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:20140101T000000
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTART;TZID=UTC:20160701T150000
DTEND;TZID=UTC:20160701T150000
DTSTAMP:20260418T131354
CREATED:20170124T101756Z
LAST-MODIFIED:20170124T101756Z
UID:531-1467385200-1467385200@optimisation.doc.ic.ac.uk
SUMMARY:Seminar: Scheduling Algorithms for Energy Efficiency in Computing Systems
DESCRIPTION:Title: Scheduling Algorithms for Energy Efficiency in Computing SystemsSpeaker: Dimitrios LetsiosAffiliation: Department of Computing – Imperial College LondonLocation: Huxley BuildingTime: 3:00pm \nAbstract. Energy consumption of computing devices has become an important issue nowadays. A major tool for efficient energy management in the system level is dynamic speed (frequency) scaling combined with job scheduling. In this context\, the processing time of a job is not fixed\, but it depends on the speed at which it is processed while the energy is a convex function of the speed. The main goal is the design efficient algorithms which compute good trade-off solutions with respect to performance and energy. Our focus will be algorithmic techniques with provably good performances for fundamental problems of the area. \nAbout the speaker. Dimitrios Letsios has very recently joined Imperial College as a postdoctoral researcher. Previously\, he has served as a postdoctoral researcher with teaching duties at the University of Nice – Sophia Antipolis (2015-2016)\, at the Technical University of Munich (2014-2015) and at University Pierre and Marie Curie (2013-2014). Before\, he obtained his PhD degree at the University of Evry in Paris (2010-2013) and his MSc and BSc degrees at Athens University of Economics and Business (2004-2010). His research interests lie in theoretical computer science and\, more specifically\, the design of algorithms with proven performance guarantees (approximation algorithms). He has mainly worked on scheduling problems taking into account the energy consumption and communication costs of computing systems.
URL:https://optimisation.doc.ic.ac.uk/event/seminar-scheduling-algorithms-for-energy-efficiency-in-computing-systems/
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=UTC:20160531T170000
DTEND;TZID=UTC:20160531T170000
DTSTAMP:20260418T131354
CREATED:20170124T102131Z
LAST-MODIFIED:20170124T102131Z
UID:534-1464714000-1464714000@optimisation.doc.ic.ac.uk
SUMMARY:Seminar: Symmetry Groups and Topological Structure of Optimisation Problems
DESCRIPTION:Title: Symmetry Groups and Topological Structure of Optimisation ProblemsSpeaker: Georgia KouyialisAffiliation: Department of Computing – Imperial College LondonLocation: Huxley BuildingTime: 5:00pm \nAbstract.  \nAbout the speaker. Georgia Kouyialis is a PhD student in the Department of Computing (QUADS group)\, at Imperial College\, under the supervision of Dr. Ruth Misener. She obtained the MSci (Hons) degree in Mathematics from University College London (UCL). She received the EPSRC DTA funding and her research evolves around Mixed Integer Nonlinear Programming.
URL:https://optimisation.doc.ic.ac.uk/event/seminar-symmetry-groups-and-topological-structure-of-optimisation-problems/
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=UTC:20160531T160000
DTEND;TZID=UTC:20160531T160000
DTSTAMP:20260418T131354
CREATED:20170124T102131Z
LAST-MODIFIED:20170124T102131Z
UID:535-1464710400-1464710400@optimisation.doc.ic.ac.uk
SUMMARY:Seminar: Asymptotic Error Bounds for Control Constrained Singularly Perturbed Linear Quadratic Optimal Control Problems
DESCRIPTION:Title: Asymptotic Error Bounds for Control Constrained Singularly Perturbed Linear Quadratic Optimal Control ProblemsSpeaker: Sei HoweAffiliation: Department of Computing – Imperial College LondonLocation: Huxley BuildingTime: 4:00pm \nAbstract.  \nAbout the speaker. Sei Howe is a PhD student in the QUADS group at Imperial College. She received her B.A in pure mathematics from Reed College\, USA in 2011 and her M.Sc. in pure mathematics from Imperial College in 2012. Her supervisor is Dr. Panos Parpas and her research focuses on stochastic optimization of multi-scale processes.
URL:https://optimisation.doc.ic.ac.uk/event/seminar-asymptotic-error-bounds-for-control-constrained-singularly-perturbed-linear-quadratic-optimal-control-problems/
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=UTC:20160517T170000
DTEND;TZID=UTC:20160517T170000
DTSTAMP:20260418T131354
CREATED:20170124T102131Z
LAST-MODIFIED:20170124T102131Z
UID:536-1463504400-1463504400@optimisation.doc.ic.ac.uk
SUMMARY:Seminar: A Parametric Approach to Solving the Pooling Problem
DESCRIPTION:Title: A Parametric Approach to Solving the Pooling ProblemSpeaker: Radu Baltean LugojanAffiliation: Department of Computing – Imperial College LondonLocation: Huxley BuildingTime: 5:00pm \nAbstract. We develop an algorithm solving specialised pooling problem instances and generating cutting planes for more generic instances. The approach parameterises the optimisation problem with respect to the pool concentration variables and uncovers embedded sparsity and polyhedral/topological properties for a variety of instances. The presentation generalises and extends recent work analysing computational complexity of the pooling problem [Boland et al. 2015\, Haugland 2016]. Our analysis also integrates source-to-output streams and both upper and lower bounds on the network parameters. \nAbout the speaker. Radu Baltean-Lugojan is a PhD student in the Department of Computing (QUADS group) at Imperial College London\, under the supervision of Dr. Ruth Misener and Dr. Panos Parpas. He received EPSRC funding\, and previously obtained the MEng Computing degree from Imperial College London.
URL:https://optimisation.doc.ic.ac.uk/event/seminar-a-parametric-approach-to-solving-the-pooling-problem/
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=UTC:20160517T160000
DTEND;TZID=UTC:20160517T160000
DTSTAMP:20260418T131354
CREATED:20170124T102132Z
LAST-MODIFIED:20170124T102132Z
UID:537-1463500800-1463500800@optimisation.doc.ic.ac.uk
SUMMARY:Seminar: On the Convergence of Galerkin Type Multilevel Optimization Methods
DESCRIPTION:Title: On the Convergence of Galerkin Type Multilevel Optimization MethodsSpeaker: Chin Pang Ho (Clint)Affiliation: Department of Computing – Imperial College LondonLocation: Huxley BuildingTime: 4:00pm \nAbstract.  \nAbout the speaker. Chin Pang Ho (Clint) is a PhD student in the Department of Computing (QUADS group) at Imperial College\, under the supervision of Dr Panos Parpas. He received a BS in Applied Mathematics from the University of California\, Los Angeles and an MSc in Mathematical Modeling and Scientific Computing from the University of Oxford.
URL:https://optimisation.doc.ic.ac.uk/event/seminar-on-the-convergence-of-galerkin-type-multilevel-optimization-methods/
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=UTC:20160503T170000
DTEND;TZID=UTC:20160503T170000
DTSTAMP:20260418T131354
CREATED:20170124T102132Z
LAST-MODIFIED:20170124T102132Z
UID:538-1462294800-1462294800@optimisation.doc.ic.ac.uk
SUMMARY:Seminar: Integrating Mixed Integer Optimisation and Logic with Satisfiability Modulo Theories
DESCRIPTION:Title: Integrating Mixed Integer Optimisation and Logic with Satisfiability Modulo TheoriesSpeaker: Miten MistryAffiliation: Department of Computing – Imperial College LondonLocation: Room 418 Huxley BuildingTime: 5:00pm \nAbstract. Mixed integer optimisation problems\, especially those involving design or organisation\, often have an inherent logical structure. Existing frameworks to model and utilise such structure reformulate the problem into a mixed integer model or make use of specialised constraints. Using the application of two-dimensional bin packing\, we explore Satisfiability Modulo Theories (SMT) as a means to exploit logical structure. The logical connectives and reasoning provided by SMT allows us to derive cuts to strengthen a Mixed Integer Linear Programming (MILP) solver and\, by using unsatisfiability proofs\, identify new ways of traversing the search tree. \nAbout the speaker. Miten Mistry is a PhD student in the Department of Computing (QUADS group) at Imperial College London\, under the supervision of Dr Ruth Misener. He received EPSRC HiPEDs CDT funding. He previously obtained a MEng Mathematics and Computer Science degree from Imperial College London.
