BEGIN:VCALENDAR VERSION:2.0 PRODID:-//Computational Optimisation Group - ECPv6.15.11//NONSGML v1.0//EN CALSCALE:GREGORIAN METHOD:PUBLISH X-WR-CALNAME:Computational Optimisation Group X-ORIGINAL-URL:https://optimisation.doc.ic.ac.uk X-WR-CALDESC:Events for Computational Optimisation Group REFRESH-INTERVAL;VALUE=DURATION:PT1H X-Robots-Tag:noindex X-PUBLISHED-TTL:PT1H BEGIN:VTIMEZONE TZID:UTC BEGIN:STANDARD TZOFFSETFROM:+0000 TZOFFSETTO:+0000 TZNAME:UTC DTSTART:20110101T000000 END:STANDARD END:VTIMEZONE BEGIN:VEVENT DTSTART;TZID=UTC:20130822T110000 DTEND;TZID=UTC:20130822T110000 DTSTAMP:20260418T131427 CREATED:20170124T102140Z LAST-MODIFIED:20170124T102140Z UID:574-1377169200-1377169200@optimisation.doc.ic.ac.uk SUMMARY:Seminar: Added value of scenario tree based stochastic optimization in long and medium term planning of hydro power systems DESCRIPTION:Title: Added value of scenario tree based stochastic optimization in long and medium term planning of hydro power systemsSpeaker: Dr. Georg OstermaierAffiliation: Location: Room 217 Huxley BuildingTime: 11:00am \nAbstract. The changes in the dynamics of power prices in Germany within the last few years have implied significant decreases of revenues of pumped storage hydro systems. The earnings from daily peak/offpeak price spreads have declined whereby at the same time long term management of large reservoirs has become even more difficult due hardly predictable power price and inflow evolutions. Stochastic Optimization has been applied for hydro system management for a long time already\, but only recently operators expect additional benefits given the changed and even more uncertain power market situation. Besides stochastic dual dynamic programming\, scenario tree based methods are meanwhile applied\, which so far have hardly been used for complex hydro systems due to the curse of dimensionality in stochastic programming. We apply efficient discretization methods for the generation of multi-dimensional scenario trees of power prices and inflows\, based on moment matching and multinomial distributions. Furthermore\, by using increased hardware efficiency it is meanwhile possible to set up and solve scenario tree based stochastic optimization models for even complex pumped storage systems. Ex post analyses have shown that outperformance over deterministic optimization in the range of 1-5 percent is achievable. \nAbout the speaker. Georg Ostermaier is founder\, owner and managing director of Decision Trees GmbH\, a Munich based firm focussing on the practical application of mathematical and stochastic programming in the European Energy industry. Georg received his graduate degree in Electrical Engineering from the Technical University of Munich and his PhD in Operations Research from the University of St.Gallen\, Switerland. Since 2006 he and his team have been developing stochastic optimization software systems for thermal power generation portfolios\, hydro power generation systems\, gas storage and gas contract valuation and gas procurment portfolio optimisation. Decision Trees GmbH has today a solid customer base in Germany\, Austria\, Switzerland\, Norway and the United Kingdom and has proven to contribute to enhanced profits for energy producers in practice through stochastic optimization. URL:https://optimisation.doc.ic.ac.uk/event/seminar-added-value-of-scenario-tree-based-stochastic-optimization-in-long-and-medium-term-planning-of-hydro-power-systems/ END:VEVENT BEGIN:VEVENT DTSTART;TZID=UTC:20130812T140000 DTEND;TZID=UTC:20130812T140000 DTSTAMP:20260418T131427 CREATED:20170124T102141Z LAST-MODIFIED:20170124T102141Z UID:575-1376316000-1376316000@optimisation.doc.ic.ac.uk SUMMARY:Seminar: Managing with Incomplete Inventory Information DESCRIPTION:Title: Managing with Incomplete Inventory Information Speaker: Prof. Suresh P. SethiAffiliation: Center for Intelligent Supply Networks – University of Texas at DallasLocation: Room 217-218 Huxley BuildingTime: 2:00pm \nAbstract. A critical assumption in the vast literature on inventory management has been that the current level of inventory is known to the decision maker. Some of the most celebrated results such as the optimality of base-stock policies have been obtained under this assumption. Yet it is often the case in practice that the decision makers have incomplete or partial information about their inventory levels. The reasons for this are many: Inventory records or cash register information differ from actual inventory because of a variety of factors including transaction errors\, theft\, spoilage\, misplacement\, unobserved lost demands\, and information delays. As a result\, what are usually observed are some events or surrogate measures\, called signals\, related to the inventory level. These relationships can provide the distribution of current inventory levels. Therefore\, the system state in the inventory control problems is not the current inventory level\, but rather its distribution given the observed signals. Thus\, the analysis for finding optimal production or ordering policies takes place generally in the space of probability distributions. The purpose of this talk is to review recent developments in the analysis of inventory management problems with incomplete information. \nAbout the speaker. Suresh P. Sethi is Eugene McDermott Professor of Operations Management and Director of the Center for Intelligent Supply Networks at The University of Texas at Dallas. He has written 7 books and published nearly 400 research papers in the fields of manufacturing and operations management\, finance and economics\, marketing\, and optimization theory. He teaches a course on optimal control theory/applications and organizes a seminar series on operations management topics. He initiated and developed the doctoral programs in operations management at both University of Texas at Dallas and University of Toronto. He serves on the editorial boards of several journals including Production and Operations Management and SIAM Journal on Control and Optimization. He was named a Fellow of The Royal Society of Canada in 1994. Two conferences were organized and two books edited in his honor in 2005-6. Other honors include: IEEE Fellow (2001)\, INFORMS Fellow (2003)\, AAAS Fellow (2003)\, POMS Fellow (2005)\, IITB Distinguished Alum (2008)\, SIAM Fellow (2009)\, POMS President (2012). URL:https://optimisation.doc.ic.ac.uk/event/seminar-managing-with-incomplete-inventory-information/ END:VEVENT BEGIN:VEVENT DTSTART;TZID=UTC:20130808T140000 DTEND;TZID=UTC:20130808T140000 DTSTAMP:20260418T131427 CREATED:20170124T102141Z LAST-MODIFIED:20170124T102141Z UID:576-1375970400-1375970400@optimisation.doc.ic.ac.uk SUMMARY:Seminar: Risk neutral and risk averse approaches to multistage stochastic programming with applications to hydrothermal operation planning problems DESCRIPTION:Title: Risk neutral and risk averse approaches to multistage stochastic programming with applications to hydrothermal operation planning problemsSpeaker: Dr. Wajdi TekayaAffiliation: Decision Trees GmbHLocation: Room 217 Huxley BuildingTime: 2:00pm \nAbstract. This talk gives an overview of risk neutral and risk averse approaches to multistage stochastic programming with applications to hydrothermal operation planning problems. In the first part of this talk\, we discuss risk neutral and risk averse approaches using coherent risk measures to multistage (linear) stochastic programming problems based on the Stochastic Dual Dynamic Programming (SDDP) method. We give a general description of the algorithm and present computational studies related to planning of the Brazilian interconnected power system.In the second part of this talk\, we discuss multistage programming with the data process subject to uncertainty. We consider a situation where the data process can be naturally separated into two components\, one can be modeled as a random process\, with a specified probability distribution\, and the other one can be treated from a robust (worst-case) point of view. We formulate this in a time consistent way and derive the corresponding dynamic programming equations. In order to solve the obtained multistage problem we develop a variant of the (SDDP) method. We give a general description of the algorithm and present computational studies related to planning of the Brazilian interconnected power system. \nAbout the speaker. Wajdi Tekaya is currently an HPCfinance postdoctoral fellow at Cambridge Systems Associates. He obtained his B.S. in industrial engineering from Tunisia Polytechnic School\, M.S. in Operations Research from Paris IX University\, Georgia Institute of Technology and Ph.D. in Operations Research from Georgia Institute of technology. His research interests are in computational approaches to stochastic programming with applications in energy and finance. URL:https://optimisation.doc.ic.ac.uk/event/seminar-risk-neutral-and-risk-averse-approaches-to-multistage-stochastic-programming-with-applications-to-hydrothermal-operation-planning-problems/ END:VEVENT BEGIN:VEVENT DTSTART;TZID=UTC:20130703T150000 DTEND;TZID=UTC:20130703T150000 DTSTAMP:20260418T131427 CREATED:20170124T102141Z LAST-MODIFIED:20170124T102141Z UID:577-1372863600-1372863600@optimisation.doc.ic.ac.uk SUMMARY:Seminar: Interdiction Games on Markovian PERT Networks DESCRIPTION:Title: Interdiction Games on Markovian PERT NetworksSpeaker: Eli GutinAffiliation: School of Industrial and Systems Engineering – Georgia Institute of TechnologyLocation: Room 218 Huxley BuildingTime: 3:00pm \nAbstract. In a stochastic interdiction game a proliferator aims to minimize the expected duration of a nuclear weapons development project\, while an interdictor endeavors to maximize the project duration by delaying some of the project tasks. We formulate static and dynamic versions of the interdictor’s decision problem where the interdiction plan is either pre-committed or adapts to new information revealed over