HGS MathComp Curriculum & Events

Computational modeling has become more and more important in the life sciences. One important class of models are compartment models and ordinary differential equations are widely used to describe the time evolution of the model_s components in a deterministic way. This compact course will introduce stochastic compartmental modeling. The objectives of the course are to learn a) methods that allow to simulate stochastic models and b) which effects so called intrinsic stochasticity can have on systems_ dynamics. These effects will make it evident that specific calibration techniques are needed in order to be able to cope with stochastic effects and exploit their information. The course will c) give a flavor of how calibration can be performed. Time will also be devoted to let the participants learn d) when stochastic modeling is necessary and beneficial.
This course will consist of lectures as well as practical exercises. Therefore, participants are encouraged to bring laptops (please contact me in case laptop sharing is desired). There is no prior software or programming experience necessary.

Please register here

Virtual Screening has become an essential element of the drug
discovery process. Virtual screening is used to search a large
library of small molecules for binding to a target protein and
select a small subset of compounds for subsequent experimental
validation and optimization. In this block course we will
discuss the methodological basis and practical applications of
structure-based and ligand-based virtual screening methods such as
docking, shape-based, pharmacophore and fingerprint concepts.
The course alternates lectures and practical sections in the
computer lab

Please register here

I will present a survey of the main results about first and second order models of swarming where repulsion and attraction are modeled through pairwise potentials. We will mainly focus on the stability of the fascinating patterns that you get by random particle simulations, flocks and mills, and their qualitative behavior. Qualitative properties of local minimizers of the interaction energies are crucial in order to understand these complex behaviors. Compactly supported global minimizers determine the flock patterns whose existence is related to the classical H-stability in statistical mechanics and the classical obstacle problem for differential operators.

The Center for Digital Research in the Humanities (CDRH) at the University of Nebraska-Lincoln (UNL) - founded in 2005 - was one of the earliest Digital Humanities (DH) centers in the world, and is supported by the University as a Center of Excellence as well as by private funds and grants. The Center is a founding member of centerNet, an international network of digital humanities centers, and is an institutional member of the TEI Consortium, the National Humanities Alliance, and the National Humanities Alliance. While the Center’s roots are in Library Science and English, the mission of CDRH is to promote collaborative and transdisciplinary digital humanities research. The Center houses over fifty scholarly projects ranging in scale, topic, and purpose. In the first part of the talk, I will present an overview of these diverse projects, some of their challenges, and their wide-spread impact in the humanities and beyond.

In the second half, I will focus specifically on Digital Cultural Heritage (DCH). Recent DH cluster hires in Anthropology, Classics & Religious Studies, History, and Art & Art History at UNL are facilitating innovative research in DCH. In particular, CDRH scholars are applying and developing Geographic Information Systems (GIS), 3D Modeling, and Virtual Reality (VR) methods and tools to foster new avenues of scholarly research. Underlying much of this research is the need to unite quantitative and qualitative data—requiring new computational methods and 3DGIS tools. I will present some of my experiences, outcomes, and ongoing challenges for three DCH projects — MayaArch3D (2009-2015), MayaCityBuilder (2016-present), and Keeping Data Alive (2017-present) - situating them within the larger framework of Digital Humanities.

Biography:
Heather Richards-Rissetto is an archaeologist specializing in the ancient Maya of Central America. She is Assistant Professor in Anthropology, a Faculty Fellow in the Center for Digital Research in the Humanities (CDHR), and holds a Courtesy Appointment in the School of Natural Resources (SNR) at the University of Nebraska-Lincoln. She received her Ph.D. in Anthropology from the University of New Mexico and her undergraduate degree in Anthropology and Geography from the University of Southern Maine. She uses Geographic Information Systems (GIS) and 3D visualization to investigate how the accessibility and visibility of architecture communicated information and structured social experience and sent political and ideological messages in past societies. She is the Director of the MayaCityBuilder Project that uses procedural modeling for 3D Visualization, Analysis, and Discourse on Ancient Maya cityscapes. The MayaCityBuilder Project builds on the data and results of the MayaArch3D Project (2009-2015) — of which she was the GIS Director. She is also the CoPI on the Project “Keeping data alive: Supporting reuse and repurposing of 3D data in the humanities” — supported by a National Endowment for the Humanities (NEH) Tier I Research and Development Grant, Division of Preservation and Access. Her interests include using gesture-based and immersive technologies such as Microsoft Kinect, Leap Motion, and Oculus Rift to explore new avenues of digital scholarship.