URL:https://optimisation.doc.ic.ac.uk/event/seminar-integrating-mixed-integer-optimisation-and-logic-with-satisfiability-modulo-theories/
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=UTC:20160503T160000
DTEND;TZID=UTC:20160503T160000
DTSTAMP:20260418T131354
CREATED:20170124T102133Z
LAST-MODIFIED:20170124T102133Z
UID:539-1462291200-1462291200@optimisation.doc.ic.ac.uk
SUMMARY:Seminar: Multi-Level Accelerated Algorithm for Large-Scale Convex Composite Minimization
DESCRIPTION:Title: Multi-Level Accelerated Algorithm for Large-Scale Convex Composite MinimizationSpeaker: Vahan HovhannisyanAffiliation: Department of Computing – Imperial College LondonLocation: Room 418 Huxley BuildingTime: 4:00pm \nAbstract. We propose a multi-level algorithm for solving convex composite optimization problems. Our method exploits the fact that many applications that give rise to large-scale problems can be modelled using varying degrees of fidelity. We show that it converges to a minimizer with optimal rate. Using numerical experiments we show that on large-scale computer vision problems our algorithm is several times faster than the state of the art. \nAbout the speaker. Vahan Hovhannisyan is a PhD student in the QUADS group at Imperial College\, under the supervision of Dr Panos Parpas. He received a BS in Applied Mathematics from the State Engineering University of Armenia and an MSc in Applied Mathematical (with application area in operations management) from ETH Zurich. His research interests are convex robust optimization with applications in machine learning.
URL:https://optimisation.doc.ic.ac.uk/event/seminar-multi-level-accelerated-algorithm-for-large-scale-convex-composite-minimization/
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=UTC:20160427T160000
DTEND;TZID=UTC:20160427T160000
DTSTAMP:20260418T131354
CREATED:20170124T102133Z
LAST-MODIFIED:20170124T102133Z
UID:540-1461772800-1461772800@optimisation.doc.ic.ac.uk
SUMMARY:Seminar: Parallel Algorithms and Applications in Structured Large-scale Optimization
DESCRIPTION:Title: Parallel Algorithms and Applications in Structured Large-scale OptimizationSpeaker: Prof. Carl Laird Affiliation: School of Chemical Engineering – Purdue UniversityLocation: Lecture Theatre 2 ACEX BuildingTime: 4:00pm \nAbstract. Mathematical programming has proven to be an efficient tool for design and operation of chemical processes. However\, engineering and scientific needs continue to push the boundaries of existing mathematical programming tools\, often outstripping the capabilities of a single CPU workstation. Furthermore\, computer chip manufacturers are no longer focusing on increasing clock speeds\, and the free performance improvements that we have historically enjoyed will no longer be available\, unless we develop algorithms that are capable of utilizing modern parallel architectures. This presentation discusses advances in parallel algorithms for structured nonlinear mathematical programming problems\, along with a few applications of large-scale optimization. Large-scale optimization formulations arise from a number of different problem classes\, including design and operations under uncertainty\, optimization of complex networks\, and optimization of discretized systems. In this presentation\, I will outline applications in each of these areas. In design of process safety systems\, we have developed advanced stochastic programming formulations for the optimal placement of gas detectors in chemical process facilities based on data from CFD simulations of leak dispersion. As well\, we have partnered with both industry and federal agencies to develop a suite of tools for protecting drinking water distribution systems in the event of accidental or intentional contamination. Our research has focused on improved simulation capabilities\, optimal placement of booster response units\, real-time determination of contamination sources\, and response optimization. Finally\, we have been working with epidemiologists at Johns Hopkins University to develop improved models of infectious disease spread. These dynamic optimization formulations find seasonal patterns in inputs by solving inverse problems based on observed case counts. In particular\, these results help quantify the importance of school-term holiday schedules on the spread of childhood infectious diseases. \nAbout the speaker. Carl Laird is an associate professor in the School of Chemical Engineering at Purdue University. Dr. Laird’s research interests include large-scale nonlinear optimization and parallel scientific computing. Focus areas include chemical process systems\, homeland security applications\, and large-scale infectious disease spread. Dr. Laird is the recipient of several research and teaching awards\, including the CAST Division Outstanding Young Researcher Award\, National Science Foundation Faculty Early Development (CAREER) Award and the Montague Center for Teaching Excellence Award. He is also a recipient of the prestigious Wilkinson Prize for Numerical Software and the IBM Bravo award for his work on IPOPT\, a software library for solving nonlinear\, nonconvex\, large-scale continuous optimization problems. Dr. Laird earned his Ph.D. in Chemical Engineering from Carnegie Mellon in 2006 and his Bachelor of Science in Chemical Engineering from the University of Alberta.
URL:https://optimisation.doc.ic.ac.uk/event/seminar-parallel-algorithms-and-applications-in-structured-large-scale-optimization/
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=UTC:20160321T120000
DTEND;TZID=UTC:20160321T120000
DTSTAMP:20260418T131354
CREATED:20170124T102133Z
LAST-MODIFIED:20170124T102133Z
UID:541-1458561600-1458561600@optimisation.doc.ic.ac.uk
SUMMARY:Seminar: On some tractable optimization models dealing with uncertainty
DESCRIPTION:Title: On some tractable optimization models dealing with uncertaintySpeaker: Prof. Patrick JailletAffiliation: Laboratory for Information and Decision Systems – MITLocation: LG19 seminar room Business SchoolTime: 12:00pm \nAbstract.  In the first part of the talk we consider the minmax regret model for combinatorial optimization problems under uncertainty\, which can be viewed as a zero-sum game played between an optimizing player and an adversary\, where the optimizing player selects a solution and the adversary selects costs with the intention of maximizing the regret of the player. The conventional model considers only deterministic solutions/strategies\, and minmax regret versions of most polynomial solvable problems are NP-hard. In this talk\, we consider a randomized model where the optimizing player selects a probability distribution (corresponding to a mixed strategy) over solutions and the adversary selects costs with knowledge of the player’s distribution\, but not its realization. We show that under this randomized model\, the minmax regret version of any polynomial solvable combinatorial problem becomes polynomial solvable. This holds true for both interval and discrete scenario representations of uncertainty. In the second part of the talk we consider satisficing models\, which\, as an approach to decision-making under uncertainty\, aims at achieving solutions that satisfy the problem’s constraints as much as possible. Mathematical optimization problems that are related to this form of decision-making include the P-model of Charnes and Cooper (1963)\, where satisficing is the objective\, as well as chance-constrained and robust optimization problems\, where satisficing is articulated in the constraints. In this talk\, we introduce the most general framework of a satisficing model\, termed the S-model\, which seeks to maximize a satisficing decision criterion in its objective\, and the corresponding satisficing-constrained optimization problem that generalizes robust optimization and chance-constrained optimization problems. We then focus on a tractable probabilistic S-model\, termed the T-model whose objective is a lower bound of the P-model. \nAbout the speaker. Patrick Jaillet is the Dugald C. Jackson Professor in the Department of Electrical Engineering and Computer Science and a member of the Laboratory for Information and Decision Systems at MIT. He is also one of the two Directors of the MIT Operations Research Center. Before MIT\, he held faculty positions at the University of Texas at Austin and at the Ecole Nationale des Ponts et Chaussees\, Paris. He received a Diplôme d’Ingénieur from France\, and a PhD in Operations Research from MIT. His current research interests include on-line and data-driven optimization under uncertainty. He is a Fellow of INFORMS and a member of SIAM.