time\, respectively. The static model gives rise to a stochastic program\, while the dynamic model is formalized as a multiple optimal stopping problem in continuous time and with decision-d ependent information. Under a Markov assumption\, we prove that the static model reduces to a mixed-integer linear program\, while the dynamic model reduces to a finite Markov decision process in discrete time that can be solved via efficient value iteration. We then generalize the dynamic model to account for uncertainty in the outcomes of the interdiction actions. We also discuss a crashing game where the proliferator can use limited resources to expedite tasks so as to counterbalance the interdictor’s efforts. The resulting problem can be formulated as a robust Markov decision process. \nAbout the speaker. Eli Gutin completed his MEng in Computing at Imperial College in 2012. His prize-winning final year project on “Interdiction Games on Markovian PERT networks” was supervised by Daniel Kuhn & Wolfram Wiesemann. A year later\, it was submitted for publication and is the subject of today’s talk. URL:https://optimisation.doc.ic.ac.uk/event/seminar-interdiction-games-on-markovian-pert-networks/ END:VEVENT BEGIN:VEVENT DTSTART;TZID=UTC:20130620T160000 DTEND;TZID=UTC:20130620T160000 DTSTAMP:20260418T131427 CREATED:20170124T102141Z LAST-MODIFIED:20170124T102141Z UID:578-1371744000-1371744000@optimisation.doc.ic.ac.uk SUMMARY:Seminar: Robust Data-Driven Approach in Decision Making Under Uncertainty DESCRIPTION:Title: Robust Data-Driven Approach in Decision Making Under UncertaintySpeaker: Grani Adiwena HanasusantoAffiliation: Department of Computing – Imperial College LondonLocation: Room 301 William PenneyTime: 4:00pm \nAbstract. We investigate robust data-driven approach in stochastic optimization problems where partial knowledge on the exogenous uncertainties is available to the decision maker. In contrast to the traditional model-based approach\, a data-driven approach requires no assumptions on the underlying distribution of exogenous uncertainties. Estimation of conditional expectation is achieved using kernel regression scheme which evaluates the cost function solely at historical observations. If sparse historical observations are available\, however\, the estimation is inaccurate and the resulting decision performs poorly in out-of-sample tests. To alleviate this unfavourable outcome\, we ‘robustify’ the decision against estimation errors by utilizing techniques from robust optimization. We show that the arising min-max problem can be reformulated as a tractable conic program. We further extend the proposed approach to multi-period settings and introduce an approximate dynamic programming framework that retains the tractability of the formulation and that is amenable to efficient parallel implementation. The proposed approach is tested across several application domains and is shown to outperform various non-robust schemes in terms of standard statistical benchmarks. \nAbout the speaker. Grani Hanasusanto is a PhD student at the Department of Computing\, Imperial College London\, under the supervision of Dr. Daniel Kuhn. He obtained the BEng (Hons) degree in Electrical and Electronic Engineering from Nanyang Technological University\, Singapore\, and the MSc degree in Financial Engineering from National University of Singapore. His research interests are in numerical and computational methods and their applications. URL:https://optimisation.doc.ic.ac.uk/event/seminar-robust-data-driven-approach-in-decision-making-under-uncertainty/ END:VEVENT BEGIN:VEVENT DTSTART;TZID=UTC:20130620T140000 DTEND;TZID=UTC:20130620T140000 DTSTAMP:20260418T131427 CREATED:20170124T102142Z LAST-MODIFIED:20170124T102142Z UID:579-1371736800-1371736800@optimisation.doc.ic.ac.uk SUMMARY:Seminar: Parallel block coordinate descent methods for huge-scale partially separable problems DESCRIPTION:Title: Parallel block coordinate descent methods for huge-scale partially separable problemsSpeaker: Martin TakacAffiliation: School of Mathematics – University of EdinburghLocation: CPSE Seminar roomTime: 2:00pm \nAbstract. In this work we show that randomized block coordinate descent methods can be accelerated by parallelization when applied to the problem of minimizing the sum of a partially block separable smooth convex function and a simple block separable convex function. We give a generic algorithm and several variants thereof based on the way parallelization is performed. In all cases we prove iteration complexity results\, i.e.\, we give bounds on the number of iterations sufficient to approximately solve the problem with high probability. Our results generalize the intuitive observation that in the separable case the theoretical speedup caused by parallelization must be equal to the number of processors. We show that the speedup increases with the number of processors and with the degree of partial separability of the smooth component of the objective function. Our analysis also works in the mode when the number of blocks being updated at each iteration is random\, which allows for modelling situations with variable (busy or unreliable) number of processors. We conclude with some encouraging computational results applied to huge-scale LASSO and sparse SVM instances. This is a joint work with Dr. Peter Richtarik\, University of Edinburgh. \nAbout the speaker. URL:https://optimisation.doc.ic.ac.uk/event/seminar-parallel-block-coordinate-descent-methods-for-huge-scale-partially-separable-problems/ END:VEVENT BEGIN:VEVENT DTSTART;TZID=UTC:20130619T140000 DTEND;TZID=UTC:20130619T140000 DTSTAMP:20260418T131427 CREATED:20170124T102142Z LAST-MODIFIED:20170124T102142Z UID:580-1371650400-1371650400@optimisation.doc.ic.ac.uk SUMMARY:Seminar: Performance-based regularization in mean-CVaR portfolio optimization DESCRIPTION:Title: Performance-based regularization in mean-CVaR portfolio optimizationSpeaker: Prof. Gah-Yi VahnAffiliation: Management Science and Operations – London Business SchoolLocation: Room 145 HuxleyTime: 2:00pm \nAbstract. Regularization is a technique widely used to improve the stability of solutions to statistical problems. We propose a new regularization concept\, performance-based regularization (PBR)\, for data-driven stochastic optimization. The goal is to improve upon Sample Average Approximation (SAA) in finite-sample performance while maintaining minimal assumptions about the data. We apply PBR to mean-CVaR portfolio optimization\, where we penalize portfolios with large variability in the constraint and objective estimations\, which effectively constrains the probabilities that the estimations deviate from the respective true values. This results in a combinatorial optimization problem\, but we prove its convex relaxation is tight. We show via simulations that PBR substantially improves upon SAA in finite-sample performance for three different population models of stock returns. We also prove that PBR is asymptotically optimal\, and further derive its first-order behavior by extending asymptotic analysis of M-estimators. This is joint work with Noureddine El Karoui (UC Berkeley Statistics) and Andrew EB Lim (NUS Business School) \nAbout the speaker. Gah-Yi Vahn is an Assistant Professor of Management Science and Operations at London Business School. She has a BSc (1st Class Hons. with Univ. Medal) from the University of Sydney (2007)\, an MA in Statistics (2011) and a PhD in Operations Research (2012) from the University of California\, Berkeley. Gah-Yi’s research interest is data-driven decision-making\, in particular optimization with complex\, high dimensional\, and/or highly uncertain data\, with applications to finance and operations management. URL:https://optimisation.doc.ic.ac.uk/event/seminar-performance-based-regularization-in-mean-cvar-portfolio-optimization/ END:VEVENT BEGIN:VEVENT DTSTART;TZID=UTC:20130614T140000 DTEND;TZID=UTC:20130614T140000 DTSTAMP:20260418T131427 CREATED:20170124T102142Z LAST-MODIFIED:20170124T102142Z UID:581-1371218400-1371218400@optimisation.doc.ic.ac.uk SUMMARY:Seminar: Distributionally robust control of constrained stochastic systems DESCRIPTION:Title: Distributionally robust control of constrained stochastic systemsSpeaker: Bart Van ParysAffiliation: Automatic Control Laboratory at Swiss Federal Institute of TechnologyLocation: Room 217-218 Huxley BuildingTime: 2:00pm \nAbstract. We investigate the control of constrained stochastic linear systems when faced with limited information regarding the disturbance process\, that is\, when only the first and second-order moments of the disturbance distribution are known. We employ two types of soft constraints to prevent the state from falling outside a prescribed target domain: distributionally robust chance constraints require the state to remain within the target domain with a given high probability\, while distributionally robust conditional value-at-risk constraints impose an upper bound on the state’s expected distance to the target domain conditional on that distance being positive. The attribute 'distributionally robust' reflects the requirement that the constraints must hold for all disturbance distributions sharing the known moments. We argue that the design of controllers for systems accommodating these types of constraints is both computationally tractable and practically meaningful for both finite and infinite horizon problems. The proposed methods are illustrated in the context of a wind turbine blade control design case study where flexibility issues play an important role and for which the distributionally robust constraints make sensible design objectives. \nAbout the speaker. Bart holds a BA degree in electrical engineering\, and a MA degree in applied/engineering mathematics\, both from the University of Leuven. Since September 2011 he has been a PhD student in the Swiss Federal Institute of Technology (ETH Zürich) under the supervision of Prof. Manfred