HGS MathComp Curriculum & Events

Registration required

This course consists of three parts. The first part provides a brief introduction to the subject of probability theory, reaching the intermediate graduate level in the subject; the prerequisite for this part of the course entails only a good background in calculus.

The second part of the course will cover applications of probability theory to the analysis of certain financial derivatives and structured investment products that have played central roles in the extreme behavior of the financial markets since 2000.

In the last part of the course, the class will analyze the prospectuses of some financial derivatives and structured products. Students will be carry out a real-time analysis of these financial instruments. Thus, the overall goal of the course is to instruct students in the importance of basic probability theory for the everyday analysis of novel financial
instruments.

Acquire knowledge of image processing and computer vision approaches for robotic applications

Linear model predictive control (LMPC) should be a special case of nonlinear model predictive control (NMPC), but the formulation is rather different. This is a tutorial presentation of the usual mathematics associated with LMPC, starting with differential equations fed by a zero order hold, conversion to discrete time, quadratic cost optimal control properties with penalty functions, conversion of standard difference equation model to the LMPC form, design of the ILMC control law, and some pitfalls and difficulties with the approach.

Scope of the conference
The construction of humanoid and bipedal robots based on humans and legged animals, is a complex, interdisciplinary and rapidly evolving research field. There are diverse scientific challenges in designing a robot that can walk, run, jump, and move in a robust manner similar to its biological counterpart. Technical applications of such robots are expected in the near future in fields ranging from service robotics (households, public or industrial environments, hazardous sites) to medical robotics (neuro-prosthetics, exo-skeletons etc).
This conference aims to bring together leading researchers from France, Germany and Japan to hold technical talks and discussions on the topics related to Humanoid and Legged Robots.

List of the conference topics:
• Locomotion of humanoid robots on varying terrains
• Manipulation and grasping for humanoid robots
• Cognitive skills for humanoid robots
• Walking models (biological & human) for robots
• Stability analysis and Robustness of robot motion
• Efficient control algorithms for robot motions
• Compliance for walking robots and humanoid robots
• Modeling in robotics - model reduction/simplification
• Simulation- and optimization methods for walking and humanoid robots
• New technologies for actuation and sensor-systems

HLR 2014 is the follow-up of two successful French-German HLR Conferences in Karlsruhe in 2006 and in Paris in 2011

HLR 2014 is divided in two parts:
• A public inaugural session on May 12 taking place at the Alte Aula
• A workshop on May 13 & 14 for invited participants, taking place at IWH

Simulation software for complex phenomena based on models involving partial differential equations (PDE) has become an important topic in modern research from mathematics and scientific computing providing methods and algorithms to application fields utilizing codes and defining requirements on their technical abilities. Powerful, but also complex mathematical algorithms and advanced hardware architectures have widened the span of practically computable problems considerably. But this power does not come for free and is typically related to complex implementations which have to be developed over many years. The development of powerful, academic, multi-purpose libraries for PDE simulation began about 1990. Nowadays, there are several widespread codes which are well developed and well maintained. These codeas have different characteristics, making them better suited for different problems.

PDESoft conferences provide a discussion venue for developers and users of such libraries as well as researchers who study implementation of computer algorithms for PDE. They serve as a focus point, where the state-of-the-art implementation of state-of-the-art algorithms is discussed. They give developers a chance to share ideas with developers of other packages and users. And users of such sofware have an opportunity to learn about the development principles of software and contribute their knowledge from the point of view of application.

The PDESoft conferences started in 2012 at Universität Münster. After the conference, a considerable number of participants gave positive feedback and it was agreed, that a series of conferences on the topic of PDE-Software was needed. The conference this year is the second in this series.

The 2014 conference will be hosted jointly by the Heidelberg Institute for Theoretical Studies (HITS) and Heidelberg University_s Interdisciplinary Center for Scientific Computing (IWR). The presentations and discussions will take place at the Studio Villa Bosch in Heidelberg, Germany.

The conference is followed by coding days (July 17/18) at the IWR. Two days of joint coding. We will set up infrastructure (power and ethernet) in a single room such that participants can discuss while writing code on their laptops and other devices.

