Balanced Atmospheric Data Assimilation

Prospective time: Monday, 11:00 am

Organizers: Gottfried Hastermann, Rupert Klein, Sebastian Reich
Speakers: Tommaso Benacchio (Met Office, United Kingdom)
Stefanie Hollborn (Deutscher Wetterdienst)
Maria Reinhardt (Universität Potsdam)
Nils Wedi (European Centre for Medium-Range Weather Forecasts)

The atmosphere's multi-scale structure poses several major challenges in numerical weather prediction. One of these arises in the context of data assimilation. The large scale dynamics of the atmosphere is balanced in the sense that acoustic or rapid internal wave oscillations generally come with negligibly small amplitudes. If triggered artificially, however, through inappropriate initialization or in the course of data assimilation, such oscillations can have a detrimental effect on forecast quality as they interact with the moist aerothermodynamics of the atmosphere.
The aim of this minisymposium is to discuss this issue from two different perspectives. First we will discuss extensions of existing assimilation strategies to avoid the artificial fast scale dynamics, secondly we will consider the problem from a model perspective, since multi-scale numerical methods may also provide an answer to the balancing problem.

Organization, Shape, and Dynamics of Biological Membranes

Prospective time: Monday, 11:00 am

Organizers: Carsten Hartmann, Ralf Kornhuber, Frank Noé, Christof Schütte, Thomas Weikl
Speakers: Carsten Gräser (Freie Universität Berlin)
Christoph Haselwandter (University of Southern California)
Michael Kozlov (Tel Aviv University)
Mohsen Sadeghi (Freie Universität Berlin)

The morphology and molecular organization of biological membranes is resulting from the interplay of a diversity of processes on multiple temporal and spatial scales. Typical modelling approaches involve and sometimes combine molecular dynamics and coupled systems of geometrical pdes. This minisymposium is devoted to the computational modelling of the organization, shape, and dynamics of biological membranes by chemical and mechanical interaction of proteins in bilipid layers.

Information-theoretical Methods for Complex Systems

Prospective time: Monday, 2:30 pm

Organizers: Sebastian Reich, Carsten Hartmann
Speakers: Michal Branicki (University of Edinburgh)
Valerio Lucarini (Universität Hamburg)
Manfred Opper (Technische Universität Berlin)
Tim Sullivan (Freie Universität Berlin)

Many natural systems are characterized by their high dimensionality and the presence of many characteristic spatio-temporal scales. A wealth of analytical and computational techniques has been proposed to extract low dimensional models that accurately capture the evolution of certain observables. The systematic use of data in the derivation and validation of the effective models based on information-theoretic tools has been a particularly important development. The mini symposium presents some of the most recent advances in the field.

Multiscale Methods for PDEs: Methodology and Applications

Prospective time: Monday, 2:30 pm

Organizers: Thomas von Larcher, Reinhold Schneider, Harry Yserentant, Sebastian Wolf
Speakers: Sergey Dolgov (University of Bath)
Vladimir Kazeev (University of Geneva)
Boris Khoromskij (Max-Planck-Institute for Mathematics in the Sciences, Leipzig)
Benjamin Stamm (RWTH Aachen)
Thomas von Larcher (Freie Universität Berlin)

Low-rank multilevel approximation methods are an important tool in numerical analysis and in scientific computing. Those methods are particularly suited to attack high-dimensional problems and solutions for nonlinear partial differential equations (PDEs) with high efficiency. In that research area, novel approaches as, e.g., hierarchical tensor product decomposition methods, are promising tools also for application to data that are concerned with cascade-of-scales problems, e.g., in molecular dynamics and in turbulent fluid dynamics. The interdisciplinary character of this session provides mathematical and computational aspects as well as application to data aspects with room for interchanges of ideas and discussions of future work.

Practical Approximations for Monte Carlo Simulations of Complex Systems

Prospective time: Tuesday, 11:00 am

Organizers: Han Cheng Lie, Tim Sullivan, Carsten Hartmann
Speakers: Ankush Agarwal (Ecole Polytechnique)
Yannis Pantazis (University of Crete)
Aretha Teckentrup (University of Edinburgh)

Many phenomena in the natural sciences and in engineering may be modeled using stochastic dynamical systems. Plain Monte Carlo simulation is the most straightforward computational method for studying the statistics of such systems. However, many of the research projects of the Collaborative Research Centre involve problems caused by rare events, coupled scales, or high-dimensional spaces. The computational cost renders plain Monte Carlo impractical for such problems. This minisymposium focuses on identifying potentially useful tools for and perspectives on Monte Carlo simulation. It emphasises the stochastic nature of the problems, with the purpose of developing better models and more efficient methods.

Deformation Accumulation in Seismic Faults and Networks

Prospective time: Tuesday, 11:00 am

Organizers: Onno Oncken, Matthias Rosenau, Alexander Mielke, Ralf Kornhuber
Speakers: Georg Dresen (Helmholtz Zentrum Potsdam)
Alice Gabriel (LMU München)
Elias Pipping (Freie Universität Berlin)
Jean Paul Ampuero (California Institute of Technology)

Earthquake statistics reveal scale invariance over 10 orders of magnitude of the eartquake strength as expressed, e.g., by the famous Gutenberg-Richter power law. However, the scaling properties underlying deformation accumulation in fault networks over longer time scales are virtually unknown. Moreover, because of the incompleteness of the real-world record of earthquakes and deformation accumulation beyond the instrumental and historical time scales (decades to centuries), there is a fundamental lack of insight into the multiscale nature of these processes. In this minisymposium we address mathematical modelling and numerical simulation as well as laboratory scale analogue models and experiments to explore the scaling properties of deformation accumulation in subduction zones and fault networks.