URL:https://optimisation.doc.ic.ac.uk/event/seminar-on-some-tractable-optimization-models-dealing-with-uncertainty/
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=UTC:20160309T160000
DTEND;TZID=UTC:20160309T160000
DTSTAMP:20260418T131354
CREATED:20170124T102134Z
LAST-MODIFIED:20170124T102134Z
UID:542-1457539200-1457539200@optimisation.doc.ic.ac.uk
SUMMARY:Seminar: On Optimal selection of fixed-size populations: an application to tree breeding
DESCRIPTION:Title: On Optimal selection of fixed-size populations: an application to tree breedingSpeaker: Dr. Pietro BelottiAffiliation: FICO companyLocation: CPSE seminar room (C615 Roderic Hill)Time: 4:00pm \nAbstract. One of the problems that tree breeders face is the selection of a pedigree of trees with two aims: 1) conserving genetic diversity; 2) maximize response to selection. We tackled the problem of selecting a fixed-size breeding population while imposing a constraint on relatedness of the population members. The problem is expressed as a Mixed Integer Quadratically Constrained Optimization (MIQCO)\, in which a function is maximized subject to nonlinear quadratic constraints an discreteness of some variables\, and solved using a variant of the branch-and-bound method that uses a linear relaxation of the original problem. I will discuss details of the problem and of the algorithm (including a fast heuristic to find feasible solutions). I will also illustrate case studies of the selection of breeding populations for Scots pine and loblolly pine (Joint work with Tim Mullin\, Skogforsk\, the Swedish Forestry Research Institute). \nAbout the speaker. Pietro Belotti received a PhD in Computer Engineering in 2003 from the Technical University of Milan with a dissertation on optimal network design under survivability constraints. He has subsequently held a postdoctoral position at the Tepper School of Business\, Carnegie Mellon University\, a Visiting Professor post at the Department of Industrial and Systems Engineering\, Lehigh University\, and then an Assistant professor position at the department of Mathematical Sciences of Clemson University. He is currently working at Fair Isaac\, in the development team of the Xpress Optimizer. His research interests lie primarily in mixed integer nonlinear optimization\, robust optimization\, and discrete bi-objective optimization.
URL:https://optimisation.doc.ic.ac.uk/event/seminar-on-optimal-selection-of-fixed-size-populations-an-application-to-tree-breeding/
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=UTC:20160226T160000
DTEND;TZID=UTC:20160226T160000
DTSTAMP:20260418T131354
CREATED:20170124T102134Z
LAST-MODIFIED:20170124T102134Z
UID:543-1456502400-1456502400@optimisation.doc.ic.ac.uk
SUMMARY:Seminar: Large-scale MILP and MINLP problems in power system planning
DESCRIPTION:Title: Large-scale MILP and MINLP problems in power system planningSpeaker: Dr. Ioannis KonstantelosAffiliation: Department of Electrical and Electronic Engineering – Imperial College LondonLocation: Room 218 Huxley BuildingTime: 4:00pm \nAbstract. Achieving the ambitious decarbonization goals set by governments worldwide will entail significant changes to the way electrical energy is generated\, transmitted and used. The cost-effective integration of inflexible low-carbon plant within conventional energy systems constitutes a significant challenge. Furthermore\, transmission planners are unable to make fully-informed decisions due to the increasing uncertainty that surrounds future system developments.  Recently\, it has been shown that stochastic system planning based on scenario trees enables the identification of strategic opportunities for the management of long-term uncertainty. However\, the description of these problems is given by large MILP models which contain thousands of binary variables and tens of millions of continuous variables and constraints. After reviewing the general characteristics of the stochastic multi-stage transmission planning problem we will present two novel solution algorithms; one based on hierarchical decomposition and one on temporal problem splitting. We will demonstrate their computational benefits and highlight how efficient solutions can inform the real-world planning process. We will also present a typology of MILP and MINLP problems encountered in energy system planning and operation. \nAbout the speaker. Ioannis Konstantelos is a Research Associate in the Control and Power group\, Electrical Engineering\, Imperial College London. He obtained a PhD from Imperial College in 2013. His work has focused on the development of optimisation models for transmission and distribution system planning and operation aimed at valuing the benefit of new technologies\, demonstrating the strategic value of storage and demand-side response and other flexible technologies when facing long-term uncertainty. His research interests include the application of decomposition and machine learning techniques to large-scale optimization problems for energy systems.
URL:https://optimisation.doc.ic.ac.uk/event/seminar-large-scale-milp-and-minlp-problems-in-power-system-planning/
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=UTC:20160224T160000
DTEND;TZID=UTC:20160224T160000
DTSTAMP:20260418T131354
CREATED:20170124T102134Z
LAST-MODIFIED:20170124T102134Z
UID:544-1456329600-1456329600@optimisation.doc.ic.ac.uk
SUMMARY:Seminar: On the standard pooling problem and strong valid inequalities
DESCRIPTION:Title: On the standard pooling problem and strong valid inequalitiesSpeaker: Dr. Claudia D AmbrosioAffiliation: Laboratory for Information – Ecole PolytechniqueLocation: Room 311 Huxley BuildingTime: 4:00pm \nAbstract. The focus of this talk will be on the standard pooling problem\, i.e.\, a continuous\, non-convex optimization problem arising in the chemical engineering context. First\, we will introduce the problem that consists of finding the optimal composition of final products obtained by blending in pools different percentages of raw materials. Bilinear terms arise from the requirements on the quality of certain attributes of the final products. The quality is a linear combination of the attributes of the raw materials and intermediate products that compose the final product. Three different classical formulations have been proposed in the literature and their characteristics will be discussed and analysed. In the second part of the talk\, strong relaxations for the pooling problem will be presented. In particular\, we studied a structured non-convex subset of some special cases to derive valid nonlinear convex inequalities that we conjecture\, and proved for a particular case\, to define the convex hull of the non-convex subset. Preliminary computational results on instances from the literature are reported and demonstrate the utility of the inequalities when used in a global optimization solver. This is a joint work with Jeff Linderoth (University of Wisconsin-Madison)\, James Luedtke (University of Wisconsin-Madison)\, Jonas Schweiger (IBM). \nAbout the speaker. Claudia D’Ambrosio is a research scientist (chargé de recherche) at CNRS affiliated at LIX\, Ecole Polytechnique (France). She holds a Computer Science Engineering Master Degree and a PhD in Operations Research from University of Bologna (Italy). Her research speciality is mixed integer nonlinear programming. During her whole carrier\, she was involved both in theoretical and applied research projects. She was awarder the EURO Doctoral Dissertation Award for her PhD thesis supervised by Professor Andrea Lodi and the 2nd award “Prix Robert Faure” (3 candidates are awarded every 3 years) granted by ROADEF society. or more detailed info:
URL:https://optimisation.doc.ic.ac.uk/event/seminar-on-the-standard-pooling-problem-and-strong-valid-inequalities/
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=UTC:20160208T160000
DTEND;TZID=UTC:20160208T160000
DTSTAMP:20260418T131354
CREATED:20170124T102135Z
LAST-MODIFIED:20170124T102135Z
UID:545-1454947200-1454947200@optimisation.doc.ic.ac.uk
SUMMARY:Seminar: Revenue-Optimising Scheduling in Parallel Stochastic Networks
DESCRIPTION:Title: Revenue-Optimising Scheduling in Parallel Stochastic NetworksSpeaker: Dr. Giuliano CasaleAffiliation: Department of Computing – Imperial College LondonLocation: Room 145 Huxley BuildingTime: 4:00pm \nAbstract. Cloud applications are often deployed on multiple virtual machines (VMs) with heterogeneous compute capacities. In this setting\, we consider the optimal static scheduling of users to application servers hosted in a set of parallel VMs. Our investigation seeks for a revenue-maximizing solution subject to resource utilization constraints\, multiple classes of users\, and a stochastic queueing-based description of latency experienced by the users at the VMs.After overviewing the general characteristics of scheduling in queueing networks\, and the underpinning optimization programs\, I will show that under a limiting regime this problem reduces to a bilinear optimization program. I will then introduce an heuristic solution for this program and determine an optimality gap. I will also demonstrate the effectiveness of this heuristic in a real system implementation and in comparison to approximate solutions that rely on convex formulations. \nAbout the speaker.