Morari and Dr. Paul Goulart. URL:https://optimisation.doc.ic.ac.uk/event/seminar-distributionally-robust-control-of-constrained-stochastic-systems/ END:VEVENT BEGIN:VEVENT DTSTART;TZID=UTC:20130528T140000 DTEND;TZID=UTC:20130528T140000 DTSTAMP:20260418T131427 CREATED:20170124T102142Z LAST-MODIFIED:20170124T102142Z UID:582-1369749600-1369749600@optimisation.doc.ic.ac.uk SUMMARY:Seminar: CVaR Approximations for Minimax and Robust Convex Optimization DESCRIPTION:Title: CVaR Approximations for Minimax and Robust Convex OptimizationSpeaker: Prof. Huifu XuAffiliation: School of Engineering and Mathematical Sciences at City University of LondonLocation: Room 218 Huxley BuildingTime: 2:00pm \nAbstract. Conditional value at risk (CVaR) has been widely used as a risk measure in finance. In this work\, we propose to randomize decision variables of a deterministic parametric maximization problem and approximate the optimal (maximum) value by the CVaR of the randomized function. One of the main advantages of this approach is that computing CVaR is down to solving a one dimensional convex optimization problem even when the original problem is multi-dimensional and nonconvex. We apply the approximation scheme to a minimax (robust) optimization problem and a convex optimization problem with semi-infinite constraints indexed by uncertain parameters. In the latter application we use CVaR to approximate the semi-infinite constraint and then apply the Monte Carlo sampling method to discretize the CVaR approximated constraint. This approach is closely related to a popular randomization approach proposed by Calafiore and Campi where the continuum of uncertainty parameters is discretized through sampling. Error bounds for the optimal solutions of the approximate problems are derived under some moderate conditions and some numerical test results are reported. The proposed methods can be applied to distributional optimization where the distribution set is constructed through moments. \nAbout the speaker. Huifu Xu is a Professor of Operational Research in the School of Engineering and Mathematical Sciences at City University of London. Before joining City University\, he was a Senior Lecturer of Operational Research in the School of Mathematics at the University of Southampton. His expertise is in continuous optimization and operational research\, including developing numerical methods and underlying theory for continuous optimization problems\, particularly those involving uncertain data and/or equilibrium constraints. Over the past ten years\, he has been actively working on stochastic mathematical programs with equilibrium problems and is recently developing interest in robust approaches for stochastic optimization and equilibrium problems. Huifu has published about 60 papers most of which are in the international journals of optimization and operational research. URL:https://optimisation.doc.ic.ac.uk/event/seminar-cvar-approximations-for-minimax-and-robust-convex-optimization/ END:VEVENT BEGIN:VEVENT DTSTART;TZID=UTC:20130522T140000 DTEND;TZID=UTC:20130522T140000 DTSTAMP:20260418T131427 CREATED:20170124T102142Z LAST-MODIFIED:20170124T102142Z UID:583-1369231200-1369231200@optimisation.doc.ic.ac.uk SUMMARY:Seminar: Hard-to-Solve Bimatrix Games DESCRIPTION:Title: Hard-to-Solve Bimatrix GamesSpeaker: Prof. Bernhard von StengelAffiliation: Depatment of Mathematics – London School of Economics and Political ScienceLocation: CPSE seminar room (C615 Roderic Hill)Time: 2:00pm \nAbstract. A bimatrix game is a two-player game in strategic form\, a basic model in game theory. A Nash equilibrium is a pair of (possibly randomized) strategies\, one for each player\, so that no player can do better by unilaterally changing his or her strategy. In this talk\, which will introduce the main concepts and geometric tools\, we show that the commonly used Lemke-Howson algorithm for finding one equilibrium of a bimatrix game is exponential. The algorithm is a pivoting method similar to the simplex algorithm for linear programming. We present a class of square bimatrix games for which the shortest Lemke-Howson path grows exponentially in the dimension d of the game. We construct the games using pairs of dual cyclic polytopes with 2d facets in d-space. The paths are recursively composed so that their lengths grow like Fibonacci numbers. We also mention subsequent results and open problems in the area. \nAbout the speaker. Diploma in Mathematics from Aachen\, MSc in Computer Sciences at Austin/Texas (student of Edsger W. Dijkstra)\, PhD in Passau\, Habilitation in Munich\, Postdoc at Berkeley\, Tilburg and ETH Zurich (with a Heisenberg grant)\, at LSE Mathematics since 1998. Interested in algorithmic game theory longer than the research area has that name. URL:https://optimisation.doc.ic.ac.uk/event/seminar-hard-to-solve-bimatrix-games/ END:VEVENT BEGIN:VEVENT DTSTART;TZID=UTC:20130516T140000 DTEND;TZID=UTC:20130516T140000 DTSTAMP:20260418T131427 CREATED:20170124T102143Z LAST-MODIFIED:20170124T102143Z UID:584-1368712800-1368712800@optimisation.doc.ic.ac.uk SUMMARY:Seminar: Alternating Maximization: Unifying Framework for 8 Sparse PCA Formulations DESCRIPTION:Title: Alternating Maximization: Unifying Framework for 8 Sparse PCA FormulationsSpeaker: Dr. Selin AhipasaogluAffiliation: Singapore University of Technology and DesignLocation: CPSE seminar room (C615 Roderic Hill)Time: 2:00pm \nAbstract. Given a multivariate data set\, sparse principal component analysis (SPCA) aims to extract several linear combinations of the variables that together explain the variance in the data as much as possible\, while controlling the number of nonzero loadings in these combinations. In this paper we consider 8 different optimization formulations for computing a single sparse loading vector; these are obtained by combining the following factors: we employ two norms for measuring variance (L2\, L1) and two sparsity-inducing norms (L0\, L1)\, which are used in two different ways (constraint\, penalty). Three of our formulations\, notably the one with L0 constraint and L1 variance\, have not been considered in the literature. We give a unifying reformulation which we propose to solve via a natural alternating maximization (AM) method. Besides this\, we provide a package which contains implementations for various parallel architectures and briefly discuss how these algorithms can be used to achieve better object recognition in challenging data sets. \nAbout the speaker. Selin Damla Ahipasaoglu is an Assistant Professor at the Singapore University of Technology and Design. She received her PhD in 2009 from Cornell University and specialises in developing algorithms for large scale optimization problems\, in particular first-order methods for convex problems and applications in image processing. She held research positions at Princeton University and London School of Economics before joining SUTD. She is also a very keen teacher and an advocate of active and innovative classroom teaching for undergraduates. URL:https://optimisation.doc.ic.ac.uk/event/seminar-alternating-maximization-unifying-framework-for-8-sparse-pca-formulations/ END:VEVENT BEGIN:VEVENT DTSTART;TZID=UTC:20130417T170000 DTEND;TZID=UTC:20130417T170000 DTSTAMP:20260418T131427 CREATED:20170124T102143Z LAST-MODIFIED:20170124T102143Z UID:585-1366218000-1366218000@optimisation.doc.ic.ac.uk SUMMARY:Seminar: On the Relation between Forecast Precision and Trading Profitability of Financial Analysts DESCRIPTION:Title: On the Relation between Forecast Precision and Trading Profitability of Financial AnalystsSpeaker: Prof. Alex WeissensteinerAffiliation: DTU Management EngineeringLocation: Room 217 Huxley BuildingTime: 5:00pm \nAbstract. We analyze the relation between earning forecast accuracy and expected profitability of financial analysts. Modeling forecast errors with a multivariate Gaussian distribution\, a complete characterization of the payoff of each analyst is provided. In particular\, closed-form expressions for the probability density function\, for the expectation\, and\, more generally\, for moments of all orders are obtained. Our analysis shows that the relationship between forecast precision and trading profitability need not to be monotonic\, and that\, for any analyst\, the impact on his expected payoff of the correlation between his forecasts and those of the other market participants depends on the accuracy of his signals. Furthermore\, our model accommodates a unique full-communication equilibrium in the sense of Radner (1979). \nAbout the speaker. Alex finished his PhD in 2003 and received his “Habilitation” from the Leopold-Franzens University Innsbruck (Austria) in 2010\, where he worked as Assistant Professor. From 2010-2013 he worked atthe Free University of Bozen-Bolzano(Italy)\, School of Economics and Management.Hisfields of research are in the intersection of finance\, optimization and mathematical modeling. Themain topics covered by his scientific publications are: asset allocation\, asset-liability management\, stochastic (linear-) programming(SLP)\, and neural networks. Since February 2013 he is working as Professor for Financial Engineering at the Technical University of Denmark (DTU). URL:https://optimisation.doc.ic.ac.uk/event/seminar-on-the-relation-between-forecast-precision-and-trading-profitability-of-financial-analysts/ END:VEVENT BEGIN:VEVENT DTSTART;TZID=UTC:20130411T140000 DTEND;TZID=UTC:20130411T140000 DTSTAMP:20260418T131427 CREATED:20170124T102143Z LAST-MODIFIED:20170124T102143Z UID:586-1365688800-1365688800@optimisation.doc.ic.ac.uk SUMMARY:Seminar: Practical Portfolio Optimization DESCRIPTION:Title: Practical Portfolio OptimizationSpeaker: Prof. Victor DeMiguelAffiliation: London Business SchoolLocation: CPSE seminar room (C615 Roderic Hill)Time: 2:00pm \nAbstract. The Nobel laureate Harry Markowitz showed that an investor who cares only about the mean and variance of portfolio returns