum means that industry must look to new mathematical frameworks in order to achieve performance levels required for operational equipment. This presentation provides an introduction to ESM systems, the context in which they are used and some of the challenges that are faced in the processing and analysis of received data. 15:00-15:20 Tea break 15:20-15:50 Sullivan, T (University of Warwick) Optimal Uncertainty Quantification Sem 1 Uncertainty Quantification lies at the interface of applied mathematics, probability, computation and applied sciences, and has been called the end-to-end study of the reliability of scientific inferences.'' It is the understanding of how information (or uncertainty) propagates through systems to produce information (or uncertainty) about output quantities of interest (e.g. structural failure risks or financial portfolio returns), and corresponding inverse problems. In many real-world applications, this information propagation spans multiple components or scales and is probabilistic in nature, but is complicated by non-negligible uncertainty about which probability distributions and models are the correct'' ones. In the Optimal UQ (OUQ) framework, these problems are formalized as optimization problems over infinite-dimensional feasible sets of probability measures and transfer functions. 15:50-16:20 Iglesias, MA (University of Warwick) Bayesian data assimilation for uncertainty quantification Sem 1 Modelling and simulation of physical systems require the especification of parameters which are often uncertain. In geophysical applications, for example, large uncertainty in model predictions arises from the lack of information of geologic properties. When observational data of the model dynamics are available, data assimilation techniques can be used to combine model and data to reduce and quantify the uncertainty. The quantification of uncertainty in predictions is essential for optimal design and decision making. In this talk I will discuss some general aspects of Bayesian data assimilation with applications to subsurface modelling. Chair: Colm Connaughton 16:20-17:00 Discussion Session: How do we strengthen the links between academia and industry in light of today's presentations? Sem 1 17:00-18:00 Wine Reception