URL:https://optimisation.doc.ic.ac.uk/event/seminar-revenue-optimising-scheduling-in-parallel-stochastic-networks/
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=UTC:20160119T150000
DTEND;TZID=UTC:20160119T150000
DTSTAMP:20260418T131354
CREATED:20170124T102135Z
LAST-MODIFIED:20170124T102135Z
UID:546-1453215600-1453215600@optimisation.doc.ic.ac.uk
SUMMARY:Seminar: The Decision Rule Approach to Optimization Under Uncertainty: Theory and Applications
DESCRIPTION:Title: The Decision Rule Approach to Optimization Under Uncertainty: Theory and ApplicationsSpeaker: Dr. Angelos GeorghiouAffiliation: Automatic Control Laboratory – Swiss Federal Institute of Technology (ETH)Location: Room 217 Huxley BuildingTime: 3:00pm \nAbstract. Decision making under uncertainty has a long and distinguished history in operations research. However\, most of the existing solution techniques suffer from the curse of dimensionality\, which restricts their applicability to small and medium-sized problems\, or they rely on simplifying modeling assumptions (e.g. absence of recourse actions). Recently\, a new solution technique has been proposed\, which is referred to as the decision rule approach. By approximating the feasible region of the decision problem\, the decision rule approach aims to achieve tractability without changing the fundamental structure of the problem. Despite their success\, existing decision rules (a) are typically constrained by their a priori design and (b) do not incorporate in their modeling binary recourse decisions. In this talk\, we present a methodology for the near optimal design of continuous and binary decision rules using mixed-integer optimization\, and demonstrate its potential in operations management applications. \nAbout the speaker. Angelos Georghiou is a post-doctoral researcher with the Automatic Control Laboratory at ETH Zurich. He joined ETH in 2013\, having previously been a post-doctoral researcher at the Process Systems Engineering Laboratory at MIT. He received the MSci degree in Mathematics in 2008 from Imperial College London\, and the Ph.D. degree in Operations Research in 2012 from the Department of Computing at Imperial College London. Angelos’s research focuses on the development of efficient computational methods for the solution of stochastic and robust optimization problems. His work is primarily application driven\, the main application areas being energy systems\, operations management\, and control.
URL:https://optimisation.doc.ic.ac.uk/event/seminar-the-decision-rule-approach-to-optimization-under-uncertainty-theory-and-applications/
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=UTC:20160113T160000
DTEND;TZID=UTC:20160113T160000
DTSTAMP:20260418T131354
CREATED:20170124T102135Z
LAST-MODIFIED:20170124T102135Z
UID:547-1452700800-1452700800@optimisation.doc.ic.ac.uk
SUMMARY:Seminar: Ambiguous Joint Chance Constraints under Mean and Dispersion Information
DESCRIPTION:Title: Ambiguous Joint Chance Constraints under Mean and Dispersion InformationSpeaker: Prof. Daniel KuhnAffiliation: Risk Analytics and Optimization – Ecole Polytechnique Federale De Lausanne (EPFL)Location: Room 217 Huxley BuildingTime: 4:00pm \nAbstract. We study joint chance constraints where the distribution of the uncertain parameters is only known to belong to an ambiguity set characterized by the mean and support of the uncertainties and by an upper bound on their dispersion. This setting gives rise to pessimistic (optimistic) ambiguous chance constraints\, which require the corresponding classical chance constraints to be satisfied for every (for at least one) distribution in the ambiguity set. We provide tight conditions under which pessimistic and optimistic joint chance constraints are computationally tractable\, and we show numerical results that illustrate the power of our tractability results. This is joint work with Grani Hanasusanto\, Vladimir Roitch and Wolfram Wiesemann. \nAbout the speaker. Daniel Kuhn holds the Chair of Risk Analytics and Optimization at EPFL. Before joining EPFL\, he was a faculty member at Imperial College London (2007–2013) and a postdoctoral researcher at Stanford University (2005–2006). He received a Ph.D. in Economics from the University of St. Gallen in 2004 and an M.Sc. in Theoretical Physics from ETH Zürich in 1999. His research interests revolve around robust optimization and stochastic programming.
URL:https://optimisation.doc.ic.ac.uk/event/seminar-ambiguous-joint-chance-constraints-under-mean-and-dispersion-information/
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=UTC:20151201T160000
DTEND;TZID=UTC:20151201T160000
DTSTAMP:20260418T131354
CREATED:20170124T102135Z
LAST-MODIFIED:20170124T102135Z
UID:548-1448985600-1448985600@optimisation.doc.ic.ac.uk
SUMMARY:Seminar: Rescaled coordinate descent methods for Linear Programming
DESCRIPTION:Title: Rescaled coordinate descent methods for Linear ProgrammingSpeaker: Dr. Giacomo ZambelliAffiliation: Department of Management – London School of Economics and Political ScienceLocation: SALC 10 Sherfield Building Time: 4:00pm \nAbstract. Simple coordinate descent methods such as von Neumann’s algorithm or Perceptron\, both developed in the 50s\, can be used to solve linear programming feasibility problems. Their convergence rate depends on the condition measure of the problem at hand\, and is typically not polynomial. Recent work of Chubanov (2012\, 2014)\, related to prior work of Betke (2004)\, has gathered renewed interest in the application of these methods in order to obtain polynomial time algorithms for linear programming. We present two algorithms that fit into this line of research. Both our algorithms alternate between coordinate descent steps and rescaling steps\, so that either the descent step leads to a substantial improvement in terms of the convergence\, or we can infer that the problem is ill conditioned and rescale in order to improve the condition measure. In particular\, both algorithms are based on the analysis of a geometrical invariant of the LP problem\,  used as a proxy for the condition measure\, that appears to be novel in the literature. This is joint work with Daniel Dadush (CWI) and László Végh (LSE) \nAbout the speaker. Dr Zambelli is an Associate Professor (Reader) in the Department of Mathematics at the London School of Economics and Political Science\, which he joined in September 2010. Previously he was  Assistant Professor at the University of Padova. He completed his PhD in Algorithms\, Combinatorics and Optimization at the Tepper School of Business\, Carnegie Mellon University. He is a co-recipient of the 2015 INFORMS Lanchester Prize.
URL:https://optimisation.doc.ic.ac.uk/event/seminar-rescaled-coordinate-descent-methods-for-linear-programming/
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=UTC:20151016T150000
DTEND;TZID=UTC:20151016T150000
DTSTAMP:20260418T131354
CREATED:20170124T102136Z
LAST-MODIFIED:20170124T102136Z
UID:549-1445007600-1445007600@optimisation.doc.ic.ac.uk
SUMMARY:Seminar: Families of Convex and Non-Convex Composite Optimization Problems for Signal Processing and Computer Vision
DESCRIPTION:Title: Families of Convex and Non-Convex Composite Optimization Problems for Signal Processing and Computer VisionSpeaker: Dr. Stefanos ZafeiriouAffiliation: Department of Computing – Imperial College LondonLocation: Room 217 – 218 Huxley BuildingTime: 3:00pm \nAbstract.  \nAbout the speaker. Stefanos Zafeiriou is a Senior Lecturer (equivalent to Associate Professor) in Pattern Recognition/Statistical Machine Learning for Computer Vision in the Department of Computing\, Imperial College London. He has been awarded one of the prestigious Junior Research Fellowships (JRF) from Imperial College London in 2011 to start his own independent research group. He is/has participated in more than 10 EU\, British and Greek research projects. Dr. Zafeiriou currently serves as an Associate Editor in IEEE Transactions on Cybernetics and  Image and Vision Computing journal. He has been guest editor in more than four special issues and co-organized more than five workshops/ special sessions in top venues such as CVPR/FG/ICCV/ECCV.  He has co-authored more than 40 journal papers mainly on novel statistical machine learning methodologies applied to computer vision problems such as 2D/3D face and facial expression recognition\, deformable object tracking\, human behaviour analysis etc published in the most prestigious journals in his field of research (such as IEEE T-PAMI\, IJCV\, IEEE T-IP\, IEEE T-NNLS\, IEEE T-VCG\, IEEE T-IFS etc). His students are frequent recipients of very prestigious and highly competitive fellowships such as Google Fellowship\, Intel Fellowship and the Qualcomm fellowship. He has more than 2000 citations to his work\, h-index 24.