should hold a portfolio on the efficient frontier. To implement these portfolios in practice\, one needs to estimate the means and covariances of asset returns. Traditionally\, the sample means and covariances have been used for this purpose. But due to estimation error\, the portfolios that rely on the sample estimates typically perform poorly out of sample. In this talk\, we will first illustrate the difficulties inherent in estimating mean-variance portfolios\, and then we will discuss several approaches that can be used to overcome these difficulties in practice. \nAbout the speaker. Victor DeMiguel is Professor of Management Science and Operations at London Business School. He holds a PhD in Management Science and Engineering from Stanford University\, and an MS in Industrial Engineering from Universidad Politecnica de Madrid. Victor's research focuses on the design\, analysis\, and application of optimization models for managerial decision making. Applications include financial portfolio selection and competition modelling. His papers have been published in journals such as Management Science\, Operations Research\, and Mathematics of Operations Research. One of his most popular papers is “Optimal Versus Naive Diversification: How Inefficient is the 1/N Portfolio Strategy”\, which received the Best Paper Award from the Institute for Quantitative Investment Research and was published in The Review of Financial Studies. Victor teaches the MBA courses Decision and Risk Analysis and Financial Modelling with Spreadsheets\, and the Strategic Decision Making module for Executive Education. He is the recipient of the Junior Faculty Teaching Award and the Outstanding Core Course Teaching Award at London Business School. More information about Victor DeMiguel can be found here: URL:https://optimisation.doc.ic.ac.uk/event/seminar-practical-portfolio-optimization/ END:VEVENT BEGIN:VEVENT DTSTART;TZID=UTC:20130306T150000 DTEND;TZID=UTC:20130306T150000 DTSTAMP:20260418T131427 CREATED:20170124T102144Z LAST-MODIFIED:20170124T102144Z UID:587-1362582000-1362582000@optimisation.doc.ic.ac.uk SUMMARY:Seminar: OR and Optimization under Uncertainty using R DESCRIPTION:Title: OR and Optimization under Uncertainty using RSpeaker: Dr. Ronald Hochreiter Affiliation: WU Vienna University of Economics and Business Location: Room 218 Huxley BuildingTime: 3:00pm \nAbstract. The statistical computing package R is well-known in the Statistics community as well as in the Data Science and Business Analytics community\, however Operations Researchers usually use MatLab and Python to solve their problems or are even using C++/Java directly to interfere with optimization solvers. In this talk\, we will outline the convenience of using R for dealing with OR tasks by solving Optimization under Uncertainty problems\, especially problems arising in the area of Stochastic Programming applied to Finance. \nAbout the speaker. Ronald Hochreiter is Assistant Professor at the WU Vienna University of Economics and Business and Visiting Professor at the University of Bergamo. His research interests includes Operations Research (Optimization under Uncertainty) as well as Data Science and Business Analytics. Furthermore he is consultant for various companies in Austria. URL:https://optimisation.doc.ic.ac.uk/event/seminar-or-and-optimization-under-uncertainty-using-r/ END:VEVENT BEGIN:VEVENT DTSTART;TZID=UTC:20130220T150000 DTEND;TZID=UTC:20130220T150000 DTSTAMP:20260418T131427 CREATED:20170124T102144Z LAST-MODIFIED:20170124T102144Z UID:588-1361372400-1361372400@optimisation.doc.ic.ac.uk SUMMARY:Seminar: Applications and Beyond DESCRIPTION:Title: Applications and BeyondSpeaker: Amit M. ManthanwarAffiliation: On Industrial Process Automation and Control TheoryLocation: CPSE seminar room (C615 Roderic Hill)Time: 3:00pm \nAbstract. At the end of second millennia advent of the technology of Model Predictive Control (MPC) has revolutionized the way process industry is controlled and operated. This technological breakthrough for industrial process automation asks the much sought-after control objective question\, how to attain desired profit margins while satisfying all the conditions of feasibility\, stability\, operability\, performance and safety from plants operating under the influence of uncertainty? Towards answering this question\, first generation of MPC development was focused on the various regulation types of issues. Next\, we focused on the development of MPC tuning engine that drives the dynamic economics of process control–delivering true economic value of control strategy. Now in the second generation of its evolution questions are raised on how MPC could be administered to the applications having fast dynamics (e.g.\, fuel cells\, automobile and biomedical applications) or systems to which on-line implementation of MPC is expensive or not possible. This has lead to the development of off-line (explicit) implementation of MPC via –you-solve-only-once– multi-parametric optimisation techniques. In this new role of MPC-on-a-chip\, we aim to develop robust control algorithms and tools for nonlinear systems operating under uncertainty. Using these tools and underlined computer-aided technologies of process automation we can minimise use of energy\, minimise exploitation of environment and maximise the economic profit incentives that are consistently demanded from modern industrial plants while operating at their optimal performance and safety conditions. In this seminar we will present historical developments and major milestones in modern control theory. Furthermore\, a theoretical perspective and computational challenges to account for uncertainty in process operation for various classes of nonlinear systems will be outlined. From the practical application standpoint\, we will present a generic control automation framework for fuel cell system and its integrated subsystems. To achieve this objective a fully integrated state-of-the-art process automation laboratory for fuel cell system is developed. This testing and validation facility will be used for characterising various designs as well as benchmarking control policies of the polymer electrolyte membrane fuel cell system. Finally\, we will present how the proposed control platform will be able to communicate with the integrated subsystems at the different length-time scales while improving operational performance of fuel cell system by guaranteeing efficiency\, material stability and longevity. As a concluding remark we will highlight our vision and goal of this research program so as to contribute significantly to the knowledge while fostering new technological discoveries in the area of process systems engineering by symbiotic\, synergetic and sustainable development of industrial process automation technology. \nAbout the speaker. Amit received his Master's Degree in Chemical Engineering from Illinois Institute of Technology\, Chicago. He worked as a senior distributed control systems engineer and automation consultant with RasGas in Qatar before joining Invensys as a Software Technology Developer for Advanced Process Performance Suit and Model Predictive Control product 'Connoisseur™'. He was Lecturer and Assistant Professor at the College of Engineering\, Pune. Currently he is working with Professor Efstratios N. Pistikopoulos at the Centre for Process Systems Engineering at Imperial College London. His research is focused on the development of economically and environmentally conscious process design\, global optimisation and robust control theory. His research work carried out in collaboration with Professor Donald J. Chmielewski has been incorporated in the book titled: 'Smart Process Plants: Software and Hardware Solutions for Accurate Data and Profitable Operations'. He has published 7 scientific papers with G-index 3 and H-index 2. He is a recipient of Hamid Arastoopour Excellence in Teaching Award at Illinois Institute of Technology\, Chicago and is listed in 2006 Marquis Who's who in Science and Engineering. He is a recipient of IChemE Journals' 2011 Best Reviewers Award. URL:https://optimisation.doc.ic.ac.uk/event/seminar-applications-and-beyond/ END:VEVENT BEGIN:VEVENT DTSTART;TZID=UTC:20130214T110000 DTEND;TZID=UTC:20130214T110000 DTSTAMP:20260418T131427 CREATED:20170124T102145Z LAST-MODIFIED:20170124T102145Z UID:589-1360839600-1360839600@optimisation.doc.ic.ac.uk SUMMARY:Seminar: Time Consistency in Multistage Stochastic Optimization DESCRIPTION:Title: Time Consistency in Multistage Stochastic OptimizationSpeaker: Prof. Georg PflugAffiliation: Department of Statistics and Operations Research – University of Vienna Location: CPSE seminar room (C615 Roderic Hill)Time: 11:00am \nAbstract. A fundamental principle for dynamic optimization is Bellman’s priciple\, stating that at any initially optimal decision sequence is also optimal at later stages. This time-consistency principle is still valid for stochastic programs if the criterion is to minimze expected costs or to maximize expeted profits. However\, as simple examples show\, it is not longer valid\, if risk-sensitive criteria are chosen such as the optimization of the average value at risk. It turns out that most risk functionals are time inconsistent\, e.g.\, it may happen that today some loss distribution appears to be less risky than another\, but looking at the conditional distribution at a later time\, the opposite relation holds. We demonstrate that this time inconsistency disappears if the conditional functionals are defined in an extended manner\, i.e.