IPNet Digest Volume 19, Number 03 March 31, 2012 Today's Editor: Patricia K. Lamm, Michigan State University Today's Topics: Workshop: 100 years of Electrical Imaging Conference: Inverse Problems and Applications Workshop Update: Optimization & Inverse Problems in Electromagnetism Conference Update: Inverse Problems Symposium PhD Positions: High-Definition Tomography New Book on Inverse Problems Submissions for IPNet Digest: Mail to ipnet-digest@math.msu.edu Information about IPNet: http://www.math.msu.edu/ipnet ----------------------------- Subject: 100 years of Electrical Imaging From: Bill Lionheart Date: 3/6/2012 100 years of electrical imaging In 1912, Conrad Schlumberger made the first electric field imaging experiment at his family house in Normandy. Over the last 100 years, electrical imaging has grown to be used in many other fields such as medical and process tomography To celebrate EIT's 100th birthday, we are organizing a workshop: 9-10 July 2012 in Paris. We aim to bring together the disparate electrical imaging communities (geophysical, medical and industrial process imaging, as well as other specialist applications), and encourage interactions and knowledge transfer between the communities on image analysis and algorithm techniques. The workshop will consist of keynote speakers in various applications of electrical imaging, as well as poster presentations of new research and applications. Please see the announcement: http://100electrical.geosciences.mines-paristech.fr/first_announcement and the workshop web site: http://100electrical.geosciences.mines-paristech.fr ----------------------------- Subject: New: Conference Inverse Problems and Applications (IPA) 2-6 April, 2013, Linköping University, Sweden From: George Baravdish Date: 3/13/2012 Dear Colleagues, We gladly announce the conference Inverse Problems and Applications (IPA2013) 2-6 April, 2013, Linköping University, Sweden http://www.mai.liu.se/IPA2013 The conference is a collaboration between Linköping University and Institut Mittag-Leffler. The aim of the conference is to present the state of the art in Inverse Problems and to foster greater exchange of experience and knowledge of applying inverse problem in different areas of applications. Main topics: Inverse Problems: Theory, Algorithms and Applications; Identification in Partial Differential Equations; Inverse Scattering; Computational Methods; Regularization Techniques; Inverse problems with small parameters; Imaging Techniques. The following invited speakers will give a plenary lecture: S. Arridge (University College London, UK) G. Bal (Columbia University, USA) V. Isakov (Wichita State University, USA) M. Lassas (University of Helsinki, Finland) D. Lesnic (University of Leeds, UK) L. Ljung (Linköping University, Sweden) Z. Nashed (University of Central Florida, USA) M. Salo (University of Helsinki, Finland) O. Scherzer (University of Vienna, Austria) G. Uhlmann (University of Washington, USA) Important deadlines: Registration: April 10 2012 -- Mars 10 2013 Abstract submission: April 10 2012 -- November 5 2012. Some financial support will be available for PhD students. Organizing committee: George Baravdish (Linköping University), Vladimir Kozlov (Linköping University), Yaroslav Kurylev (University College London, UK), Lassi Päivärinta (University of Helsinki, Finland)) and Luba Kulesh (Linköping University). On the behalf of the organizing committee, George Baravdish, George.baravdish@liu.se and Vladimir Kozlov, vladimir.kozlov@liu.se ----------------------------- Subject: 2nd call for papers OIPE 2012' From: Date: 3/14/2012 Dear colleagues, We cordially invite you to participate to the 12th Workshop on Optimization and Inverse Problems in Electromagnetism, OIPE 2012, to be held on September 19th-21st, 2012, in Ghent, Belgium. Prior to the workshop, several PhD courses on the topic will be organized on September 17th-18th, 2012. We invite members of the scientific community in universities, research centers and industry to attend the workshop and present their recent achievements. Abstract submission deadline will be May 14th, 2012. More information about the workshop and PhD course can be found on http://www.oipe2012.com/. We look forward to meet all of you in Ghent at OIPE 2012. Prof. dr. ir. Luc Dupre, Chairman dr. ir. Guillaume Crevecoeur, Conference secretary ----------------------------- Subject: IPS 2012 From: "Dolan, Kirk" Date: 3/30/2012 Dear Inverse Problems Researchers and Instructors, Registration is now open for 2012 Inverse Problems Symposium, June 10-12 at Michigan State University Kellogg Center. Early registration ends May 5th. Abstract and poster submission ends April 18th. The 2012 symposium in East Lansing, Michigan, will retain the single session format of these symposia, and will have sessions addressing both the theoretical and applied aspects of inverse problems. Please circulate this announcement to interested colleagues. Agenda Sunday June 10 15:30-17:30: James Beck, tutorial