Historians and archaeologists don_t look any more at images according to the postulates of the traditional iconography. They highlight the need to take image as an indissociable whole and not to separate it from the materiality of its medium: images have to be considered in their physicality and it is as such – that is, as physical objects being acted and acting in specifics locations –that their functionalities could be better understood. In addition to remind some of the well - known reasons
to work on 3D reconstructions of archaeological heritage, the present paper will illustrate how such reconstructions can contribute to better understanding of the \"agency\" of images, taking the example of one the most extraordinary sculpted group of Ancient Cambodia: the five heavily damaged statues of the dance of Śiva at Prasathom in Koh Ker.

Den TeilnehmerInnen werden neben den grundlegenden Kenntnissen der Versicherungs-
und Finanzwirtschaft vor allem die 10 goldenen Regeln vermittelt, welche den Grundstein eines erfolgreichen Vermögensaufbaus darstellen.


Bitte hier registrieren:
http://hgs.iwr.uni-heidelberg.de/Portfolio_HGS/VERANSTALTUNGEN/reg_form/reg_form.php?id=96

/ Description:
Usually, the doctoral thesis is the first large academic writing project a doctoral student has to complete. Mastering it successfully requires high levels of motivation as well as strategies for the effective organization of work and daily life.
In this course, we will treat the dissertation as a personal project that can be planned and carried out using special techniques.

We will focus on the following topics:
- Identifying the phases of a project and their core tasks: situation analysis, goal definition, project structure plan and project action plan
- Work-Life-Balance: which goals do I want to spend my time on?
- Weekly planning and prioritising: how do I distribute my tasks over the week and avoid time pressure?
- Analysing personal strengths and weaknesses: what do I need to improve and how can I plan my development?
- Communicating effectively with my supervisor: how do I plan and carry out goal-oriented dialogs?
- Creating an academic network

In the course you will apply the techniques learned by planning your individual doctoral project and evaluate your working behaviour.

/ Description:
Writing grant proposals is a demanding task because evaluation committees expect clear and concise texts. In addition, grant proposals differ in structure and style from other forms of academic writing.

In this workshop, we answer three questions:
- What are characteristics of high-quality grant proposals?
- What are appropriate text structures for proposals?
- Which strategies help to accelerate the preparation of proposals?

Programme:
- Functions of grant proposals: planning, communication, persuasion, contract
- Refining research ideas, choosing and narrowing topics
- Section structure of proposals: problem statement, aims and objectives, literature review, methodology, timetable, budget, etc.
- Differences between proposal writing and paper writing
- Characteristics of high-quality proposals: clarity, correctness, consistency, conciseness, confidence

Activities and time duration:
The workshop includes lectures and exercises. In the exercises, participants analyse sample proposals and draft components of their own proposals.
Half day-workshop: 9 am to 12.30 am or 1.30 pm to 5 pm

Preparation:
- Please think of a research question you want to tackle in your next research project. Describe the problem your research is going to solve (a few sentences are fine). Please bring your notes to the workshop.
- Please ask your supervisor or your colleague for a well-prepared grant proposal. Note the section structure of the proposal and bring this information to the workshop.

The lecture gives an introduction to the theory and numerics of optimization problems in which partial differential equations (PDEs) occur as constraints. The following topics are covered:
- Parameter estimation: estimation of parameters in elliptic and parabolic PDEs;
- Ill-posed inverse problems: estimation of distributed parameters in elliptic PDEs;
- Optimal experimental design with PDEs: estimation and optimization of the covariance of the parameters;
- Optimal control of PDEs.

Computer simulation is everywhere in our daily life, be it the designing of novel materials or weather and climate prediction. Simulation is the new heart of science and engineering. It has largely replaced experimentation and prototyping and will play an increasing role in the future. It also has made inroads into new areas of growth, such as the social sciences, medicine and biology.

What all these complicated processes have in common is the vast array of spatial and temporal scales that they span. Brute force computation is impossible even on the biggest supercomputer, and so mathematics plays a central role to reduce the behaviour to simpler, computable models that still retain predictive power on the scales of interest. By looking at various examples, drawn from my research, I will highlight the progress and the challenges for modern scientific computing in simulating our multiscale world, in particular calling attention to the increasingly important field of uncertainty quantification.

