Le 10/09/2019 par maxime_ogier :

This is an announcement from Chris Dent, University of Edinburgh. ----------------------- Dear all, There is a Postdoctoral Fellow opportunity at Edinburgh on the above topic, through the Train@Ed scheme part-funded by the European Union. This is available to candidates holding a PhD, who have not worked in the UK for more than 1 of the last 3 years, and who have no more than 10 years of postdoctoral experience. Appointment is on the normal Edinburgh postdoc scale, with other terms and conditions improved from the normal. See https://www.ed.ac.uk/human-resources/jobs/train-ed/train-ed-application-process for more details. The topic of research is flexible within the overall field ? please see below for a general description, and some possible work package ideas (which are not exhaustive ? other good ideas are welcome). Detailed proposals are co-developed with a host academic ? so if you are interested please contact Chris Dent (Reader in Industrial Mathematics, chris.dent@ed.ac.uk) with a full CV and outline of the topics which interest you as soon as possible, certainly by the end of September. The deadline for eventual proposals is end October. The Fellow would be hosted in the School of Mathematics ? however interdisciplinary collaboration is a key part of this scheme, so provided there is a clear demonstrated interest in a relevant mathematical science-related topic it is possible to consider applicants from a wide range of backgrounds, and we can suggest relevant academic collaborators. Full details of the scheme may be found at https://www.ed.ac.uk/human-resources/jobs/train-ed. Best regards Chris -- Decarbonising energy systems, while maintaining affordability and security of supply, is one of the key issues facing today?s society. Indeed the degree of ambition of UK carbon objectives has been prominent in recent weeks, following the government?s amendment of the Climate Change Act to a new target of zero carbon by 2050. Mathematical models are used widely to process the large amounts of data involved in specification of possible futures for the energy system, and thus to understand the impacts of new policies, technologies and market arrangements on the planning and operation of energy systems. Examples include wide use of MARKAL and TIMES in UK and Scottish energy policy, ESME by the Energy Technologies Institute and Energy Systems Catapult, and the recent major IDLES grant to Imperial College. Specialist data science expertise is not well integrated into the mainstream energy research and practice. This project will collaborate with applied energy modellers in government, industry and research, to promote more efficient computation, better assessment of uncertainty in the relationship between model outputs and the world, and develop new modelling approaches involving co-design of system modelling and uncertainty considerations to maximise learning about the real world. Possible academic collaborators include Chris Dent, Mike Allerhand, Miguel Anjos, Lars Schewe, Amy Wilson (Maths); Harry van der Weijde (Engineering); Mark Winskel (Social and Political Sciences). -- The following example work packages are suggested. The final scheme of work will depend on the interests of the candidate for the Train@Ed position. Sensitivity analysis in whole energy system models. Working in existing modelling frameworks, develop schemes for identifying key drivers of results. Pilot work with Scottish Government suggests that successively applying the following will provide a means of narrowing the range of inputs of interest: one-at-a-time; global sensitivity methods such as Morris; and statistical emulation. Computation for stochastic model formulations. Real energy systems evolve under conditions of substantial uncertainty over all aspects of planning background. To produce realistic results, models must therefore consider multiple possible future backgrounds. For efficient computation, some form of decomposition of different scenarios is required, whether by formal decomposition of optimisation subproblems, or by statistical emulation of operational subproblems within the overall modelling framework. Combined design of system modelling and uncertainty considerations. Typically, specialists in uncertainty management work with system models which have already been designed before their involvement begins. This package will investigate co-design of the two aspects, with the aim of maximising learning about the real system. It will draw on ideas from the ?Evidence Based Decisions? programme at the Isaac Newton Institute in July 2019. Communication of evidence. As well as producing high quality evidence, it is also critically important that this be communicated well to a non-specialist audience, and conversely that the analysts are carrying out modelling which represents the true interests of the decision makers. Communicating scientifically subtle evidence about complex systems is an area of research in its own right, in which collaboration involving social scientists and psychologists is necessary. -- Dr. Chris Dent SMIEEE FORS CEng Chancellor?s Fellow and Reader in Industrial Mathematics, School of Mathematics, University of Edinburgh Turing Fellow at the Alan Turing Institute for Data Science School of Mathematics, University of Edinburgh, James Clerk Maxwell Building, The King''s Buildings, Peter Guthrie Tait Road, Edinburgh EH9 3FD. UK. Tel: +44 131 650 5064 Office: JCMB 6222