on the inverse heat conduction problem (IHCP) Evening: Informal dinner on our own Monday, June 11 7:45-8:30: Registration 8:00: Continental breakfast 8:30: Welcome 8:40 -9:25: Keynote address: Dr. Daniel Inman, Department Chair and Clarence "Kelly" Johnson Professor, Aerospace Engineering, University of Michigan 9:25-17:00: Oral and poster presentations, Lunch provided 19:00: Symposium Banquet Satish Udpa, University Distinguished Professor, Dean of MSU College of Engineering, speaker Tuesday, June 12 8:00: Continental breakfast 8:30 -9:15: Keynote address: Dr. Jay Frankel, Professor, Mechanical, Aerospace and Biomedical Engineering, University of Tennesee 9:25-17:00: Oral presentations, Lunch provided 17:00: Finish The early registration fee $150/$100 regular/student covers Monday/Tue continental breakfast, lunch, breaks, Monday banquet, and CD. We are interested in a wide range of topics in engineering, agriculture, natural sciences, mathematics, statistics, etc. A written paper is not required and the papers will not be subject to copyright. The website is: www.inverseproblems2012.org Kirk Dolan, Conference Chairman Associate Professor Department of Food Science & Human Nutrition Department of Biosystems & Agricultural Engineering 135 Trout Food Science Building Michigan State University ----------------------------- Subject: Two PhD Positions, Technical University of Denmark From: Kim Knudsen Date: 3/14/2012 Two PhD positions are available at DTU Informatics, Technical University of Denmark, starting August 1, 2012. Both are funded by the ERC project "High-Definition Tomography" headed by Professor Per Christian Hansen. The goal of this project is to develop the enabling mathematical technology and computational algorithms to produce a new generation of tomographic reconstruction methods that can incorporate many different kinds of available prior information in order to produce high-definition reconstructions, i.e., sharper images with more reliable details. Project 1: Statistical Priors in Variational Reconstruction Methods This PhD project aims at bridging the gap between Bayesian and variational reconstruction methods by setting up a mathematical and computational framework that utilizes the connections between parametric priors for the solutions, in the form of probability distributions, and corresponding norms or filters. See http://www.dtu.dk/English/About_DTU/vacancies.aspx?guid=d093d5ed-bfec-4bc6-ae39-86dd803a00e3 Project 2: Training Sets in Large-Scale Reconstruction Methods This PhD project aims at providing a theoretical and methodological framework for the use of training sets as non-parametric priors for the solutions in tomographic reconstruction. We will study and further develop the use of artificially generated training sets for incorporation of structural information about the solution. See http://www.dtu.dk/English/About_DTU/vacancies.aspx?guid=7f4ce369-e70f-4f4f-803a-c77b0d46fe68 For more information, see the home pages for the two projects. Applications must be written in English and submitted online via the home pages by April 22, 2012. More information can be obtained from Professor Per Christian Hansen, DTU Informatics Email pch@imm.dtu.dk Tel +45 45.25.30.97 Submitted by: Kim Knudsen, Lektor, DTU Matematik Danmarks Tekniske Universitet http://www.dtu.dk/images/DTU_email_logo_01.gif Institut for Matematik Matematiktorvet Bygning 303 S 2800 Kgs. Lyngby Direkte telefon 45253026 k.knudsen@mat.dtu.dk www.mat.dtu.dk/ ----------------------------- Subject: a new book on inverse problems From: "Klibanov, Michael" Date: 3/20/2012 A new book on Inverse Problems Title: Approximate Global Convergence and Adaptivity for Coefficient Inverse Problems Authors: Larisa Beilina and Michael V. Klibanov Publisher: Springer, New York, 2012 Hardcover, ISBN 978-1-4419-7804-2 http://www.springer.com/mathematics/dynamical+systems/book/978-1-4419-7804-2 Chapters: Two Central Questions of This Book and an Introduction to the Theories of Ill-Posed and Coefficient Inverse Problems Approximately Globally Convergent Numerical Method Numerical Implementation of the Approximately Globally Convergent Numerical Method The Adaptive Finite Element Technique and its Synthesis with the Approximately Globally Convergent Numerical Method Blind Experimental Data Backscattering Data The field of Inverse Problems is an applied one. Because of many applications, it is important to develop non-local numerical methods for Coefficient Inverse Problems (CIPs). On the other hand, the goal of the development of such methods is a very challenging one because of both nonlinearity and ill-posedness of CIPs. Both the most important and the most difficult question in this regard is about obtaining a good approximation for the unknown coefficient without any advanced knowledge of a small neighborhood of this coefficient. This is the case of many applications. And this is what “non-local” means. This is the first book in which an answer to the above