/ Description:

We kindly invite all members, students and staff of the HGS MathComp to join us at our Summer Party 2014.

As usual, in accordance with university guidelines, we have to charge € 10 per person to cover expenses and of course children are free.

Please make sure to register online for the event. (see link below)

Part-structured models were first developed for the task of photographic object recognition, but recently they have been used to solve problems in document analysis. The models allow incorporation of spatial constraints between component parts using springlike potentials, and inference on them can be efficiently performed through a form of dynamic programming. This talk will describe the use of one particular form of part-structured model, the inkball model, and examine its application to word spotting, character detection, word recognition, and writing style comparison.

The KoroiBot Summer School will be held in Heidelberg from September 22nd to September 26th 2014 at the Interdisciplinary Center for Scientific Computing at the University of Heidelberg. The Summer School will consist of lectures and hands-on tutorials designed to give students at MSc and PhD level an insight on different approaches used for analysis and generation of humanoid motions. A poster session provides the opportunity for the students to present their ongoing work.

Program
Topics will include:
- Acquisition, Representation and Mapping of Human Motion to Humanoid Motion
- Humanoid motion generation within the task-space formalism
- Dynamic modeling of anthropomorphic systems
- Offline and online optimal control methods for human and humanoid motions
- Inverse optimal control for the analysis of human movement
- Bayesian model comparison for movement primitive selection
- Whole-body torque estimation / whole-body feedback linearization.

Random projection is widely used as a method of dimension reduction. In recent years, its combination with standard techniques of regression and classification has been explored. In this talk we present a method for its use in anomaly detection in high-dimensional settings, in conjunction with principal component analysis (PCA) and corresponding subspace detection methods. We assume a so-called spiked covariance model for the underlying data generation process and a Gaussian random projection. We adopt a hypothesis testing perspective of the anomaly detection problem, with the test statistic defined to be the magnitude of the residuals of a PCA analysis. Under the null hypothesis of no anomaly, we characterize the relative accuracy with which the mean and variance of the test statistic from compressed data approximate those of the corresponding test statistic from uncompressed data. Furthermore, under a suitable alternative hypothesis, we provide expressions that allow for a comparison of statistical power for detection. Finally, whereas these results correspond to the ideal setting in which the data covariance is known, we show that it is possible to obtain the same order of accuracy when the covariance of the compressed measurements is estimated using a sample covariance, as long as the number of measurements is of the same order of magnitude as the reduced dimensionality. We illustrate the practical impact of our results in the context of predicting volume anomalies in Internet traffic data.

Parts of this talk is a joint work with Ahmed H. Sameh (Purdue University), Faisal Saied (NYU-AD), Georgios Kollias (IBM Thomas J. Watson Research Center), Ercan Bolukbasi (METU) and Omer Tari
(METU).

Solution of sparse linear systems in parallel computing platforms
requires two main steps which are critical: (1) Reordering the sparse
matrix and (2) direct or iterative algorithms for solving the
reordered system of equations. In this talk we will talk about novel
algorithms that improve the parallel scalability in both steps.

In the first part of this talk, we will present a new multithreaded
and recursive variation of the DS factorization based parallel direct
sparse solver. We show the improvement compared to the sparse LU
factorization based multithreaded Pardiso direct solver on a shared
memory architecture.

In the second part, we will present recent results on obtaining the
Fiedler vector and the permutation induced by the Fiedler vector
effectively on a parallel computing platform. We will show a
significant parallel improvement using our algorithm compared to a
highly effective sequential counterpart in the Harwell Subroutine
Library. We will also apply this permutation to extract a banded
preconditioner and solve the linear systems of equations in parallel
using PSPIKE scheme.
If time permits, we will also present a new reordering scheme based on
the largest eigenvector of the graph Laplacian. From any given sparse
matrix, the new reordering scheme can create a 2x2 block form with
favorable properties.