Although half a decade has passed since Frank Rosenblatt_s original work on multi-layer perceptrons, modern artificial neural networks are still surprisingly similar to his original ideas.

In this talk I will give a brief introduction on deep neural networks and will question one of their most fundamental design aspects. As networks have become much deeper than had been possible or had even been imagined in the 1950s, it is no longer clear that the layer by layer connectivity pattern is a well-suited architectural choice. In the first part of the talk I will show that randomly removing layers during training can speed up the training process, make it more robust, and ultimately lead to better generalization. We refer to this process as learning with stochastic depth -- as the effective depth of the networks varies for each minibatch. In the second part of the talk I will propose an alternative connectivity pattern, Dense Connectivity, which is inspired by the insights obtained from stochastic depth. Dense connectivity leads to substantial reductions in parameter sizes, faster convergence, and further improvement in generalization. Finally, I
will show examples of problems that were considered challenging but have become surprisingly easy in the light of deep learning.

Biography:
Kilian Weinberger is an Associate Professor in the Department of Computer Science at Cornell University. He received his Ph.D. from the University of Pennsylvania in Machine Learning under the supervision of Lawrence Saul and his undergraduate degree in Mathematics and Computer Science from the University of Oxford. During his career he has won several best paper awards at ICML, CVPR, AISTATS and KDD (runner-up award). In 2011 he was awarded the Outstanding AAAI Senior Program Chair Award and in 2012 he received an NSF CAREER award. He was elected co-Program Chair for ICML 2016 and for AAAI 2018. Kilian Weinberger_s research focuses on Machine Learning and its applications. In particular, he focuses on learning under resource constraints, metric learning, machine learned web-search ranking, computer vision and deep learning. Before joining Cornell University, he was an Associate Professor at Washington University in St. Louis and before that he worked as a research scientist at Yahoo! Research in Santa Clara.

The idea of preconditioning iterative methods for the solution of linear systems goes back to Jacobi (1845), who used rotations to obtain a system with more diagonal dominance, before he applied what is now called Jacobi_s method. The preconditioning of linear systems for their solution by Krylov methods has become a major field of research over the past decades, and there are two main approaches for constructing preconditioners: either one has very good intuition and can propose directly a preconditioner which leads to a favorable spectrum of the preconditioned system, or one uses the splitting matrix of an effective stationary iterative method like multigrid or domain decomposition as the preconditioner.

Much less is known about the preconditioning of non-linear systems of equations. The standard iterative solver in that case is Newton_s method (1671) or a variant thereof, but what would it mean to precondition the non-linear problem? An important contribution in this field is ASPIN (Additive Schwarz Preconditioned Inexact Newton) by Cai and Keyes (2002), where the authors use their intuition about domain decomposition methods to propose a transformation of the non-linear equations before solving them by an inexact Newton method. Using the relation between stationary iterative methods and preconditioning for linear systems, we show in this presentation how one can systematically obtain a non-linear preconditioner from classical fixed point iterations, and present as an example a new two level non-linear preconditioner called RASPEN (Restricted Additive Schwarz Preconditioned Exact Newton) with substantially improved convergence properties compared to ASPIN.

"Mathematik, Informatik und Scientific Computing zum Anfassen"

Am 8. Juli laden die Fakultät für Mathematik & Informatik und das Interdisziplinäre Zentrum für Wissenschaftliches Rechnen (IWR) gemeinsam zum Tag der Offenen Tür im Mathematikon der Universität Heidelberg ein. Von Mathematik zum Anfassen über Forschung zu Künstlicher Intelligenz bis zum Robotiklehrlabor - Wissenschaftler geben spannende Einblicke in die Forschung und Praxis rund um Mathematik, Informatik und Scientific Computing. Führungen durch das neueste Gebäude der Universität Heidelberg ergänzen das abwechslungsreiche Programm.