\, are evaluated under a specific change of measure. It follows from our results that\, for consistency reasons\, the revelation of partial information in time must dramatically change a decision maker's preferences among the remaining courses of action to keep time consistency. We extend conditional risk functionals to allow a temporal decomposition of the initial risk functional in time consistent way. With this extension\, a Bellmann priciple may be proved. Counterexamples show that without change of measures the only time consistent risk functionals are the expectation and the essential supremum. \nAbout the speaker. \nBorn on June 10th\, 1951 in Vienna. Study of Law\, Mathematics and Statistics at the University of Vienna\, Magister iuris (1974)\, Ph.D in Mathematics (1976). Assistant Professor University of Vienna (1976-81). Professor\, University of Giessen\, Germany (1981-1989). Full Professor\, University of Vienna (1989 – ). \n \nVisting Professor at the University of Bayreuth (1979)\, Michigan State University (1985)\, University of California Davis (1993)\, Universite de Rennes (1994)\, Technion Haifa\, Israel (1996)\, Princeton University (2001)\, University of California Davis (2006). \n \nDean\, Department of Mathematics\, University of Giessen\, Germany (1987-88); Dean\, Faculty of Business\, Economics and Statistics\, University of Vienna\, Austria (2008-2010); Head\, Department of Statistics and Decision Support Systems\, University of Vienna (2000-2003); Member of Senate\, University of Vienna (200-2003); Research scholar\, International Institute of Applied Systems Analysis\, IIASA (1990 – ) – Risk\, Modelling and Society (RMS)\, Risk and Vulnerability Project (RAV)\, . \n \nEditor in chief: Statistics and Decisions\, Central European Journal of Operations Research \n \nAssociate editor: Statistics and Probability Letters (1994-2007)\, Stochastic Programming Electronic Publication Series\, Austrian Journal of Statistics\, Mathematics of Operations Research (1994 – 1997)\, Mathematical Methods of OR\, Computational Optimization and Applications\, Computational Management Science\, Energy Systems: Optimization\, Modeling\, Simulation and Economic Aspects; Vestnik of the Finance Academy\, Moscow. \n \nMember\, Council of Scientists\, INTAS\, Bruessel (1999 – 2002); Fellow\, International Statistical Institute; Member\, executive board of the international committee on stochastic programming; Member\, central research council\, University of Bolzano/Bozen. \n \nAuthor Author of 4books\, editor of 5 books\, and more than 70 publications in refereed journals\, such as: Annals of Statistics\, Annals of OR\, Probability Theory\, J Statist. Planning and Inference\, J. ACM\, Parallel Computing\, The Computer Journal\, Math. Programming\, Mathematics of Optimization\, SIAM J. on Optimization\, Computational Optimization and Applications\, J. Applied Probability\, Stoch. Processes and Applications\, Graphs and Combinatorics\, J. Theoretical Computer Science\, Quantitative Finance etc. \n \nOrganizer of COMETT II Workshop "Simulation and Optimization"\, Raach (1992); Workshop"Computer Intensive Methods in Simulation and Optimization"\, IIASA (1994); EURO Winter School "Stochastic Optimization"\, Semmering (1996); Fourth World Congress of the Bernoulli Society\, Vienna (1996); Workshop Stochastic Dynamic Optimization\, IIASA (2002); Workshop series "Mathematical Optimization for Financial Models"\, Semmering (2003)\, Cyprus (2003)\, Bergamo (2004); 11th International Conference on Stochastic Programming\, Vienna (2007); APMOD\, Vienna and Bratislava (2008). \n \nProject leader of past and present projects: Project leader of past and present projects: Statistical pattern recognition (Austrian National Bank); Pension fund management (BVP pension fund); Data dependency in financial optimization (FWF- Austrian Science Fund); Optimal network design and marketing strategies (Telekom Austria); AURORA-Advanced parallel and distributed algorithms for Computational Finance (FWF); Unit life insurance with guarantee (Austrian National Bank); WEBOPT (European Commission-subproject leader)\, Risk management in liberalized electricity markets (WWTF); Seeds in Finance (Austrian National Bank)\, RISKPLAN (Asia-Link)\, Long-term risk management (FWF)\, Pension fund management (Bridge Program). URL:https://optimisation.doc.ic.ac.uk/event/seminar-time-consistency-in-multistage-stochastic-optimization/ END:VEVENT BEGIN:VEVENT DTSTART;TZID=UTC:20121211T160000 DTEND;TZID=UTC:20121211T160000 DTSTAMP:20260418T131427 CREATED:20170124T102145Z LAST-MODIFIED:20170124T102145Z UID:590-1355241600-1355241600@optimisation.doc.ic.ac.uk SUMMARY:Seminar: Optimal control of Weakly Connected Markov Decision Processes DESCRIPTION:Title: Optimal control of Weakly Connected Markov Decision ProcessesSpeaker: Dr. Panos ParpasAffiliation: Department of Computing – Imperial College LondonLocation: Room 140 HuxleyTime: 4:00pm \nAbstract. Weakly connected Markov processes are used to model stochastic dynamics across different scales. They are widely used in electrical engineering\, finance and molecular dynamics simulations. In many of these applications it is becoming increasingly important to both efficiently simulate these processes as well as to control them. Classical algorithms for the control of Markov Processes cannot be directly applied to weakly connected Markov processes. The problem with existing approaches is that they exhibit an extremely slow convergence rate due to the existence of multiscale effects. We show why existing algorithms are slow\, and propose a new class of algorithms with more favourable convergence properties and computational complexity. In our approach we use spectral graph theory to derive a hierarchy of models that are valid at different resolutions. We then propose a polynomial time algorithm that uses the finest resolution model only when required. The rate of convergence of the algorithm is discussed as well as its complexity. \nAbout the speaker. Panos Parpas is a Lecturer in the Quantitative Analysis and Decision Science (QUADS) section of the Department of Computing at Imperial College London. Before joining Imperial College he was a postdoctoral fellow at the MIT Energy Initiative (2009-2011). Before that he was a quantitative associate at Credit-Suisse (2007-2009). He completed his PhD in computational optimization at Imperial College in 2006. Panos Parpas is interested in the development and analysis of quantitative optimization models under uncertainty. Stochastic optimization models are used in many areas such as economics\, finance\, engineering\, and energy systems. Realistic models have a large number of variables\, and multiple interactions across time and space. Advanced computational methods\, and analytical approximations that take advantage of problem structure are needed in order to analyze realistic models. I am interested in both the development of computational methods and applications. URL:https://optimisation.doc.ic.ac.uk/event/seminar-optimal-control-of-weakly-connected-markov-decision-processes/ END:VEVENT BEGIN:VEVENT DTSTART;TZID=UTC:20121129T150000 DTEND;TZID=UTC:20121129T150000 DTSTAMP:20260418T131427 CREATED:20170124T102145Z LAST-MODIFIED:20170124T102145Z UID:591-1354201200-1354201200@optimisation.doc.ic.ac.uk SUMMARY:Seminar: The Branch-and-Sandwich Algorithm for Mixed-Integer Nonlinear Bilevel Programming Problems DESCRIPTION:Title: The Branch-and-Sandwich Algorithm for Mixed-Integer Nonlinear Bilevel Programming ProblemsSpeaker: Dr. Polyxeni-Margarita KleniatiAffiliation: Centre for Process Systems Engineering at Imperial College LondonLocation: Room 144 HuxleyTime: 3:00pm \nAbstract. We extend our recently introduced algorithm for general bilevel programming problems\, Branch-and-Sandwich (Kleniati and Adjiman\, J. Global Optim.\, 2012)\, to the class of mixed-integer nonlinear bilevel problems. As in the original algorithm\, auxiliary inner lower and upper bounding problems are constructed in order to bound the inner value function and provide constant bound cuts for the outer upper and outer lower bounding problems. The KKT-based relaxations\, originally proposed for the inner upper bounding and the outer lower bounding problems\, are applicable with respect to the lower-level continuous variables based on appropriate constraint qualifications\, but are no longer required. In the extension that we present here\, a robust counterpart approach is employed to formulate the inner upper bounding problem and the resulting bound cut may be the only constraint added to the proposed outer lower bounding problem. The branching framework with auxiliary lists of nodes\, as developed for the original Branch-and-Sandwich\, is also applied to the discrete case. The algorithm is used to solve successfully ten literature problems. \nAbout the speaker. Dr. Kleniati is undertaking her second postdoctoral research position with Prof. Adjiman at the Chemical Engineering department of Imperial College London. She received her PhD in Computing and Optimisation research in 2010 under the supervision of Prof. Rustem at the department of Computing in Imperial College London. The research of Polyxeni Kleniati is currently focused on the global optimisation of bilevel programming problems with applications to chemical engineering. URL:https://optimisation.doc.ic.ac.uk/event/seminar-the-branch-and-sandwich-algorithm-for-mixed-integer-nonlinear-bilevel-programming-problems/ END:VEVENT BEGIN:VEVENT DTSTART;TZID=UTC:20121127T140000 DTEND;TZID=UTC:20121127T140000 DTSTAMP:20260418T131427 CREATED:20170124T102145Z LAST-MODIFIED:20170124T102145Z UID:592-1354024800-1354024800@optimisation.doc.ic.ac.uk SUMMARY:Seminar: Matrix Learning Problems and First-Order Optimization DESCRIPTION:Title: Matrix Learning Problems and First-Order OptimizationSpeaker: Dr. Andreas Argyriou Affiliation: Toyota Technological Institute at ChicagoLocation: Room 217 Huxley BuildingTime: 2:00pm \nAbstract. In the past few years\, there has been significant interest in nonsmooth convex optimization problems involving matrices\, especially in the areas of machine learning\, statistics and control. Instances of such problems are multitask learning and matrix completion\, robust PCA\, sparse inverse covariance selection etc. I will