URL:https://optimisation.doc.ic.ac.uk/event/seminar-families-of-convex-and-non-convex-composite-optimization-problems-for-signal-processing-and-computer-vision/
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=UTC:20151014T153000
DTEND;TZID=UTC:20151014T153000
DTSTAMP:20260418T131354
CREATED:20170124T102136Z
LAST-MODIFIED:20170124T102136Z
UID:550-1444836600-1444836600@optimisation.doc.ic.ac.uk
SUMMARY:Seminar: Generating structured music with local search and machine learning
DESCRIPTION:Title: Generating structured music with local search and machine learningSpeaker: Dr. Dorien HerremansAffiliation: School of Electronic Engineering and Computer Science – Queen Mary UniversityLocation: LT 144 Huxley BuildingTime: 3:30pm \nAbstract. Many state of the art music generation/improvisation systems generate music that sounds good on a note-to-note level. However\, these compositions often lack long term structure or coherence. By looking at generating music as an optimization problem\, this research overcomes this problem and generates music that has a larger structure. A powerful variable neighbourhood search algorithm (VNS) was developed\, which is able to generate a range of musical styles based on it’s objective function\, whilst constraining the music to a structural template. In the first stage of the project\, an objective function based on rules from music theory was used to generate counterpoint. In this research\, a machine learning approach is combined with the VNS in order to generate structured music for the bagana\, an Ethiopian lyre. Different ways are explored in which a Markov model can be used to construct quality metrics that represent how well a fragment fits the chosen style (e.g. music for bagana). Current research that aims to extend the objective function with models such as recursive neural networks is also briefly discussed. The approach followed in this research allows us to combine the power of machine learning methods with optimization algorithms. \nAbout the speaker. Dorien Herremans is currently a Marie Skodowska-Curie Postdoctoral Fellow at C4DM\, Queen Mary University of London. She got her PhD in Operations Research on the topic of Computer Generation and Classification of Music through Operations Research Methods (Compose: Compute – Generating and Classifying Music through Operations Research Methods). She graduated as a commercial engineer in management information systems at the University of Antwerp in 2005. After that\, she worked as a Drupal consultant and was an IT lecturer at the Les Roches University in Bluche\, Switzerland. She also worked as a mandaatassistent at the University of Antwerp\, in the domain of operations management\, supply chain management and operations research (OR). Her current research focuses on applications of OR in the field of music.
URL:https://optimisation.doc.ic.ac.uk/event/seminar-generating-structured-music-with-local-search-and-machine-learning/
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=UTC:20150902T110000
DTEND;TZID=UTC:20150902T110000
DTSTAMP:20260418T131354
CREATED:20170124T102136Z
LAST-MODIFIED:20170124T102136Z
UID:551-1441191600-1441191600@optimisation.doc.ic.ac.uk
SUMMARY:Seminar: A two-phase proximal augmented Lagrangian method for large scale convex composite quadratic programming
DESCRIPTION:Title: A two-phase proximal augmented Lagrangian method for large scale convex composite quadratic programmingSpeaker: Prof. Kim Chuan TohAffiliation: Department of Mathematics – National University of SingaporeLocation: CPSE seminar room (C615 Roderic Hill)Time: 11:00am \nAbstract. We consider an important class of high dimensional convex composite quadratic optimization problems with large numbers of linear equality and inequality constraints. Our work is motivated by the recent interests in convex quadratic conic programming problems\, as well as from convex quadratic programming problems with dual block angular structures such as those arising from two stage stochastic programming problems. In this talk\, we first introduce a symmetric Gauss-Seidel (sGS) decomposition theorem for solving an unconstrained convex composite programming problem whose objective is the sum of a multi-block quadratic function and a non-smooth function involving only the first block. Then\, based on the sGS decomposition theorem\, we propose a two phase proximal augmented Lagrangian method to efficiently solve the targeted problem to high accuracy. Specifically\, in Phase I\, we design an inexact sGS-based semi-proximal ADMM to generate a reasonably good initial point to warm-start the algorithm in Phase II\, which is a semi-smooth NewtonCG based proximal augmented Lagrangian method capable of computing a high accuracy solution efficiently. \nAbout the speaker. Kim-Chuan Toh is a Professor at the Department of Mathematics\, National University of Singapore (NUS). He obtained his Bachelor degree from NUS in 1990 and the PhD degree from Cornell University in 1996 under the guidance of Professor Nick Trefethen. He is currently an Area Editor for Mathematical Programming Computation\, and an Associate Editor for the SIAM Journal on Optimization. His research focuses on designing efficient algorithms and software for convex programming\, particularly large scale matrix optimization problems such as semidefinite programming (SDP) and convex quadratic SDP.
URL:https://optimisation.doc.ic.ac.uk/event/seminar-a-two-phase-proximal-augmented-lagrangian-method-for-large-scale-convex-composite-quadratic-programming/
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=UTC:20150519T140000
DTEND;TZID=UTC:20150519T140000
DTSTAMP:20260418T131354
CREATED:20170124T102136Z
LAST-MODIFIED:20170124T102136Z
UID:552-1432044000-1432044000@optimisation.doc.ic.ac.uk
SUMMARY:Seminar: Estimating variance matrices
DESCRIPTION:Title: Estimating variance matricesSpeaker: Prof. Karim Abadir Affiliation: Imperial College Business SchoolLocation: Room 218 Huxley BuildingTime: 2:00pm \nAbstract.  This talk introduces a new method for estimating variance matrices. Starting from the orthogonal decomposition of the sample variance matrix\, we exploit the fact that orthogonal matrices are never ill-conditioned and therefore focus on improving the estimation of the eigenvalues. We estimate the eigenvectors from just a fraction of the data\, then use them to transform the data into approximately orthogonal series that deliver a well-conditioned estimator (by construction)\, even when there are fewer observations than dimensions. We also show that our estimator has lower error norms than the traditional one. Our estimator is design-free: we make no assumptions on the distribution of the random sample or on any parametric structure the variance matrix may have. Simulations confirm our theoretical results and they also show that our simple estimator does very well in comparison with other existing methods\, especially when the data are generated from fat-tailed densities.  \nAbout the speaker.  Karim Abadir is the Chair of Financial Econometrics at Imperial College\, London. He obtained his DPhil from Oxford University. His MA (Economics) and BA (Major in Economics\, Minor in Business) are from the American University in Cairo. He started his academic career as a lecturer in Economics at Lincoln College\, Oxford. He then joined the University of Exeter as a Senior Lecturer in Statistics and Econometrics\, rising to the position of Reader in Econometrics. From 1996-2005\, he held a Chair in Econometrics and Statistics at the University of York\, joint between the Departments of Mathematics and Economics.He is credited with having solved in his DPhil a major long-standing problem in Mathematical Statistics and Time Series that was open since the 1950’s. More recently\, he has predicted the timing of the 2008 recession a year in advance\, and the different timings of the recoveries in various Western countries.He is a founding member of the liberal party Al Masreyeen Al Ahrrar (translates as Free/Liberal Egyptians)\, established in 2011.