Wir betrachten Optimierungsprobleme, in denen partielle
Differentialgleichungen als Nebenbedingungen auftreten. In den letzten
Jahren wurden erhebliche Fortschritte bei der Entwicklung von
Multilevel-Optimierungsverfahren gemacht, die adaptive
Diskretisierungstechniken, reduzierte Modelle und a
posteriori-Fehlerschätzer in effizienter Weise verbinden. Diese
Verfahren eröffnen das Potential, Optimierungsiterationen auf Basis
relativ billiger Simulationsmodelle zu berechnen und das
Optimierungsproblemen im Aufwand weniger Simulationen der
zugerundeliegenden partiellen Differentialgleichung zu lösen.
Wir diskutieren einen Multilevel-Ansatz, der eine Hierarchie adaptiver
Diskretisierungen während der Optimierungsiteration auf Basis
adaptiver FE-Diskretisierungen und reduzierter Modelle wie POD
erzeugt. Die adaptive Verfeinerungsstrategie beruht auf a
posteriori-Fehlerschätzern für die zugrundeliegende Zustandsgleichung,
adjungierte Gleichung und das Stationaritätsmaß. Der resultierende
Algorithmus kann bestehende adaptive PDE-Löser und Fehlerschätzer
modular einsetzen. Neben der Behandlung von Steuerungsnebenbedingungen
gehen wir auch auf die Behandlung von Zustandsschranken ein. Wir
demonstrieren die Effizienz des Ansatzes
anhand von numerischen Ergebnissen.

Historians and archaeologists don_t look any more at images according to the postulates of the traditional iconography. They highlight the need to take image as an indissociable whole and not to separate it from the materiality of its medium: images have to be considered in their physicality and it is as such – that is, as physical objects being acted and acting in specifics locations –that their functionalities could be better understood. In addition to remind some of the well - known reasons
to work on 3D reconstructions of archaeological heritage, the present paper will illustrate how such reconstructions can contribute to better understanding of the \\\"agency\\\" of images, taking the example of one the most extraordinary sculpted group of Ancient Cambodia: the five heavily damaged statues of the dance of Śiva at Prasathom in Koh Ker.

Maintaining a good model of a controlled plant is an important challenge
in the process industries, model quality being one determining factor
for the performance of a model predictive controller (MPC). Feldbaum was
the first to recognize that an optimal controller for a system with
unknown parameters has two conflicting tasks: directing the output
toward a reference, and exciting the system for learning purposes so
that better control decisions can be made in the future. A dual
controller is optimal in the sense that it finds the best trade-off or
balance between control and excitation.

In this talk we will look at three variations of an MPC based approach
to the problem of controlling a system with unknown model parameters
while simultaneously exciting the system so that the parameter estimates
can be improved and their error variances reduced. Excitation is
explicitly included in the cost functions and balanced against standard
control objectives.

All three algorithms can be implemented with minor modifications to an
existing MPC framework. We look numerical results for each of the three
approaches and compare some advantages and disadvantages. The results
show that parameter identification can be greatly enhanced with only a
small effect on the system output by allowing the controllers to
slightly increase the system excitation in a smart way, and that this
can be achieved without requiring persistent excitation.

The last part of the presentation contains some promising preliminary
experimental results.

Advances in combustion science and technology depend ever more critically on our ability to formulate descriptions that enable predictive simulations for fundamental insights and for the design of technical combustion systems. Experimentally discovering and characterizing new insights into the physics and chemistry of combustion is foundational in that context. Over the past decade, advances in instrumentation have enabled measurements to image mixing, ignition, and flame evolution at relevant real-time scales. New laser sources, digital cameras, and image processing capabilities have lead to a rapid growth in capability and widespread use of high-speed imaging diagnostics in combustion research. Now, the extensions of these capabilities are being pursued to enable experiments that capture the 3D nature of combustion in single-shot measurements.


This presentation describes examples of 2D and 3D high-speed imaging techniques and their application to study spark-ignited direct-injected internal combustion engines to understand the causes of combustion instabilities that are inherent to these engines.

Abstract:
A new class of spanning trees, called Majority Spanning trees, and Majority Cotrees have been defined and their existence in digraphs, with non-negative weights
on edges, have been established. The applications of Majority Spanning Trees to Minimum Connection Time Problem, Optimal Ranking of a Round Robin Tour- nament and settling debts have been shown.

Hinweis für Teilnehmer:
For this lunch-time event we will provide desserts for every attendee. We cordially invite all members of HGS MathComp and IWR to join us!