Eine Auswahl an Speisen und Getränken wird durch die Fachschaft MathPhys angeboten.

Der Eintritt ist frei.

Weitere Informationen zum Programm werden in Kürze verfügbar sein.

Digital documentation and preservation is a key knowledge for archaeologists, no matter in which research area they are working in. This summer school gives a short introduction in the main digital methods currently used along with practical experience in real environments producing real data. The methods we want to show you are broad:

Structure from motion: This method uses a series of images to create 3D models. Can be used for small finds to whole buildings (with drones)
Drones: With drones, Structure from motion can acquire whole buildings or even landscapes. Several methods, from fixed wing (small airplanes) to multirotor drones will be shown.
Laser scanning: This device can create high precision point clouds in very short time and is very useful for all kind of documentation tasks.
Structured light scanner: This device is especially suited for small finds and objects. It projects actively light on the object to detect surfaces. The precision and quality is very high.
David scanner: A low-budget version of the Structured Light Scanner using laser beams. The results are very good.
Ground Penetrating Radar: Makes structures in the earth visible without excavating. One of the so called “non invasive methods”
Analysis of airborne LiDAR: Covering large areas, airborne LiDAR can help to detect a lot of archaeological remains, even in forested areas.
Geographic Information Systems: With the help of Geographic information systems, the analysis of historic movements, distribution systems and many more became possible.

The course teaches modern C++ programming skills from a Scientific Computing perspective. Special emphasis is given to modernizations and innovations from recent standards (C++11, C++14, C++17), topics that are often skipped in introductory courses due to high complexity and time constraints.

Topics that are discussed in detail include best practices in Scientific Computing, class concepts, dynamic memory allocation, exception
handling, safe handling of resources, template programming, static vs. dynamic polymorphism, traits and policies, the Standard Template Library, and template metaprogramming.

Cell nuclei are not always spherical; some are highly irregular. The structural basis for this multiplicity of shapes appears to derive from the components of the nuclear envelope and their interactions with cytoskeletal elements. Describing these shapes mathematically is a challenge; explaining their biochemical origin remains a mystery. The goal of this seminar is to discuss aspects of these problems and to provoke comments and ideas from the audience.

In this talk, I will provide an overview of solution techniques and iterative techniques employed to solve the first-order form of the radiation transport equation on massively parallel machines. A review of scaling efficiency for transport sweeps (up to order 1-million processes) will be provided for logically Cartesian grids. Challenges posed by the need to move to unstructured (load-unbalanced) grids and ongoing research will be discussed. Diffusion-based synthetic accelerators for the one-speed (within-group) and multigroup transport equations will be presented and issues related to massively parallel diffusion-accelerated transport sweeps be analyzed.

Biography:
Dr. Jean Ragusa specializes in computational methods for radiation (neutron, photon, coupled electron-photon) transport, radiative transfer, and multiphysics applications (e.g., radiation-hydrodynamics and two-phase flow modeling using a seven-equation model). Dr. Ragusa obtained his PhD from the University of Grenoble in 2001 and was a visiting assistant professor in the scholar of nuclear engineering at Purdue in 2001. From 2002 until 2004, he was a research engineer at the CEA-Saclay, France, in the reactor physics and applied mathematics division. In September 2004, he joined Texas A&M University where he is a professor of Nuclear Engineering and, since 2009, the associate director of the Institute for Scientific Computation.

The human brain has a complex geometric structure consisting of white and gray matter, blood vessels, ventricles, skull etc. It forms a highly anisotropic medium. Glioma in the brain are known to invade along white matter tracks and along other brain structures. Using diffusion tensor imaging (DTI) it is now possible to obtain directional information of the brain geometry. In my talk I will show how this DTI information can be used to parametrize a fully anisotropic diffusion equation for glioma spread. We validate the model on clinical data of glioma patients and discuss the future use in treatment design. (joint work with A. Swan, K.J. Painter, C. Surulescu, C. Engwer, M. Knappitsch, A. Murtha).