URL:https://optimisation.doc.ic.ac.uk/event/seminar-estimating-variance-matrices/
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=UTC:20150513T110000
DTEND;TZID=UTC:20150513T110000
DTSTAMP:20260418T131354
CREATED:20170124T102136Z
LAST-MODIFIED:20170124T102136Z
UID:553-1431514800-1431514800@optimisation.doc.ic.ac.uk
SUMMARY:Seminar: Computational Progress in Linear and Mixed Integer Programming
DESCRIPTION:Title: Computational Progress in Linear and Mixed Integer ProgrammingSpeaker: Dr. Robert BixbyAffiliation: GurobiLocation: CPSE seminar room (C615 Roderic Hill)Time: 11:00am \nAbstract.  We will look at the progress in linear and mixed-integer programming software over the last 25 years.   As a result of this progress\, modern linear programming codes are now capable of robustly and efficiently solving instances with multiple millions of variables and constraints.   With these linear programming advances as a foundation\, mixed-integer programming then provides the modeling framework and solution technology that enables the overwhelming majority of present-day business planning and scheduling applications\, and is the key technology behind prescriptive analytics.   The performance improvements in mixed-integer programming code overs the last 25 years have been nothing short of remarkable\, well beyond those of linear programming and have transformed this technology into an out-of-the box tool with applications to an almost unlimited range of real-world problems.   \nAbout the speaker.
URL:https://optimisation.doc.ic.ac.uk/event/seminar-computational-progress-in-linear-and-mixed-integer-programming/
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=UTC:20150507T110000
DTEND;TZID=UTC:20150507T110000
DTSTAMP:20260418T131354
CREATED:20170124T102136Z
LAST-MODIFIED:20170124T102136Z
UID:554-1430996400-1430996400@optimisation.doc.ic.ac.uk
SUMMARY:Seminar: A bilevel programming problem occurring in smart grids
DESCRIPTION:Title: A bilevel programming problem occurring in smart gridsSpeaker: Prof. Leo LibertiAffiliation: Ecole Polytechnique ParisLocation: CPSE seminar room (C615 Roderic Hill)Time: 11:00am \nAbstract. A key property to define a power grid “smart” is its real-time\, fine-grained monitoring capabilities. For this reason\, a variety of monitoring equipment must be installed on the grid. We look at the problem of fully monitoring a power grid by means of Phasor Measurement Units (PMUs)\, which is a graph covering problem with some equipment-specific constraints. We show that\, surprisingly\, a bilevel formulation turns out to provide the most efficient algorithm. \nAbout the speaker. Leo Liberti obtained his B.Sc. in Mathematics and his Ph.D. in Process Systems Engineering from Imperial College. He became a professor at Ecole Polytechnique (France)\, then a Research Staff Member at IBM Research (USA). He was recently appointed Research Director at CNRS and part-time professor back at Ecole Polytechnique. His research interests are Mixed-Integer Nonlinear Programming and Distance Geometry.
URL:https://optimisation.doc.ic.ac.uk/event/seminar-a-bilevel-programming-problem-occurring-in-smart-grids/
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=UTC:20150505T150000
DTEND;TZID=UTC:20150505T150000
DTSTAMP:20260418T131354
CREATED:20170124T102137Z
LAST-MODIFIED:20170124T102137Z
UID:555-1430838000-1430838000@optimisation.doc.ic.ac.uk
SUMMARY:Seminar: A cycle-based formulation and valid inequalities for DC power transmission problems with switching
DESCRIPTION:Title: A cycle-based formulation and valid inequalities for DC power transmission problems with switchingSpeaker: Prof. Jeff LinderothAffiliation: Departments of Industrial and Systems Engineering and Computer Sciences (by courtesy) – University of Wisconsin-MadisonLocation: LT 145 Huxley BuildingTime: 3:00pm \nAbstract. It is well-known that optimizing network topology by switching on and off transmission lines improves the efficiency of power delivery in electrical networks. Many authors have studied the problem of determining an optimal set of transmission lines to switch off to minimize the cost of meeting a given power demand under the direct current (DC) model of power flow. This problem is known in the literature as the Direct-Current Optimal Transmission Switching Problem (DC-OTS). Most research on DC-OTS has focused on heuristic algorithms for generating quality solutions or on the application of DC-OTS to crucial operational and strategic problems. The mathematical theory of the DC-OTS problem is less well-developed. In this work\, we formally establish that DC-OTS is NP-Hard\, even if the power network is a series-parallel graph with at most one load/demand pair. Inspired by Kirchoff’s Voltage Law\, we give a cycle-based formulation for DC-OTS\, and we use the new formulation to build a cycle-induced relaxation. We characterize the convex hull of the cycle-induced relaxation\, and the characterization provides strong valid inequalities that can be used in a cutting-plane approach to solve the DC-OTS. We give details of a practical implementation\, and we show promising computational results on standard benchmark instances.  Co-authors:This is joint work with: Burak Kocuk\, Santanu Dey\, Andy Sun (Georgia Tech)\, Hyemin Jeon\, and Jim Luedtke (Wisconsin) \nAbout the speaker. Jeff Linderoth is a Professor in the departments of Industrial and Systems Engineering and Computer Sciences (by courtesy) at the University of Wisconsin-Madison\, joining both departments in 2007. Dr. Linderoth received his Ph.D. degree from the Georgia Institute of Technology in 1998.  He was awarded an an Early Career Development Award from the Department of Energy\, and he has won the SIAM/Activity Group on Optimization Prize and the INFORMS Computing Society ICS Prize.  Dr. Linderoth currently serves on the editorial boards of 4 journals\, including Operations Research and Mathematical Programming Computation.
URL:https://optimisation.doc.ic.ac.uk/event/seminar-a-cycle-based-formulation-and-valid-inequalities-for-dc-power-transmission-problems-with-switching/
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=UTC:20150413T140000
DTEND;TZID=UTC:20150413T140000
DTSTAMP:20260418T131354
CREATED:20170124T102137Z
LAST-MODIFIED:20170124T102137Z
UID:556-1428933600-1428933600@optimisation.doc.ic.ac.uk
SUMMARY:Seminar: Are Targets for Renewable Portfolio Standards Too Low? - The Impact of Market Structure on Energy Policy? (joint work with Makoto Tanaka and Yihsu Chen)
DESCRIPTION:Title: Are Targets for Renewable Portfolio Standards Too Low? – The Impact of Market Structure on Energy Policy? (joint work with Makoto Tanaka and Yihsu Chen)Speaker: Dr Afzal SiddiquiAffiliation: Department of Statistical Science – University College LondonLocation: Room 217 Huxley BuildingTime: 2:00pm \nAbstract. In order to limit climate change from greenhouse gas emissions\, governments have introduced renewable portfolio standards (RPS) to incentivize renewable energy production. While the response of industry to exogenous RPS targets has been addressed in the literature\, setting RPS targets from a policymaker’s perspective has remained an open question. Using a bi-level model\, we prove that the optimal RPS target for a perfectly competitive electricity industry is higher than that for a benchmark centrally planned one. Allowing for market power by the non-renewable energy sector within a deregulated industry lowers the RPS target vis-à-vis perfect competition. Moreover\, to our surprise\, social welfare under perfect competition with RPS is lower than that when the non-renewable energy sector exercises market power. In effect\, by subsidizing renewable energy and taxing the non-renewable sector\, RPS represents an economic distortion that over-compensates damage from emissions. Thus\, perfect competition with RPS results in “too much” renewable energy output\, whereas the market power of the non-renewable energy sector mitigates this distortion\, albeit at the cost of lower consumer surplus and higher emissions. Hence\, ignoring the interaction between RPS requirements and the market structure could lead to sub-optimal RPS targets and substantial welfare losses. \nAbout the speaker. Afzal Siddiqui is a Senior Lecturer in the Department of Statistical Science. Previously\, he was a Lecturer in Statistics at UCL (2005-2010) and a College Lecturer in the Department of Banking and Finance at University College Dublin. After having completed his Ph.D. in Industrial Engineering and Operations Research from the University of California at Berkeley in 2002\, Afzal served as a Visiting Assistant Professor in the Department of Industrial Engineering and Operations Research at UC Berkeley (2002) and a Visiting Post-doctoral Researcher at the Ernest Orlando Lawrence Berkeley National Laboratory (2002-2003). In addition\, he is a Professor (20% time) at the Department of Computer and Systems Sciences of Stockholm University and a Visiting Professor at the Systems Analysis Laboratory of Aalto University.
