HGS MathComp Curriculum & Events

The course gives an introduction into concepts and algorithms of mixed-integer programming and combinatorial optimization.

Topics are linear programming and duality, polyhedral theory, postoptimal analysis, mixed-integer modelling, computation of optimal solutions, polyhedral combinatorics and combinatorial
polytopes, implementation of branch-and-cut algorithms, valid inequalities and cuts.

Two invited speakers will give a detailed presentation of two practical applications of mixed-integer models. In the afternoons of the first two days participants will learn how to model problems and solve them with commercial software.

Speakers:
Gerhard Reinelt, Heidelberg University
Elena Fernandez, Polytechnic University of Barcelona
Martine Labbe, Free University of Brussels

Time and locations:
Monday, February 10 2014:
09:00 - 12:00 OMZ, INF 350, Room U013
13:00 - 16:00 OMZ, INF 350, Room U012

Tuesday, February 11 2014:
09:00 - 12:00 OMZ, INF 350, Room U013
13:00 - 16:00 OMZ, INF 350, Room U012

Wednesday, February 12 2014:
09:00 - 12:00 OMZ, INF 350, Room U013

Thursday, February 13 2014:
09:00 - 12:00 OMZ, INF 350, Room U013

Objectives:
gain familiarity with some essential pattern recognition techniques. These are essential tools to address problems in prediction, tracking and exploratory data analysis.

Content:
mostly pattern recognition techniques that can be formulated as directed probabilistic graphical models:
- logistic regression
- Gaussian mixture models
- Hidden Markov models
- Kalman filter
- topic models
plus applications and inference schemes

Prerequisites:
basic notions of probability theory (joint distribution, marginal distribution). There will be a brief reminder, but it helps to have studied these beforehand.

Thursday, 28.11.2013

13:30-15:10 - "Basic principles of lasers and of spectroscopic signal detection."
Parts I-II (Prof. Dr. Rainer Suntz, KIT, Karlsruhe)

15:10-15:30 Coffee break

15:30-16:15 "Basic principles of lasers and of spectroscopic signal detection."
Part III (Prof. Dr. Rainer Suntz, KIT, Karlsruhe)


Friday, 29.11.2013

9.00-10.40 Laser diagnostic methods for the investigation of gas turbine combustion
Parts I-II (Dr. Wolfgang Meier, DLR Stuttgart)

10.40-11.00 Coffee break

11:00-12:00 Nanoparticles generated by Flame-Spray-Pyrolysis.
(Dr. Lydia Achelis, Universität Bremen)

12.00 End of the short course

Goals: The workshop will enable participants
(1) to choose the proper visual representation of data to illustrate or validate findings in publications,
(2) to perform data analysis tasks using established
visualization tools, and
(3) to design and implement new interactive visualizations.


Topics:
- dataset types, charts, and chart type selection
- avoiding common chart mistakes, graphical excellence
- models of visualization and interaction
- introduction to the programming language Processing
- developing interactive charts using Processing
- analyzing continuous spatial data using VTK/Paraview


Prerequisites: basic knowledge of computer programming concepts helpful

see web link

Description
In this key competence course we will utilize Osterwald_s Business Model canvas and Customer Value Canvas as strategic management and entrepreneurial tools. The process and tools will allow us to invent, describe, design and challenge our business models - from a vague idea to a blueprint fuelling a business plan.

Goals: You will learn about and apply a practical language to describe challenge and design business models. You will discover powerful examples for business model innovation. You will explore and prototype business models to boost future growth of your idea, project or solution.

Content:
- The common ground of profitable business models
- On finding a business model language
- A design thinking attitude
- testing and perpetuating business models.

The workshop discusses various aspects of the PhD training and education. It gives students an overview over planning a PhD, scientific writing and general organization of research work. Several well-known speakers and lecturers give first-hand insight and valuable information for a successful PhD phase.

Absolventen der Naturwissenschaften haben nicht nur das Vorstellungsgespräch im Bewerbungsprozess vor der Brust, sondern werden oftmals mit einem Assessment Center (AC) konfrontiert. Mit Hilfe von verschiedenen Übungen und Aufgaben werden die Fähigkeiten und Kompetenzen der AC-TeilnehmerInnen im Rahmen eines Bewerberauswahlverfahrens eingeschätzt. Ziel dieses Workshops ist es, den Teilnehmern sicheres und überzeugendes Auftreten sowohl im Vorstellungsgespräch als auch im AC beizubringen.