present PRISMA\, a new optimization algorithm for minimizing a convex objective which decomposes into three parts: a smooth part\, a simple non-smooth Lipschitz part\, and a simple nonsmooth non-Lipschitz part. Our algorithm combines the methodologies of smoothing and accelerated proximal methods. Moreover\, our convergence result removes the assumption of bounded domain required by Nesterov's smoothing methods. I will show how PRISMA can be applied to the problems of max-norm regularized matrix completion and clustering\, robust PCA and sparse inverse covariance selection\, and compare to state of the art methods. \nAbout the speaker. Andreas Argyriou has received degrees in Computer Science from MIT and a PhD in Computer Science from UCL. The topic of his PhD work has been on machine learning methodologies integrating different tasks and data sources. He has held postdoctoral and research faculty positions at UCL\, TTI Chicago\, KU Leuven and is currently in Ecole Centrale Paris with an RBUCE-UP fellowship. His current interests are in the areas of machine learning with big and complex data\, compressed sensing and convex optimization methods. URL:https://optimisation.doc.ic.ac.uk/event/seminar-matrix-learning-problems-and-first-order-optimization/ END:VEVENT BEGIN:VEVENT DTSTART;TZID=UTC:20121121T140000 DTEND;TZID=UTC:20121121T140000 DTSTAMP:20260418T131427 CREATED:20170124T102145Z LAST-MODIFIED:20170124T102145Z UID:593-1353506400-1353506400@optimisation.doc.ic.ac.uk SUMMARY:Seminar: Algorithms and computer architectures for efficient real-time optimization and linear algebra solvers DESCRIPTION:Title: Algorithms and computer architectures for efficient real-time optimization and linear algebra solversSpeaker: Eric C KerriganAffiliation: Department of Aeronautics and Department of Electrical and Electronic Engineering – Imperial College LondonLocation: Room 408 Electrical & Electronic Engineering DepartmentTime: 2:00pm \nAbstract. In many engineering applications where one would like to implement control and signal processing algorithms\, one needs to use the latest measurements to update and solve a sequence of numerical optimization or linear algebra problems. Solving these problems in a computationally efficient and numerically reliable fashion on an embedded computing system is a challenging task. One of the key choices that an engineer has to make in order to determine the speed\, cost and power consumption of a microprocessor is the number representation that will be used in the arithmetic units. CPUs within modern desktop or laptop PCs provide hardware support for double precision floating-point arithmetic. However\, most microprocessors in embedded systems do not support double precision floating-point arithmetic; they often only support single-precision floating-point and/or fixed-point arithmetic. It is therefore possible that\, because of a significant decrease in precision or dynamic range\, a numerical algorithm that gives reliable results on the office PC or laptop might give completely different results when implemented on an embedded computing system. We will present novel mathematical formulations\, computer architectures\, optimization solvers and linear algebra solvers to show that computational resources can be reduced significantly using very low precision arithmetic\, without sacrificing accuracy. We will also present new mathematical results that allow one to use fixed-point arithmetic to make impressive computational savings in iterative linear algebra solvers. Our theoretical results will be supported by implementations on a Field Programmable Gate Array (FPGA) and we will show that it is possible to exceed the peak theoretical performance of a 1TFLOP/s general-purpose GPU. \nAbout the speaker. Dr Kerrigan’s research includes the optimal and robust control of nonlinear\, constrained and distributed parameter systems. His research is focused on the development of efficient numerical methods and computational hardware architectures for solving the resulting problems and is applied to a variety of problems in aerospace and fluid dynamics. URL:https://optimisation.doc.ic.ac.uk/event/seminar-algorithms-and-computer-architectures-for-efficient-real-time-optimization-and-linear-algebra-solvers/ END:VEVENT BEGIN:VEVENT DTSTART;TZID=UTC:20121108T150000 DTEND;TZID=UTC:20121108T150000 DTSTAMP:20260418T131427 CREATED:20170124T102146Z LAST-MODIFIED:20170124T102146Z UID:594-1352386800-1352386800@optimisation.doc.ic.ac.uk SUMMARY:Seminar: Reflections on robustness in stochastic programs with risk constraints DESCRIPTION:Title: Reflections on robustness in stochastic programs with risk constraintsSpeaker: Dr. Milos KopaAffiliation: Department of Probability and Mathematical Statistics – Charles University in PragueLocation: Room 301 William PenneyTime: 3:00pm \nAbstract. This paper is a contribution to the robustness analysis for stochastic programs whose set of feasible solutions depends on the probability distribution P. For various reasons\, probability distribution P may not be precisely specified and we study robustness of results with respect to perturbations of P. The main tool is the contamination technique. For the optimal value\, local contamination bounds are derived and applied to robustness analysis of the optimal value of a portfolio performance under risk-shaping constraints. To illustrate the theoretical results\, numerical examples for several mean-risk models are presented. Finally\, under suitable conditions on the structure of the problem and for discrete distributions we shall suggest a new robust portfolio efficiency test with respect to the first (second) order stochastic dominance criterion and we shall exploit the contamination technique to analyze the resistance with respect to additional scenarios. \nAbout the speaker. Dr. Milos Kopa is an assistant professor at Charles University in Prague. He received his Ph.D. degree in Econometrics and Operations research in 2006 (supervisor: Prof. Jitka Dupacova\, Charles University in Prague). He is a member of several scientific societies: Stochastic Programming Community\, EURO working group on financial modelling\, EUROPT. Recently he has become an elected member (and secretary) of Managerial Board of new EURO working group on stochastic programming. He is vice-head of the Center of Excellence “Dynamic models in economics” that comprises about 40 leading researchers (in quantitative economics and finance) from Czech Republic. The research of Milos Kopa is focused on: stochastic programming theory and applications\, especially financial applications. In recent years he has published several papers dealing with portfolio efficiency with respect to stochastic dominance criteria; data envelopment analysis and its relation to stochastic dominance; robustness (contamination) in stochastic programs with risk constraints\, etc. URL:https://optimisation.doc.ic.ac.uk/event/seminar-reflections-on-robustness-in-stochastic-programs-with-risk-constraints/ END:VEVENT BEGIN:VEVENT DTSTART;TZID=UTC:20121030T140000 DTEND;TZID=UTC:20121030T140000 DTSTAMP:20260418T131427 CREATED:20170124T102146Z LAST-MODIFIED:20170124T102146Z UID:595-1351605600-1351605600@optimisation.doc.ic.ac.uk SUMMARY:Seminar: Multi-task Learning and Matrix Regularization DESCRIPTION:Title: Multi-task Learning and Matrix RegularizationSpeaker: Prof. Massimiliano PontilAffiliation: Department of Computer Science – University College LondonLocation: Room 301 William PenneyTime: 2:00pm \nAbstract. We discuss the problem of estimating a structured matrix with a large number of elements. A key motivation for this problem occurs in multi-task learning. In this case\, the columns of the matrix correspond to the parameters of different regression or classification tasks\, and there is structure due to relations between the tasks. We present a general method to learn the tasks’ parameters as well as their structure. Our approach is based on solving a convex optimization problem\, involving a data term and a penalty term. We highlight different types of penalty terms which are of practical and theoretical importance. They implement structural relations between the tasks and achieve a sparse representations of parameters. We address computational issues as well as the predictive performance of the method. Finally\, we describe some recent applications of these methods to computer vision and human computer interaction. \nAbout the speaker. Massimiliano Pontil is Professor of Computational Statistics and Machine Learning in the Department of Computer Science at University College London. His research interests are in the field of machine learning with a focus on regularization methods\, convex optimization and statistical estimation. He has published about 100 research papers on these topics\, is regularly in the programme committee of the leading conferences in the field\, is an associate editor of the Machine Learning Journal and is a member of the scientific advisory board of the Max Planck Institute for Biological Cybernetics\, Germany. URL:https://optimisation.doc.ic.ac.uk/event/seminar-multi-task-learning-and-matrix-regularization/ END:VEVENT BEGIN:VEVENT DTSTART;TZID=UTC:20120913T150000 DTEND;TZID=UTC:20120913T150000 DTSTAMP:20260418T131427 CREATED:20170124T102146Z LAST-MODIFIED:20170124T102146Z UID:596-1347548400-1347548400@optimisation.doc.ic.ac.uk SUMMARY:Seminar: Performance bounds in robust & stochastic optimal control problems DESCRIPTION:Title: Performance bounds in robust & stochastic optimal control problemsSpeaker: Bart Van ParysAffiliation: Automatic Control Laboratory at Swiss Federal Institute of TechnologyLocation: Room 308 Computing DepartmentTime: 3:00pm \nAbstract. We present a new method to bound the performance of controllers for uncertain linear systems with mixed state and input constraints and bounded persistent disturbances. We take as a performance metric either an