URL:https://optimisation.doc.ic.ac.uk/event/seminar-are-targets-for-renewable-portfolio-standards-too-low-the-impact-of-market-structure-on-energy-policy-joint-work-with-makoto-tanaka-and-yihsu-chen/
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=UTC:20150325T140000
DTEND;TZID=UTC:20150325T140000
DTSTAMP:20260418T131354
CREATED:20170124T102137Z
LAST-MODIFIED:20170124T102137Z
UID:557-1427292000-1427292000@optimisation.doc.ic.ac.uk
SUMMARY:Seminar: The Complexity of Primal-Dual Fixed Point Methods for Ridge Regression
DESCRIPTION:Title: The Complexity of Primal-Dual Fixed Point Methods for Ridge RegressionSpeaker: Prof. Ademir RibeiroAffiliation: Department of Mathematics – Federal University of ParanaLocation: Room 218 Huxley BuildingTime: 2:00pm \nAbstract.  We study the ridge regression (L2 regularized least squares) problem and its dual\, which is also a ridge regression problem. We observe that the optimality conditions can be formulated in several different but equivalent ways\, in the form of a linear system involving a structured matrix depending on a single “stepsize” parameter which we introduce for regularization purposes. This leads to the idea of studying and comparing\, in theory and practice\, the performance of the fixed point method applied to these reformulations. We compute the optimal stepsize parameters and uncover interesting connections between the complexity bounds of the variants of the fixed point scheme we consider. These connections follow from a close link between the spectral properties of the associated matrices. For instance\, some reformulations involve purely imaginary eigenvalues; some involve real eigenvalues and others have all eigenvalues on the complex circle. We show that the deterministic Quartz method—which is a special case of the randomized dual coordinate ascent method with arbitrary sampling recently developed by Qu\, Richtarik and Zhang—can be cast in our framework\, and achieves the best rate in theory and in numerical experiments among the fixed point methods we study. This is joint work with Peter Richtarik (Edinburgh). \nAbout the speaker.  I am an Associate Professor at Department of Mathematics\, Federal University of Parana\, Brazil\, since 1992. I got my undergraduate degree in Mathematics at Federal University of Parana in 1989. In 1993 I finished my MSc in Mathematics\, at IMPA – National Institute for Pure and Applied Mathematics. I got my PhD in Optimization at Federal University of Parana in 2005.  My current research interests are applied mathematics\, continuous optimization\, global and local convergence of algorithms as filter and trust region methods for nonlinear programming and convex optimization\, complexity of direct search methods\, among others. I have been published papers in journals like Applied Mathematics and Computation\, Applied Mathematical Modelling\, Optimization\, Computational Optimization and Applications\, Mathematical Programming and SIAM Journal on Optimization. I have been given talks in some meetings like Optimization Conference (Porto 2007 and Guimarães 2014)\, Brazilian Workshop on Continuous Optimization (Rio de Janeiro 2009 and Florianópolis 2014) and International Symposium on Mathematical Programming (Rio de Janeiro 2006 and Berlin 2012).  I have supervised 2 PhD and 6 MSc students.  In joint work with Elizabeth Wegner Karas\, I also have published a book (in Portuguese)\, called Continuous Optimization: Theoretical and computational aspects. Cengage Learning\, Sao Paulo\, Brazil\, 2013.
URL:https://optimisation.doc.ic.ac.uk/event/seminar-the-complexity-of-primal-dual-fixed-point-methods-for-ridge-regression/
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=UTC:20150317T140000
DTEND;TZID=UTC:20150317T140000
DTSTAMP:20260418T131354
CREATED:20170124T102137Z
LAST-MODIFIED:20170124T102137Z
UID:558-1426600800-1426600800@optimisation.doc.ic.ac.uk
SUMMARY:Seminar: Bifurcation Analysis Using Complete Search Methods
DESCRIPTION:Title: Bifurcation Analysis Using Complete Search Methods Speaker: Dr. Mario Eduardo VillanuevaAffiliation: Centre for Process Systems Engineering at Imperial CollegeLocation: Room 217 Huxley BuildingTime: 2:00pm \nAbstract. When studying a non-linear dynamic system it is important to locate and characterise its equilibrium manifold and its bifurcation regions within a pre-specified computational domain. Except for very few simple cases\, an algebraic characterisation of such manifold is impossible. In this seminar a methodology for locating the equilibrium manifold of non-linear dynamic systems defined by parametric ODEs is presented. This methodology is based on a set-inversion approach which uses state-of-the-art bounding techniques within a complete search algorithm. The efficacy of this approach is illustrated with a challenging non-linear model of an anaerobic digestion process.  \nAbout the speaker. Mario Eduardo Villanueva (MEV) graduated with a BEng in Biochemical Engineering from the Instituto Tecnologico de Veracruz\, Mexico and an MSc in Advanced Chemical Engineering with Process Systems Engineering from Imperial College London. He is currently a PhD student in the Department of Chemical engineering\, under the supervision of Dr. B. Chachuat. His PhD project is concerned with the development of methods and tools for complete search in uncertain dynamic systems.
URL:https://optimisation.doc.ic.ac.uk/event/seminar-bifurcation-analysis-using-complete-search-methods/
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=UTC:20150313T150000
DTEND;TZID=UTC:20150313T150000
DTSTAMP:20260418T131354
CREATED:20170124T102137Z
LAST-MODIFIED:20170124T102137Z
UID:559-1426258800-1426258800@optimisation.doc.ic.ac.uk
SUMMARY:Seminar: Supply Chain Optimization - from strategic to operational decision levels
DESCRIPTION:Title: Supply Chain Optimization – from strategic to operational decision levelsSpeaker: Prof. Ana Barbosa-PovoaAffiliation: University of LisbonLocation: CPSE seminar room (C615 Roderic Hill)Time: 3:00pm \nAbstract. Supply Chains are complex systems that involve challenging problems. Such problems require suitable answers so as to guarantee efficiency and responsiveness improvements of the involved systems. When answering to such problems optimization is a possible path to follow aiming at building tools that can help the decision makers on the problems solutions that span from strategic to operational levels. The scientific community has been exploring this pathway but much more is required\, especially due to the outer shell of new emerging problems. The present talk characterizes supply chain decisions and presents some of the work that has been done on the optimization of supply chains detailing specially the work developed by the Operations and Logistics Group of the Centre for Management Studies at Instituto Superior Técnico (IST) in Lisbon. We conclude with a discussion of the tendencies and future challenges in the area. \nAbout the speaker. Ana Barbosa-Póvoa obtained her PhD in Engineering from Imperial College of Science Technology and Medicine. She is currently a Full Professor of Operations and Logistics at the Department of Management and Engineering of Instituto Superior Técnico (IST)\, University of Lisbon\, Portugal where she is the director of the BSc and Master Programs in Engineering and Management. She is a member of the scientific council of IST and of the University Senate. She is also the Vice-president of the Portuguese Association for Operational Research. She has been acting as reviewer to several national and international research scientific boards on research projects. Her research interests are on the supply chain management\, where both forward and reserve structures are included and on the design\, planning and scheduling of flexible systems. Ana has published widely in these areas and supervised several Master and PhD students. Ana in 2008 has received the scientific award of Technical University of Lisbon in the scientific area of Industrial Management.