Inhalt
Das Training bietet nach einer sehr kurzen theoretischen Einführung zum Thema Assessment Center die Möglichkeit, einzelne Übungen zu durchlaufen. Einige Sequenzen werden für Feedbackrunden gefilmt, wie z.B. die persönliche Vorstellung vor Gruppen. Die TeilnehmerInnen lernen souveränes Auftreten vor Gruppen, in Drucksituationen gelassen zu bleiben und überzeugender zu argumentieren. Ziel ist es Körpersprache und Gesagtes zu harmonisieren. Sie trainieren u. a. Selbstpräsentation und Gruppendiskussionen, Rhetorik sowie wichtige Aufgaben aus Gruppenauswahlverfahren wie z.B. Rollenspiele. Ebenso werden Teile des Vorstellungsgespräches simuliert. Darüber hinaus erhalten die Teilnehmer Informationen zu den Erwartungen der Personalverantwortlichen.

In einem separaten Gespräch erhält jede(r) TeilnehmerIn ein ausführliches, individuelles Feedback, um sich verbessern zu können.

Bitte hier registrieren:

http://hgs.iwr.uni-heidelberg.de/Portfolio_HGS/VERANSTALTUNGEN/reg_form/reg_form.php?id=85

Target group:
Members of the university who have to talk a lot and are interested in improving their voice competences in communication.

Aims:
The workshop contains manifold exercises for a healthy and powerful voice. Learning how your own voice can be used as a strong communication and expression tool. As well as experiencing the influence of a physiological breathing and posture that affects your expressiveness during presentations and communication.

Contents of the seminar:
* Exercises that aim for the improvement of vocal sound, resonance, dynamics and articulation
* Individual voice analysis and consultation
* Practice of the acquired skills in lecture or talk
* Developing an improved perception of your own and foreign voices
* Facts about physiological aspects of the voice (for example process of voice production and health of the voice)

Goal and Content:
Knowledge of common numerical methods for solving problems in Linear Algebra and applications:
I. Linear systems and eigenvalue problems
II. Matrix decompositions
III. General iterative methods
IV. Krylow space methods
V. Iterative methods for eigenvalue problems
VI. Applications

Course language: English

I will present quantitative results for the stochastic
homogenization of linear elliptic equations with random coefficients in a discrete setting. Classical qualitative homogenization theory states that on large length scales the random coefficients can be replaced by homogenized coefficients that are deterministic and constant in space.
The homogenized coefficients are characterized by a formula that involves the solution to the so called \"corrector problem\". In contrast to periodic homogenization, in the stochastic setting the corrector problem is a highly degenerate elliptic equation on a probability space.
In this talk I will explain how to obtain various optimal estimates on the corrector, on approximations of the homogenized coefficients and on the homogenization error based on quantification of ergodicity. The talk is based on joint work with A. Gloria (Université Libre de Bruxelles) and F. Otto (MPI Leipzig).

The Roman poet Ovid (43 BC – AD 17/18) wrote eloquently that in life, \"Medio tutissimus ibis\": \"You will go safest in the middle.\" But where is the middle to be found, and how do we know when we have found it? These questions are difficult to answer if only because the search for the middle requires that we hold in check our basic human impulses.

In this inaugural lecture, I will elucidate the importance of searching for the middle in several contexts. I will provide examples of ways of proceeding cautiously through life, instances from the process of scientific discovery, and illustrations from research in mathematical statistics.

In the case of mathematical statistics, I will demonstrate connections between the search for the middle and many problems in the integral calculus that I have studied for the past 35 years.

DataFlow computers, compared to ControlFlow computers, offer speedups of 20 to 200 (even 2000 for some applications), power reductions of about 20, and size reductions of also about 20. However, the programming paradigm is different. This talk explains the paradigm, using Maxeler as an example (Maxeler is 20% owned by JPMorgan), and sheds light on the ongoing research in the field.

The Alexander von Humboldt Foundation is an organisation to promote research cooperation between scientists from abroad and from Germany. It offers a multitude of research fellowships and prizes for individual scientists starting at the post-doctoral level. Prominent examples are the Humboldt fellowship and the Feodor-Lynen fellowship. The former is awarded to scientists from other countries to conduct a research project in Germany for up to two years, while the latter is intended for young German scientists to carry out a research project in collaboration with a former Humboldt fellow at a research institute abroad.

This presentation will introduce the Alexander von Humboldt Foundation and its philosophy. An overview of funding opportunities at different stages of a scientific career will be given with particular focus on post-doctoral research.