Followed by: “Meet the speaker” in the common room (with drinks and canapes)

With significantly increased number of graduate students entering career paths beyond academia, there is a pressing need to provide our students with a broad set of skills, in addition to their scholarly training. We enhance our students’ breadth and depth via intergenerational learning and professional development. Professor Meng will demonstrate these by discussing the formulation of, as well as his teaching experience in, Graduate Seminar in Undergraduate Education, the Harvard Horizons Program for communicating Ph.D research findings to general audiences, and the Center for Writing and Communicating Ideas.

In this talk, we will give an overview of different numerical methods we propose to carry out efficient numerical simulations of the kinetic Vlasov equation and for non-linear hyperbolic equations. The Vlasov equation is a kinetic transport equation in the position-velocity phase space. The numerical resolution of the Vlasov equation requires large computational resources combined with high accuracy for capturing the small scale dynamics (phase-space filamentation, gyrokinetic turbulence). By performing a finite-element semi-discretization in the velocity variable, we approximate the Vlasov equation by a hyperbolic system. Applying mass-lumping techniques on the resulting system, the approximate Vlasov equation reduces to several coupled transport equations. We consider either semi-Lagrangian method on Cartesian meshes or Galerkin-Discontinuous solver on multi-patch structured meshes to achieve high order accuracy in space. As an inverse strategy, we approximate non-linear hyperbolic equations by linear kinetic transport equations with a finite set of velocities and stiff relaxation (Lattice Boltzmann Method). Implicit resolution of the transport equations (by front tracking), combined with palindromic composition method in time (of order 2, 4 or 6) enables us to use large CFL numbers while being well adapted to parallelization.
In this talk I will give a basic introduction to multilevel Monte Carlo methods applied to uncertainty quantification problems in PDE applications. I will aim the talk at numerical analysts and computational scientists, familiar with numerical methods for partial differential equations. I will only assume basic knowledge of probability theory. A particular focus will be the study of the inverse problem, where PDE models with random coefficients that encapsulate the prior knowledge about the coefficient distribution are coupled with measurement data of functionals of the PDE solution. The talk will focus mainly on methodology, but I will refer to some theoretical results that underpin the new methods, as well as stress some open theoretical and computational problems.

Consider a rigid body that is fully submerged in a viscous fluid whose motion is governed by the Navier-Stokes equations.
I will discuss the asymptotic structure at spatial infinity of a steady flow around the body. In the 3D irrotational case, this classical problem has been almost fully resolved over the years, but many questions are still open in the 2D case. Indeed. the asymptotic structure of a 2D Stokes flow is closely related to the Stokes Paradox. In my talk, I will focus on bodies that are rotating with a constant angular velocity and discuss some new results for rotating planar flows.

Understanding the brain is an old and yet-unsolved problem. To understand the workings of a neural circuit, it is possibly required to know its structure, and almost surely necessary to know its connectivity.
After great progress in electron microscopy, several labs worldwide are milling away at animal brains and generating what will amount to petabytes of high-quality data. The resulting images are good enough for human tracers to consistently follow at least the majority of neural processes; unfortunately, humans would take thousands of years to complete the task for even the smallest mammalian brain.
So the quest is on for computer vision algorithms to do the same automatically and reliably. The current state of the art pipelines recur to deep neural networks and combinatorial graph partitioning problems. The former are notoriously ill understood, the latter still expensive to solve at scale.
In this talk, I will sketch the problem, a state of the art approach (which does not quite achieve human accuracy yet), and I will lay out some of the open problems in the field.

Interdisciplinary Seminar Series "Structures & Mathematics":

The workshop-seminars aim at initiating interactions between mathematicians and researchers from other sciences. We want to explore in particular questions and problems, which might be of interest to and benefit from the involvement of mathematicians of all sort. The setting of the seminar is informal, and interactive.

The term quantum turbulence denotes the turbulent motion of quantum fluids, systems such as superfluid helium and atomic Bose–Einstein condensates, which are characterized by quantized vorticity, superfluidity, and, at finite temperatures, two-fluid behaviour. In this talk we shall describe different regimes of quantum turbulence and how it compares to turbulence in a classical fluid. We shall close by discussing turbulence in a quantum analogue of a ferrofluid.