URL:https://optimisation.doc.ic.ac.uk/event/seminar-supply-chain-optimization-from-strategic-to-operational-decision-levels/
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=UTC:20150303T140000
DTEND;TZID=UTC:20150303T140000
DTSTAMP:20260418T131354
CREATED:20170124T102138Z
LAST-MODIFIED:20170124T102138Z
UID:560-1425391200-1425391200@optimisation.doc.ic.ac.uk
SUMMARY:Seminar: Formal Proofs for Nonlinear Optimization
DESCRIPTION:Title: Formal Proofs for Nonlinear OptimizationSpeaker: Dr. Victor MagronAffiliation: Department of Electrical and Electronic Engineering – Imperial College LondonLocation: Room 217 Huxley BuildingTime: 2:00pm \nAbstract. We present a formally verified global optimization framework. Given a semialgebraic or transcendental function f and a compact semialgebraic domain K\, we use the nonlinear maxplus template approximation algorithm to provide a certified lower bound of f over K. This algorithm allows to bound in a modular way some of the constituents of f by suprema of quadratic forms with a well chosen curvature. Thus\, we reduce the initial goal to a hierarchy of semialgebraic optimization problems\, solved by semidefinite relaxations.  Our implementation tool interleaves semialgebraic approximations with sums of squares witnesses to form certificates. It is interfaced with Coq and thus benefits from the trusted arithmetic available inside the proof assistant. This feature is used to produce\, from the certificates\, both valid underestimators and lower bounds for each approximated constituent. The application range for such a tool is widespread; for instance Hales’ proof of Kepler’s conjecture yields thousands of multivariate transcendental inequalities. We illustrate the performance of our formal framework on some of these inequalities as well as on examples from the global optimization literature.http://cas.ee.ic.ac.uk/people/vmagron/slides/quads.pdf \nAbout the speaker. Victor graduated from Ecole Centrale Paris Engineering School in 2010\, while receiving his MSc from the department of Systems Innovation\, Tokyo University (double diploma). In 2013\, he received his PhD in Computer Science at INRIA Saclay\, Ecole Polytechnique. In 2014\, he was a Postdoc fellow in the MAC team at LAAS in Toulouse\, France. He is currently a Research Assistant at Imperial College for the Circuits and Systems group\, in the department of Electrical and Electronic Engineering.
URL:https://optimisation.doc.ic.ac.uk/event/seminar-formal-proofs-for-nonlinear-optimization/
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=UTC:20150211T110000
DTEND;TZID=UTC:20150211T110000
DTSTAMP:20260418T131354
CREATED:20170124T102138Z
LAST-MODIFIED:20170124T102138Z
UID:561-1423652400-1423652400@optimisation.doc.ic.ac.uk
SUMMARY:Seminar: Worst-case complexity of nonlinear optimization: Where do we stand?
DESCRIPTION:Title: Worst-case complexity of nonlinear optimization: Where do we stand?Speaker: Prof. Philippe TointAffiliation: Department of Mathematics – Université de NamurLocation: CPSE seminar room (C615 Roderic Hill)Time: 11:00am \nAbstract. We review the available results on the evaluation complexity of algorithms using Lipschitz-continuous Hessians for the approximate solution of nonlinear and potentially nonconvex optimization problems. Here\, evaluation complexity is a bound on the largest number of problem functions (objective\, constraints) and derivatives evaluations that are needed before an approximate first-order critical point of the problem is guaranteed to be found. We start by considering the unconstrained case and examine classical methods (such as Newton’s method) and the more recent ARC2 method\, which we show is optimal under reasonable assumptions. We then turn to constrained problems and analyze the case of convex constraints first\, showing that a suitable adaptation ARC2CC of the ARC2 approach also possesses remarkable complexity properties. We finally extend the results obtained in simpler settings to the general equality and inequality constrained nonlinear optimization problem by constructing a suitable ARC2GC algorithm whose evaluation complexity also exhibits the same remarkable properties. \nAbout the speaker. Philippe L. Toint (born 1952) received its degree in Mathematics in the University of Namur (Belgium) in 1974 and his Ph.D. in 1978 under the guidance of Prof M.J.D. Powell. He was appointed as lecturer at the University of Namur in 1979 were he became associate professor in 1987 and full-professor in 1993. Since 1979\, he has been the co-director of the Numerical Analysis Unit and director of the Transportation Research Group in this department. He was in charge of the University Computer Services from 1998 to 2000 and director of the Department of Mathematics from 2006 to 2009. He currently serves as Vice-rector for Research and IT for the university. His research interests include numerical optimization\, numerical analysis and transportation research. He has published four books and more than 280 papers and technical reports. Elected as SIAM Fellow (2009)\, he was also awarded the Beale-Orchard-Hayes Prize (1994\, with Conn and Gould)) and the Lagrange Prize in Continuous Optimization (2006\, with Fletcher and Leyffer). He is the past Chairman (2010-2013) of the Mathematical Programming Society\, the international scientific body gathering most researchers in mathematical optimization world-wide. Married and father of two girls\, he is a keen music and poetry lover as well as an enthusiast scuba-diver.
URL:https://optimisation.doc.ic.ac.uk/event/seminar-worst-case-complexity-of-nonlinear-optimization-where-do-we-stand/
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=UTC:20150206T150000
DTEND;TZID=UTC:20150206T150000
DTSTAMP:20260418T131354
CREATED:20170124T102138Z
LAST-MODIFIED:20170124T102138Z
UID:562-1423234800-1423234800@optimisation.doc.ic.ac.uk
SUMMARY:Seminar: McCormick Relaxations: Convergence Rate and Extension to Multivariate Outer Functions
DESCRIPTION:Title: McCormick Relaxations: Convergence Rate and Extension to Multivariate Outer FunctionsSpeaker: Prof. Alexander MitsosAffiliation: Laboratory for Process Systems Engineering – RWTH Aachen UniversityLocation: CPSE seminar room (C615 Roderic Hill)Time: 3:00pm \nAbstract. Optimization is a widely used tool in process systems engineering\, but often the optimization problems have multiple suboptimal local minima. Deterministic global optimization algorithms can solve such problems\, typically employing convex/concave relaxations of the objective and constraints. Several methods have been proposed for the construction of convergent relaxations\, including the McCormick relaxations. These provide the framework for the computation of convex relaxations of composite functions. McCormick’s relaxations are clearly a very important tool\, but they have the limitation of only allowing univariate composition. Although most functions can be decomposed in a way that only univariate functions are used as building blocks\, this often results in weak relaxations. Moreover\, McCormick has not provided results for the convergence rate of these relaxations. We propose a reformulation of McCormick’s composition theorem\, which while equivalent to the original\, suggests a straight forward generalization to multi-variate outer functions. In addition to extending the framework\, the multi-variate McCormick relaxation is a useful tool for the proof of relaxations: by direct application to the product\, division and minimum/maximum of two functions\, we obtain improved relaxations when comparing with uni-variate McCormick. Furthermore\, we generalize the theory for the computation of subgradients to the multi-variate case\, envisioning practical methods that utilize the framework. Further\, we extend the notion of convergence order from interval extensions to convex relaxations in the pointwise metric and Hausdorff metric. We develop theory for the McCormick relaxations by establishing convergence rules for the addition\, multiplication and composition operations. The convergence order of the composite function depends on the convergence order of the relaxations of the factors. No improvement in the order of convergence compared to that of the underlying bound calculation\, e.g.\, via interval extensions\, can be guaranteed unless the relaxations of the factors have pointwise convergence of high order\, in which case at least quadratic conver- gence order can be guaranteed. Additionally\, the McCormick relaxations are compared with the alphaBB relaxations by Floudas and coworkers\, which also guarantee quadratic pointwise convergence. Finally\, the convergence order of McCormick-Taylor models is addressed. Illustrative and numerical examples are given and hybrid methods are discussed. The implication of the results are discussed for practical bound calculations as well as for convex/concave relaxations of factors commonly found in process systems engineering models. \nAbout the speaker. Alexander Mitsos is a Full Professor (W3) in RWTH Aachen University\, and the Director of the Laboratory for Process Systems Engineering (AVT.SVT)\, comprising 40 research and administrative staff. Mitsos received his Dipl-Ing from University of Karlsruhe in 1999 and his Ph.D.  from MIT in 2006\, both in Chemical Engineering. Prior appointments include military service\, free-lance engineering\, involvement in a start-up company\, a junior research group leader position in the Aachen Institute of Computational Engineering Science and the Rockwell International Assistant Professorship at MIT. Mitsos has over 60 publications in peer-reviewed journals and has received a number of awards. His research focuses on optimization of energy and chemical systems and development of enabling numerical algorithms.
URL:https://optimisation.doc.ic.ac.uk/event/seminar-mccormick-relaxations-convergence-rate-and-extension-to-multivariate-outer-functions/
END:VEVENT
END:VCALENDAR