Computational modeling and simulation of human cognition is an emergent multidisciplinary field of research promising breakthroughs for both basic research and applications. There is already a wide range of notable examples illustrating the successful application of computational modeling in cognition from models of working memory to intelligent tutoring systems. However, many psychiatric and neurological disorders involve characteristic cognitive deficits that still are poorly understood on a process level.
Quantitative cognitive process models promise to contribute towards filling the explanatory gap between models of basic neural mechanisms and high-level cognitive abilities. Mathematical methods and computational tools for process modeling, which have a long success story in other disciplines, e.g. chemical engineering, may help to address the challenges and potential pitfalls arising in this approach. A successful implementation, however, requires a broad range of skills and knowledge.
Therefore, in this challenge workshop, we bring together experts from the different fields of cognitive modeling, cognitive disorders, and process modeling. The aim is to facilitate the exchange about research questions in domains like psychiatry, neuroscience, or cognitive science, which may be addressed by methods of process modeling. We will discuss the available methodological approaches and tools as well as their limitations in order to foster effective future collaborations.
The focus of this workshop will be on higher cognition and its disorders, including process models of the underlying neural activity. A methodological focus will be on effective parameter estimation methods, model validation procedures, model-based optimum experimental design, and the application of these methods to models of cognition.

Target Audience
Researchers interested in quantitative models of cognition, particularly higher cognition and models of dysfunctional cognition. Relevant fields include psychiatry, geriatrics, psychology, cognitive science, neuroscience, mathematics, economics, computer science, statistics, and related fields. The workshop is also open to doctoral and graduate students.

Example topics
- Studies on process-related aspects of disordered
cognition
- Cognitive models of planning and problem-solving
in psychiatric disorders
- Tracing cognitive activity in real-time using model-
supported EEG/fMRT
- Process-interpretation of fMRT connectivity
analyses
- Modeling age-related decline in cognitive abilities
- Modeling of cognitive effects of neurotransmitter
dysbalances and psychopharmacological
interventions
- Axiomatic analyses of modeling frameworks
- Tractable mathematical formulations of cognitive
models
- Parameter estimation
- Model discrimination and validation
- Model-based optimum experimental design

Every academic culture in general as well as individual disciplinary cultures in particular are characterised by shared basic assumptions and practices that are acquired during the course of study and are handed down and upheld in the practical context at universities and in professional work. It is easy to overlook the fact that basic assumptions and practices of this type are also moulded by cultural differences that not only enrich academic exchange at the international level, but sometimes also impede it. For example, problems can arise at the social level, i.e. with regard to interaction, communication and co-operation, or can be related to the level of academic traditions, i.e. structures of thought and argument.
The goal of the workshop is to make the inherent basic assumptions, rules and expectations of the German academic culture comprehensible to international doctoral candidates and to strengthen their adaptation capabilities. One of the essential objectives of the workshop is to facilitate entry into the foreign culture for the participants in order to reduce stress, misunderstandings and uncertainty and thus to ultimately contribute towards optimal utilisation of personal, time-related and financial resources.

Main focus areas:
- The cultural framework (individual and cultural patterns)
- Underlying structures (comparison of the implicit rules, values and convictions of the German academic system)
- Action-oriented skills 1 (social interaction with colleagues and supervisors)
- Action-oriented skills 2 (academic argumentation)

>Registration required<
Please use the link below to register for the workshop:

The junior research group “From the Caribbean to North-America and Back” (Transcultural Studies, University of Heidelberg) and the Heidelberg Graduate School of Mathematical and Computational Methods for the Sciences are organizing the interdisciplinary workshop “Under construction. Analyzing postcolonial weblogs with literary and computational methods.” The workshop makes academic approaches to personal weblogs accessible which include the very divergent areas of postcolonial literary studies, computational and engineering sciences. The main object of the event is to mediate a dialogue between the various disciplines. For this purpose, three scientists will give us insights into their research: Prof. Dr. Dagmar Monett Díaz from the Berlin School of Economics and Law, who will talk about the Cuban Blogosphere and its census; Andreas Schieber from the Technical University of Dresden, who will discuss the method of content extraction in weblogs; and Chantal Marquardt from the University of Mannheim, who will elaborate on literary approaches to Francophone North African weblogs.