OPTEC this week: John Betts et al.
Dear OPTEC Members and Friends,
as previously announced, this week is full of events, altogether seven. Five are related to this winter's Stevin lecturer, John Betts, who gives the Stevin lecture tomorrow Wed at 16:00 on Algorithmic Choices when solving an optimal control problem (in CS). He also gives a short optimal control course starting already Wed morning (Wed and Thu morning, in the castle) and two scientific seminars (Thu afternoon, in CS), the last one being its famous "Discretize then optimize" Oberwolfach talk.
Before, today, we have Dr. Tobias Jensen from Aalborg talking about fast gradient
methods for L1 problems, speaking at 11:00, and from 14:00 to 18:00 there is
an internal OPTEC workshop on tensors and large scale optimization (in ESAT).
All abstracts are below, as well as on http://www.kuleuven.be/optec/event
Best regards,
Moritz Diehl
**********
Start date: 7/02/2012
11:00 - 12:00
Location: ESAT 00.62
"A Lower Complexity Bound for l1-regularized Least-squares Problems using a Certain Class of Algorithms"
Tobias Lindstrøm Jensen
Deparment of Electronic Systems at Aalborg University
The l1-regularized least-squares problem have received broad attention the last couple of years. The result is numerous approaches for reliable large-scale solvers which combines both well known methods and recently developed techniques for efficient computations. We define a class of algorithms which is not as restrictive as classic black-box algorithms and hence includes most of the recently proposed methods. We show how to obtain a worst-case convergence rate for all these methods.
**********
OPTEC MANET Workshop on Tensors and Large Scale Optimization
Tue 07 Feb 2012 14:00-18:00, ESAT 00.62
"OPTEC MANET Workshop on Tensors and Large Scale Optimization"
This internal workshop brings together OPTEC professors, postdocs, and a few PhD students that are interested in Tensors and in Large Scale Optimization Algorithms. Each of the present professors gives a 10 minute statement on large scale optimization problems and algorithms in his/her group.
The workshop is related to OPTEC's Special Interest Groups on "Tensors and Optimization" (coordinated by Lieven De Lathauwer) and "Distributed and Parallel Methods for Optimization" (coordinated by Toon van Waterschoot), as well as to the GOA project MANET. The workshop will reserve 50% of its time for discussions. Aim is to find synergies between the groups.
Program:
12:00 Joint Alma lunch for all who like to join
14:00 Start and introduction
14:10 Lieven De Lathauwer: Tensor Optimization Problems
14:20 Discussion
14:30 Johan Suykens: Large Scale Optimization in Machine Learning
14:40 Discussion
14:50 Moritz Diehl: Distributed Optimization Algorithms
15:00 Discussion
15:10 Toon van Waterschoot: Distributed Optimization Problems in Signal Processing
15:20 Discussion
15:30 Coffee break
16:00 Sabine Van Huffel: Tensor Optimization in Biomedical Applications
16:10 Discussion
16:20 Alexander Bertrand: Consensus-based Distributed Total Least Squares Estimation in Ad-hoc Networks
16:35 Laurent Sorber: Efficient Algorithms for Tensor Decompositions
16:50 Sam Weckx: Distributed Optimization in Smart Grid Applications
17:05 Marco Signoretto: Learning Tensor-based Models with Structure-inducing Penalties
17:20 future planning of cooperations and wrap up
18:00 end
Workshop Organizers: Lieven De Lathauwer, Toon van Waterschoot, Moritz Diehl
*******************
Wed 08 Feb 2012 09:00-12:00, Auditorium of the Arenberg Castle 01.07
"The Direct Transcription Method for Optimal Control"
John Betts
Abstract:
Part 1: Nonlinear Programming (Feb 8, 9-12)
An overview of the methods for solving a finite dimensional optimization problem with constraints will be presented. A brief review of fundamentals suitable for small problems will be given. Then a description of what changes when the problem becomes "large" and the underlying matrices are "sparse". Details of a large sparse sequential quadratic programming (SQP) method and a large sparse primal-dual interior point (barrier) algorithm will then be presented.
Part 2: Optimal Control (Feb 9, 9-12)
Abstract: Methods for "transcription" or conversion of a problem formulated in terms of differential and/or differential-algebraic equations will be described. A brief review of methods for numerical solution of dynamic systems, will be followed by a discussion of the optimal control problem. Various alternatives for discretization will be described, as well as a technique for mesh refinement. Extensions of the paradigm to parameter estimation or "inverse problems" will also be covered.
*******************
Simon Stevin Lecture - John Betts
Start date: 8/02/2012
16:00 - 17:30
Location: Dept. of Computer Science, Celestijnenlaan 200A, 00.225
22nd Simon Stevin Lecture on Optimization in Engineering
"Algorithmic Choices When Solving an Optimal Control Problem"
John Betts
Abstract
When designing a computational algorithm for solving an optimal control problem there are often many alternatives. Is it better to use an indirect or direct formulation? Is an interior point algorithm better than a sequential quadratic programming approach? This talk will discuss the many choices that must be made when constructing an algorithm that is robust, efficient, and accurate.
Biographical Information
John T. Betts received a B.A. degree from Grinnell College in 1965 with a major in physics and minor in mathematics. He attended graduate school at Purdue University and in 1967 received an M.S. in Astronautics with a major in orbit mechanics. He received a Ph.D. in aeronautical engineering from Purdue in 1970, specializing in optimal control theory. He joined The Aerospace Corporation in 1970 as a Member of the Technical Staff, and from 1977-1987 was manager of the Optimization Techniques Section of the Performance Analysis Department. He joined the Boeing Company, serving as manager of the Operations Research Group of Boeing Computer Services from 1987-1989. He served as a Technical Fellow in the Mathematics and Computing Technology Division, until his retirement in 2009, during which time he provided technical support to all areas of the Boeing Co. Dr. Betts is a member of AIAA and SIAM with active research in nonlinear programming and optimal control theory. In 2004, he was granted an "outstanding aerospace engineer award" by Purdue University and in 2011 was named a SIAM fellow. He has over 50 technical publications, and is the author of two books on optimal control methods.
**************
Thu 09 Feb 2012 15:00-16:00, Dept. of Computer Science, Celestijnenlaan 200A, 00.225
"Optimal Control of Partial Differential Equations with Delay "
John Betts
Abstract
The direct transcription method has proven valuable for solving optimal control problems described by ordinary differential equations. This presentation describes how the method can be extended to partial differential equations with delayed arguments. In addition to presenting results for this new class of applications, a number of generic algorithm performance issues are discussed.
**************
Thu 09 Feb 2012 16:00-17:00, Dept. of Computer Science, Celestijnenlaan 200A, 00.225
"Discretize Then Optimize"
John Betts
Abstract
Computational techniques for solving optimal control problems typically require combining a discretization technique with an optimization method. One possibility is to "Discretize Then Optimize", that is first discretize the differential equations and then apply an optimization algorithm to solve the resulting finite dimensional problem. Conversely one can "Optimize Then Discretize", that is write the continuous optimality conditions first and then discretize them. The goal of this paper is to compare the two alternatives and assess the relative merits.
**************
Please, do not reply to this message. The reply address is not monitored.