expected-value or minimax discounted cost over an infinite horizon\, and provide a method for computing a lower bound on the achievable performance of any causal control policy in either case. Our lower bound is compared to an upper performance bound provided by restricting the choice of controller to one that is affine in the observed disturbances\, and we show that the two bounds are closely related. In particular\, the lower bounds have a natural interpretation in terms of affine control policies that are optimal for a problem with a restricted disturbance set. We show that our performance bounds can be computed via solution of a finite dimensional convex optimization problem\, and provide numerical examples to illustrate the efficacy of our method. \nAbout the speaker. Bart holds a BA degree in electrical engineering\, and a MA degree in applied/engineering mathematics\, both from the University of Leuven. Since September 2011 he has been a PhD student in the Swiss Federal Institute of Technology (ETH Zürich) under the supervision of Prof. Manfred Morari and Dr. Paul Goulart. URL:https://optimisation.doc.ic.ac.uk/event/seminar-performance-bounds-in-robust-stochastic-optimal-control-problems/ END:VEVENT BEGIN:VEVENT DTSTART;TZID=UTC:20120830T140000 DTEND;TZID=UTC:20120830T140000 DTSTAMP:20260418T131427 CREATED:20170124T102146Z LAST-MODIFIED:20170124T102146Z UID:597-1346335200-1346335200@optimisation.doc.ic.ac.uk SUMMARY:Seminar: Optimisation with PDE constraints using automated consistent adjoints of finite element models DESCRIPTION:Title: Optimisation with PDE constraints using automated consistent adjoints of finite element modelsSpeaker: Simon FunkeAffiliation: Applied Modelling and Computation Group at Imperial CollegeLocation: CPSE seminar room (C615 Roderic Hill)Time: 2:00pm \nAbstract. Optimisation with partial differential equations (PDE) as constraints arise in many research areas from science engineering to finance. Typical examples are data assimilation for weather forecasting or ocean modelling and shape optimisation for wing designs in which PDEs enforce the laws of physics. The resulting optimisation problems are constrained by these nonlinear\, time dependent differential equations which can be computationally extremely demanding to solve. Therefore the usage of gradient based optimisation algorithms is usually essential to reduce the number of optimisation iterations. In this talk we present work towards a new framework for solving PDE constrained optimisation problems that aims to automate many of the steps involved in solving these kind of problems. Given a differentiable PDE model\, the parameter set and a functional of interest\, it applies gradient based optimisation to solve the optimisation problem. The key feature of this framework is the efficient gradient computation by automatically deriving and solving the associated the adjoint equation. The framework is demonstrated on examples for steady and unsteady optimal control problems. One of the major difficulties in practice is the derivation and implementation of the adjoint system to efficiently compute gradient information; Naumann (2011) described it as “[…] one of the great open challenges in the field of High-Performance Scientific Computing” for large scale simulation code. There are two current approaches to derive the adjoint equation each of which suffer from their own limitations. Algorithmic differentiation (AD) derives the adjoint model directly from the model source code. In practice this technique has strong restrictions\, and requires a major initial and ongoing investment to prepare the code for automatic adjoint generation. An alternative is to formulate the adjoint PDE and to discretise this separately. This approach\, known as the continuous adjoint has the disadvantage that two different model code bases must be maintained and manually kept synchronised as the model develops. In this talk\, we present an alternative approach where the PDE is formulated in a high level language that resembles the matematical notation. The model is automatically generated using code generation (using the FEniCS project). In this approach it is the high level code specification which is differentiated\, a task very similar to the formulation of the continuous adjoint. However since the forward and adjoint models are generated automatically\, the difficulty of maintaining them vanishes and the software engineering process is therefore robust. The scheduling and execution of the adjoint model\, including the application of an appropriate checkpointing strategy is managed by a library called libadjoint. In contrast to the conventional algorithmic differentiation description of considering a model as a series of primitive mathematical operations\, libadjoint employs a new abstraction of considering the model as a sequence of discrete equations which are assembled and solved. It is the coupling of the respective abstractions employed by libadjoint and the FEniCS project which produces the adjoint model automatically\, without further intervention from the model developer. \nAbout the speaker. Simon is a PhD student in the Applied Modelling and Computation Group at Imperial College. URL:https://optimisation.doc.ic.ac.uk/event/seminar-optimisation-with-pde-constraints-using-automated-consistent-adjoints-of-finite-element-models/ END:VEVENT BEGIN:VEVENT DTSTART;TZID=UTC:20120705T163000 DTEND;TZID=UTC:20120705T163000 DTSTAMP:20260418T131427 CREATED:20170124T102146Z LAST-MODIFIED:20170124T102146Z UID:598-1341505800-1341505800@optimisation.doc.ic.ac.uk SUMMARY:Seminar: Robust Pricing of Monopolistic Cloud Computing Services with Service Level Agreements DESCRIPTION:Title: Robust Pricing of Monopolistic Cloud Computing Services with Service Level AgreementsSpeaker: Vladimir RoitchAffiliation: Department of Computing – Imperial College LondonLocation: CPSE seminar room (C615 Roderic Hill)Time: 4:30pm \nAbstract. Cloud Computing is a new computing paradigm that gives end-users on-demand access to computing resources of companies that maintain large data centres. Here\, we address the optimal pricing of cloud computing services from the perspective of a monopolistic service provider that needs to manage demand responsiveness and uncertainty. We formulate the pricing problem for on-demand services as a multi-stage stochastic program and model service level agreements via chance constraints. Under weak assumptions about the demand uncertainty we show that the resulting model can be reduced to an equivalent two-stage stochastic program. As cloud computing is only just emerging\, it is impossible to reliably estimate demand distributions from historical data. Indeed\, such data may even be difficult to collect. We address this type of model uncertainty by adopting a distributionally robust approach\, assuming that only information about the location\, size and support (but not the shape) of the demand distribution is available. We show that the arising robust model can be reformulated as a second-order cone program\, and we analytically derive the worst-case distributions. Several extensions of the basic model are discussed. First\, we study generalized models in which higher-order moments of the demand distribution are known. Next\, we include multiple products and account for different product qualities. Finally\, we investigate the possibility of selling unused capacity (if any) on a spot market. \nAbout the speaker. URL:https://optimisation.doc.ic.ac.uk/event/seminar-robust-pricing-of-monopolistic-cloud-computing-services-with-service-level-agreements/ END:VEVENT BEGIN:VEVENT DTSTART;TZID=UTC:20120705T160000 DTEND;TZID=UTC:20120705T160000 DTSTAMP:20260418T131427 CREATED:20170124T102147Z LAST-MODIFIED:20170124T102147Z UID:599-1341504000-1341504000@optimisation.doc.ic.ac.uk SUMMARY:Seminar: A decision rule approach to medium-term hydropower scheduling under uncertainty DESCRIPTION:Title: A decision rule approach to medium-term hydropower scheduling under uncertaintySpeaker: Paula RochaAffiliation: Department of Computing – Imperial College LondonLocation: CPSE seminar room (C615 Roderic Hill)Time: 4:00pm \nAbstract. We present a multistage stochastic optimisation model for the medium-term scheduling of a cascaded hydropower system. Electricity spot prices change on a much shorter time scale than the hydrological dynamics of the reservoirs in the cascade. We exploit this property to reduce computational complexity: we partition the planning horizon into hydrological macroperiods\, and we account for intra-stage price variability by using price duration curves. Moreover\, we restrict the space of recourse decisions to those affine in the observable data\, thereby obtaining a tractable approximate problem. \nAbout the speaker. URL:https://optimisation.doc.ic.ac.uk/event/seminar-a-decision-rule-approach-to-medium-term-hydropower-scheduling-under-uncertainty/ END:VEVENT BEGIN:VEVENT DTSTART;TZID=UTC:20120703T140000 DTEND;TZID=UTC:20120703T140000 DTSTAMP:20260418T131427 CREATED:20170124T102147Z LAST-MODIFIED:20170124T102147Z UID:600-1341324000-1341324000@optimisation.doc.ic.ac.uk SUMMARY:Seminar: Protection at All Levels: Probabilistic Envelope Constraints DESCRIPTION:Title: Protection at All Levels: Probabilistic Envelope ConstraintsSpeaker: Dr. Xu HuanAffiliation: Department of Mechanical Engineering at National University of SingaporeLocation: CPSE Seminar room (C616 Roderic Hill)Time: 2:00pm \nAbstract. Optimization under chance constraints is a standard approach to ensure that bad events such as portfolio losses\, are unlikely to happen. They do nothing\, however\, to protect more against terrible events (e.g.