Die Bedeutung der mentalen Leistungsfähigkeit nimmt im Wandel der kognitiven Anforderungen einer modernen Arbeitswelt stetig zu. Der Zusammenhang von psychomentaler Überlastung und kognitiven Einschränkungen wird häufig unterschätzt, besonders bei Menschen mit hohem Leistungspotenzial bzw. im Rahmen anspruchsvoller beruflicher Tätigkeiten.
Als Weiterführung der Veranstaltung im Oktober 2016 wird im aktuellen Vortrag insbesondere das kognitive Stressmanagement im Vordergrund stehen.
Einführend soll ein Modell des Nobelpreisträgers Daniel Kahnemann Aufschluss geben, von welchen inneren Faktoren unser Denken und unsere Bewertungsstrategien abhängig sind und in welcher Weise unsere Emotionen unser Denken beeinflussen. Ein Abriss über kognitive Verzerrungen wird veranschaulichen, wie unsere Bewertungen und Schlussfolgerungen im Alltag beeinflusst bzw. verzerrt werden. Als Impuls für einen besseren Umgang mit unserem Denken werden unterschiedliche kognitive Strategien anhand praktischer Beispiele vorgestellt und diskutiert.
Abschließend werden wir der Frage nachgehen, warum Menschen manchmal völlig fern jeder Vernunft handeln, obwohl sie es „eigentlich-besser-wissen- müssten“.
Es werden keine Vorkenntnisse vorausgesetzt.
Bitte hier anmelden

Teaching undergraduate students is an effective way to enhance your own knowledge about your research field. Additionally, you develop communication skills relevant for a career within or outside academia.

This two-day course covers the basics of professional University teaching. You will improve your methodological knowledge about the teaching–learning interaction and how that setting can be influenced effectively. The aim is to reach a level of learner-centered teaching that leads to a deep-level learning approach on the side of the students. Hence, interaction with and motivation of the students is in the focus of this course.

- Basic principles of teaching and learning
- Understanding your role as a teacher
- Didactical planning of a course or lesson
- Defining learning objectives – designing learning activities
- Co-operative learning

The course work comprises of short inputs, discussion, group work and individual reflection of personal experiences. Participants are asked to be actively involved in the course by working on their own teaching tasks.

Please register: here

„Heimfahrten, Laptop, Fachliteratur & Semesterbeiträge“; kurzum an nahezu allen Kosten,
die euch vom Beginn des Masters bis zum Ende der Promotion begleiten, könnt ihr das Finanzamt beteiligen – sogar rückwirkend für die letzten Jahre. Der Effekt: Eine bis zu 5-stellige Steuerrückerstattung fließt nach dem Berufseinstieg auf euer Konto.Gleichzeitig gibt es wirtschaftliche Themen, die ihr in der Promotion bewegen solltet, um
frühzeitig und langfristig von wiederkehrenden Erstattungen zu profitieren. Dies wird ebenfalls aufgezeigt und besprochen.

Inhalt
Nach dem Training werden die Teilnehmer:
- die Vorgehensweise kennen, wie sie ihre Studienkosten rückerstattet bekommen.
- für sich wichtige Finanzthemen erkennen und einen der größten Steuerhebel nutzen können.
-wissen, wie sie einen Verlustvortrag generieren & Werbungskosten und Sonderausgaben steueroptimiert einordnen können.

Bitte hier anmelden

Research Posters have to meet several requirements: raising attention for your research project through an attractive design, summarizing the relevant information concisely and self-explaining, and generating a discussion about your work.
In the workshop you will learn about basic design aspects of research posters and receive feedback on your own draft. The course content will be as following:

- Part 1: „Design“– reducing complex content; layout principles; use of visual elements; technical tips; working on a first draft
- In-between: creating your own poster
- Part 2: short presentations; feedback on drafts/posters

You may bring along posters in English or German. Please note that software related questions (e.g. MS Powerpoint, InDesign, …) are not addressed in the course.

Please register
http://hgs.iwr.uni-heidelberg.de/Portfolio_HGS/VERANSTALTUNGEN/reg_form/reg_form.php?id=191