question can be found. Two new concepts of numerical solutions of multidimensional CIPs for a hyperbolic PDE are presented here: Approximate Global Convergence and the Adaptive Finite Element Method (adaptivity for brevity). Chapter 1 might be used as an introductory graduate course to the theories of Ill-Posed and Coefficient Inverse Problems. In the beginning of this chapter two central questions of this book are presented (see below). The major part of this chapter is devoted to an introduction in the theory of Ill-Posed Problems. In addition, some uniqueness theorems for CIPs with single measurement data are proved in this chapter via the method of Carleman estimates. T The book combines a detailed convergence analysis with recipes for various numerical implementations of developed algorithms. A reader who would want to focus on numerical implementations, might skip reading proofs of convergence theorems. Many numerical examples are presented. It is important that all numerical results are in a good agreement with the convergence analysis. Only CIPs with single measurement data are considered in this book. "Single measurement" means that the data are generated by either a single position of the point source or a single direction of the incident plane wave. This is the most economical way of data collection, which is preferable in many applications. For example, in military applications one wants to minimize both the number of measurements and the measurement angle: because of dangers on the battlefield. The single measurement case means the minimal amount of the available information. Therefore, this case is the most challenging one to handle. The numerical technique of this book is applied to two types of experimental data. The most challenging case of blind data is considered in both cases. A clear advantage of the blind data case is that it is unbiased. The first type of experimental data is collected in a controlled laboratory environment (Chapter 5). The second type (section 6.9) is collected in the field in a cluttered environment by the Forward Looking Radar of US Army Research Laboratory. The result for the second type addresses a real World problem. This is the problem of imaging of dielectric constants in shallow explosive-like targets using the data collected by the above radar. In both types of experimental data huge discrepancies between the measured data and computationally simulated ones are evident. This is the main challenge here. Those discrepancies are addressed via two new data pre-processing procedures. The noise in the resulting pre-processed data includes both: the natural measurement noise and the modeling noise. Therefore, the noise level is unknown and is very large. Nevertheless, accurate imaging results are obtained in blind data cases by the approximately globally convergent method of this book. This points towards a high stability of algorithms of this book. In fact, the stability observed in the studies of experimental data is better than the one predicted by the convergence analysis. This is because any convergence analysis assumes that the noise in the data is small, whereas it is large in our case. In 2008-2011 the authors have developed a new technique, which has addressed the above question for n-dimensional (n=2,3) CIPs for an important hyperbolic PDE. The book addresses the following two central questions for the above CIPs: Question 1. How to obtain a good approximation for the exact solution without any knowledge of a small neighborhood of this solution? Question 2. Given that approximation, how to refine it? It is well known that the first question is an enormously challenging one. Therefore, one inevitably faces a tough dilemma: either ignore this question, or try to address it in the expense of using some approximate mathematical models. The authors have chosen the second option. These models amount to the truncation of a certain asymptotic series. In the case of second and third approximate models, this truncation is done only on the first iteration, and all subsequent iterations do not use it. As a result, the authors have introduced a new term "approximate global convergence" for corresponding numerical methods addressing Question 1. Justification of those approximate models is done via numerical studies of computationally simulated data and, most persuasively, of blind experimental data. Question 2 is addressed via the adaptivity. The adaptivity is likely the best technique for refinements of images obtained on the first stage. Thus, a natural two-stage numerical procedure is developed. On the first stage the approximately globally convergent method provides a guaranteed good approximation for the exact solution. And on the second stage the locally convergent adaptivity technique refines that approximation. Rigorous convergence analysis is conducted for both stages. Some new analytical results for the adaptivity for CIPs are presented. Michael V. Klibanov ------- end -------