We simulate slightly viscous turbulent low Mach number 3d bluff body flow (including streamlined bodies) by computational solution of the incompressible Navier-Stokes equations with a slip boundary condition modeling observed small skin friction, by using a residual stabilized adaptive finite element method, referred to as Direct Fem-Simulation since no turbulence model beyond automatic residual stabilization is used. We find by duality based a posteriori estimation that mean value quantities such as drag and lift are computable to accuracies comparable to experiments. As a key example, we show that the turbulent flow around a complete airplane is computable and inspecting solutions leads to a new theory of flight essentially different from the accepted theory by Kutta-Zhukovsky-Prandtl developed 100 years ago. We find that turbulent bluff body flow in general can be described as potential flow modified by rotational slip separation as a flow which is resolvable computationally using millions of mesh points, except in a far-field wake of little influence on lift and drag, and also is understandable through a mathematical stability analysis.

In contrast to Ordinary Differential Equations (ODEs), Delay Differential Equations (DDEs) allow for the inclusion of past actions into mathematical models, thus making the model closer to the real-world phenomenon. Most of the results on DDEs have been developed only in the last fifty years (Kuang, 1993), however there is not yet such a comprehensive theory, as for ODEs. In particular, big challenges are given by equations with state-dependent delay. For this kind of problems, indeed, there is nowadays no unified theory, but rather results worked out on the basis of particular classes of equations (Hartung et al., 2006). Applications of DDEs with state-dependent delay is a very modern topic in mathematics and might offer the chance for significant steps forward.
In this talk we will present several applications of delay differential equations with constant and state- dependent delay in the fields of population dynamics and mathematical epidemiology:
• In the context of population dynamics, DDEs with constant delay can be obtained, e.g., from the balance laws of age-structured population dynamics, assuming that birth rates and death rates, as functions of age, are piece-wise constant. The delay arises naturally from biology as the age-at- maturity of individuals. We shall see applications of this modeling approach in population dynamics of isolated populations (Barbarossa et al., 2013), interplay of predators and prey (Mohr et al., 2013), tumor modeling (Barbarossa et al., 2012), as well as for the control of mosquitoes by mean of sterile insect technique.
• In epidemiology a delay can represent for example the duration in time of incubation or time a host stays infected. However, delay equations can be also used to investigate the phenomenon of waning immunity. When the body gets infected by a virus, indeed, the immune system develops a certain resistance against it. As a matter of fact disease-induced immunity tends to wane and, long time after recovery, an individual might become again susceptible to the virus. Exposure to the pathogen boosts the immune system, thus prolonging the time in which the individual is immune (Heffernan et al., 2009). We shall model the feedback mechanism which makes possible for certain individuals to have lifelong immunity, being regularly exposed to the infection.

During the upcoming Fireside Chat we are going to discuss a hotly debated topic among young scientists - the challenge of combining career and family.

For this special occasion we have invited three renowned scientists who are going to share their personal experience with you!

Our guests will be:

Prof. Fred Hamprecht, Professor for Multidimensional Image Processing, Heidelberg Collaboratory for Image Processing (HCI)

Prof. Angelika Altmann-Dieses, Vice Dean Faculty of Management Science and Engineering, University of Applied Sciences Karlsruhe

Prof. Heike Leitte, Junior Professor Computer Graphics and Visualization (IWR)

http://hgs.iwr.uni-heidelberg.de/Portfolio_HGS/VERANSTALTUNGEN/reg_form/reg_form.php?id=91

Abstract:
This is a slightly unusual talk but we think the audience will enjoy the following three related presentations. The first speaker, Florian Jug, will start by showing some biological data in order to make the point that BioImaging is essential for gaining insights into underlying biological processes. It will be apparent that segmentation, matching, and tracking are of central importance for most questions to be answered. The second part of our talk, held by Dagmar Kainmüller, gives one concrete and detailed example of a recently developed matching procedure we call \"Active Graph Matching\". This method combines the well-established Active Shape Models and graph matching, more precisely with wide baseline matching via dual decomposition. We successfully applied Active Graph Matching for the task of simultaneous segmentation and annotation of nuclei in C. elegans larvae. In the last part of our presentation, Carsten Rother will briefly introduce the Computer Vision Lab Dresden. He will also present some initial results on projects which run in his group, which are mainly about correspondence problems in RGBD images and on detection and pose estimation of chairs and other objects (joint work with Oxford University)

Discussion of topical challenges to mathematical modeling and simulation of the cardiovascular system with focus on the following subjects:

Modeling, Analysis
- Biomechanical interaction of blood flow with the vessel wall and tissue reactions and transport processes in blood flow and the vessel wall
- Mechanics of the vessel walls including information on microstructures
- Deposition and chemical processes and their influence on the vessel wall
- Thrombus and plaque formation
- Vessel occlusion and infarcts
- Methods and techniques for calibration and validation based on medical data

Numerical methods and simulation:
- Methods for long time processes
- Simulations of fluid-structure interaction and related free boundary problems
- Methods for processes with large deformations
- Report and discussion of simulations results and their medical implications