\, deep portfolio losses\, or bankruptcy). In this talk\, we will propose a new decision concept\, termed "probabilistic envelop constraint"\, which extends the notion of chance constraints\, to a formulation that provides different probabilistic guarantees at each level of constraint violation. Thus\, we develop a notion of guarantee across the spectrum of disasters\, or rare events\, ensuring these levels of protection hold across the curve\, literally. We further show that the corresponding optimization problem can be reformulated as a semi-infinite optimization problem\, and provide conditions that guarantee its tractability. Interestingly\, the resulting formulation is what is known as a comprehensive robust optimization in literature. This work thus provides a new fundamental link between two main methodologies in optimization under uncertainty: stochastic optimization and robust optimization. This is a joint work with Constantine Caramanis (UT-Austin) and Shie Mannor (Technion). \nAbout the speaker. Huan Xu has been an assistant professor of Mechanical Engineering of National University of Singapore since 2011. He obtained his Ph. D. degree in ECE from McGill University\, Canada\, in 2009\, and was a postdoctral research fellow of the University of Texas at Austin prior to joining NUS. His current research interest focuses on learning and decision-making in large-scale complex systems\, including machine learning\, high-dimensional statistics\, robust and adaptable optimization\, robust sequential decision making\, and applications to large-scale systems. He has published in leading operations research and machine learning journals including Operations Research\, Math. Oper. Res.\, IEEE Info. Theory\, JMLR\, and conferences including ICML\, NIPS\, and COLT. URL:https://optimisation.doc.ic.ac.uk/event/seminar-protection-at-all-levels-probabilistic-envelope-constraints/ END:VEVENT BEGIN:VEVENT DTSTART;TZID=UTC:20120702T140000 DTEND;TZID=UTC:20120702T140000 DTSTAMP:20260418T131427 CREATED:20170124T102147Z LAST-MODIFIED:20170124T102147Z UID:601-1341237600-1341237600@optimisation.doc.ic.ac.uk SUMMARY:Seminar: Time-Critical Cooperative Path-Following Control of Multiple UAVs DESCRIPTION:Title: Time-Critical Cooperative Path-Following Control of Multiple UAVsSpeaker: Prof. Naira HovakimyanAffiliation: Department of Mechanical Science and Engineering at University of Illinois at Urbana-ChampaignLocation: Room 212 William PennyTime: 2:00pm \nAbstract. Worldwide\, there has been growing interest in the use of autonomous vehicles to execute cooperative missions of increasing complexity without constant supervision of human operators. Despite significant progress in the field of cooperative control\, several challenges need to be addressed to develop strategies capable of yielding robust performance of a fleet in the presence of complex vehicle dynamics\, communications constraints\, and partial vehicle failures. In this talk\, we will present a theoretical framework for the development of decentralized strategies for cooperative motion control of multiple vehicles that must meet stringent spatial and temporal constraints. The approach adopted applies to teams of heterogeneous systems\, and does not necessarily lead to swarming behavior. Flight test results of a coordinated road search mission involving multiple small tactical UAVs will be discussed to demonstrate the efficacy of the multi-vehicle cooperative control framework presented. \nAbout the speaker. Naira Hovakimyan received her M.S. in Theoretical Mechanics and Applied Mathematics in 1988 from Yerevan State University in Armenia. She received her Ph.D. in Physics and Mathematics in 1992\, in Moscow\, from the Institute of Applied Mathematics of Russian Academy of Sciences\, majoring in optimal control and differential games. In 1997 she was awarded a governmental postdoctoral scholarship to work in INRIA\, France. In 1998 she was invited to the School of Aerospace Engineering of Georgia Tech\, where she worked as a research faculty member until 2003. In 2003 she joined the Department of Aerospace and Ocean Engineering of Virginia Tech\, and in 2008 she moved to the University of Illinois at Urbana-Champaign\, where she is a professor and Schaller faculty scholar. She is the 2011 recipient of the AIAA Mechanics and Control of Flight Award. She has coauthored one book and more than 250 refereed publications. Her research interests are in the theory of robust adaptive control and estimation with an emphasis on aerospace applications\, control in the presence of limited information\, networks of autonomous systems and game theory. She is an associate fellow and life member of AIAA\, a Senior Member of IEEE\, and a member of AMS and ISDG. URL:https://optimisation.doc.ic.ac.uk/event/seminar-time-critical-cooperative-path-following-control-of-multiple-uavs/ END:VEVENT BEGIN:VEVENT DTSTART;TZID=UTC:20120621T140000 DTEND;TZID=UTC:20120621T140000 DTSTAMP:20260418T131427 CREATED:20170124T102147Z LAST-MODIFIED:20170124T102147Z UID:602-1340287200-1340287200@optimisation.doc.ic.ac.uk SUMMARY:Seminar: Modeling and Solution Methods for Stochastic Programming Problems Under Endogenous Observation of Uncertainty DESCRIPTION:Title: Modeling and Solution Methods for Stochastic Programming Problems Under Endogenous Observation of UncertaintySpeaker: Dr. Christos MaraveliasAffiliation: Department of Chemical and Biological Engineering at the University of Wisconsin-Location: Room 219A William PennyTime: 2:00pm \nAbstract. We first present an overview of optimization problems under uncertainty and multi-stage stochastic programming methods. We then discuss applications where the decision maker alters the underlying stochastic process by affecting the timing of uncertainty observation\, i.e. problems under endogenous observation of uncertainty. We discuss how this important but less studied class of problems can be formulated as a large-scale multi-stage stochastic programming model. To address this challenging problem\, we develop a number of theoretical results\, modeling methods and computational techniques. First\, we show how the structure of the problem can be exploited to formulate substantially smaller yet tighter models. Second\, we discuss a number of approximations (e.g. finite-horizon approximation for rolling-horizon approaches) that can be used to obtain solutions of high quality. Third\, we present a novel branch-and-cut algorithm where we start from a reduced model and add essential constraints only if they are violated. The presented methods are applied to the planning of clinical trials in the pharmaceutical research and development pipeline. \nAbout the speaker. Christos obtained his Diploma in Chemical Engineering from the National Technical University of Athens. He then moved to the London School of Economics (London\, UK)\, where he received an MSc in Operational Research. After completing his military service in Greece\, he went to Carnegie Mellon University where he worked towards his PhD under the supervision of Professor Grossmann. In the fall of 2004 he joined the faculty of the Department of Chemical and Biological Engineering at the University of Wisconsin - Madison. He is a recipient of the Inaugural Olaf A. Hougen Fellowship\, an NSF CAREER award\, and the 2008 W. David Smith Jr. Award from the CAST division of ACIChE. Christos' primary research interests are in the areas of a) chemical production planning and scheduling\, b) process synthesis\, with an emphasis on biofuels\, c) optimization under uncertainty\, and d) the development of computational tools for the design of chemical and biological catalysts. URL:https://optimisation.doc.ic.ac.uk/event/seminar-modeling-and-solution-methods-for-stochastic-programming-problems-under-endogenous-observation-of-uncertainty/ END:VEVENT BEGIN:VEVENT DTSTART;TZID=UTC:20120525T140000 DTEND;TZID=UTC:20120525T140000 DTSTAMP:20260418T131427 CREATED:20170124T102148Z LAST-MODIFIED:20170124T102148Z UID:603-1337954400-1337954400@optimisation.doc.ic.ac.uk SUMMARY:Seminar: On Control and Random Dynamical Systems in Reproducing Kernel Hilbert Spaces DESCRIPTION:Title: On Control and Random Dynamical Systems in Reproducing Kernel Hilbert SpacesSpeaker: Dr. Boumediene HamziAffiliation: Department of Mathematics of Imperial CollegeLocation: CPSE seminar room (C615 Roderic Hill)Time: 2:00pm \nAbstract. We introduce a data-based approach to estimating key quantities which arise in the study of nonlinear control systems and random nonlinear dynamical systems. Our approach hinges on the observation that much of the existing linear theory may be readily extended to nonlinear systems – with a reasonable expectation of success – once the nonlinear system has been mapped into a high or infinite dimensional Reproducing Kernel Hilbert Space. In particular\, we develop computable\, non-parametric estimators approximating controllability and observability energy functions for nonlinear systems\, and study the ellipsoids they induce. It is then shown that the controllability energy estimator provides a key means for approximating the invariant measure of an ergodic\, stochastically forced nonlinear system. We also apply this approach to the problem of model reduction of nonlinear control systems. In all cases the relevant quantities are estimated from simulated or observed data. This is joint work with J. Bouvrie (Duke University). \nAbout the speaker. Boumediene Hamzi is a Marie Curie Fellow at the Department of Mathematics of Imperial College. He obtained his Ph.D. in Control Theory from the University of Paris-Sud and then held different visiting academic positions at UCDavis\, the MSRI and then Duke University. His current research interests are the analysis of control and random dynamical systems in Reproducing Kernel Hilbert Spaces in view of developing data-based methods for the analysis and prediction of random dynamical systems and control strategies for nonlinear systems on the basis of observed data (rather than a pre-described model). URL:https://optimisation.doc.ic.ac.uk/event/seminar-on-control-and-random-dynamical-systems-in-reproducing-kernel-hilbert-spaces/ END:VEVENT END:VCALENDAR