Stochastic Parametrizations for Convective Processes

Prospective time: Tuesday, 2:30 pm

Organizers: Uwe Ulbrich, Henning Rust, Christian Franzke (Uni Hamburg), Rupert Klein
Speakers: Boualem Khouider (University of Victoria, Canada)
Alexander Kuhn (Zuse Insitute Berlin)
Robert Plant (University of Reading)
Axel Seifert (Deutscher Wetterdienst)
Glen Shutts (Met Office, United Kingdom)

Convective processes play an important role in atmospheric dynamics. However, even state-of the art global circulation models still exhibit systematic biases in representing these processes since they act on scales comparable to or smaller than the grid scale. Alternatively to capturing their effect by parametric functions, stochastic models are now increasingly used to systematically represent the highly variable nature of these processes. Our Mini-Symposium will discuss a wide range of stochastic approaches.

Coarse Graining in Molecular Systems

Prospective time: Tuesday, 2:30 pm

Organizers: Stefan Klus, Christof Schütte, Ralf Kornhuber
Speakers: Cecilia Clementi (Rice University, Houston)
Michael Dellnitz (Universität Paderborn)
Stefan Klus (Freie Universität Berlin)
Wei Zhang (Freie Universität Berlin)

The ever-increasing complexity of molecular systems renders an efficient numerical analysis virtually impossible. The basic idea behind coarse graining is to replace the high-dimensional all-atom description of such systems by a reduced representation that preserves the properties of interest with a sufficient accuracy. The goal of the minisymposium is to present and discuss novel approaches for the systematic derivation of coarse grained models, their numerical realization with a focus on data-driven methods, and their validation based on both mathematical insight and real-life experiments.

Open Systems

Prospective time: Wednesday, 11:00 am

Organizers: Rupert Klein, Luigi Delle Site, Christof Schütte, Carsten Hartmann
Speakers: Jens Eisert (Freie Universität Berlin)
Patrick Ilg (University of Reading)
Michela Ottobre (Heriot Watt University, Edinburgh)
Matej Praprotnik (National Institute of Chemistry, Ljubljana, Slovenia)

Physical modelling, mathematical formalisation and numerical simulation of open systems is becoming more and more important for a broad field of applications, ranging from quantum computers to biological systems out of equilibrium. A common theme is that the applications require a precise and systematic modelling of the interaction of the system of interest with its environment, while being subject to computational constraints that limit the achievable level of detail in these models.
The talks in this minisymposium offer a look at the most recent advances in the field and include state-of-the art methods for modelling, simulating and analysing open systems.

Molecular Kinetics and Kinetic Models from MD simulations

Prospective time: Wednesday, 11:00 am

Organizers: Roland Netz, Robert Schulz, Julian Kappler, Jan Daldrop, Frank Noe, Robert Patterson
Speakers: Omar Valsson (ETH Zurich and USI Lugano)
Matthias Heyden (Max Planck Institut für Kohlenforschung)
Nuria Plattner (Freie Universität Berlin)

Gaining molecular insights into nanoscale kinetics from Molecular Dynamics (MD) simulations is challenging due to the large number of degrees of freedom involved and the limited computational power. This mini-symposium addresses applied and conceptual questions: An important issue of all-atom MD simulations is that physically relevant processes typically happen on time scales much larger than simulation time. Novel techniques to overcome this sampling problem, based on applying external bias potentials to collective variables, will be discussed. A separate issue concerns the meaningful analysis of simulation trajectories and the connection to experimental data. As an example, the translational and orientational dynamics of hydration water close to a biomolecular surface will be discussed.

Stochastic Modeling Approaches for Turbulent Flows

Prospective time: Wednesday, 2:30 pm

Organizers: Thomas von Larcher, Nikki Vercauteren, Rupert Klein
Speakers: Fenwick Cooper (University of Oxford)
Davide Faranda (LSCE laboratory, University of Paris-Saclay)
Christian Franzke (Universität Hamburg)
Nikki Vercauteren (Freie Universität Berlin)
Jeroen Wouters (Universität Hamburg, currently Marie Curie Fellowship in Sydney, AUS)

Turbulent flows are characterised by complex spatio-temporal dynamics involving multifaceted cascade-of-scales problems. The efficient modeling particularly of sub-grid scale turbulence requires understanding of the physics and statistics of scale interactions and has been a subject of intensive research over the past decades. Novel approaches as, e.g., stochastic/statistical methods and data-driven models, have been recently developed to analyse and represent scale interactions in hydrodynamic turbulence. The session provides discussions about recent developments and gives room for an exchange of ideas for future work.

Multi-scale Reaction Kinetics

Prospective time: Wednesday, 2:30 pm

Organizers: Mohsen Sadeghi, Frank Noé, Christoph Fröhner, Luigi Sbailò, Moritz Hoffmann
Speakers: Pieter Rein ten Wolde (AMOLF, Amsterdam)
Matthias Weiss (Universität Bayreuth)
Stefanie Winkelmann (Freie Universität Berlin)

Multi-scale reaction kinetics methods are used to study complex reactive systems. Examples are chemical signaling in living cells, ligand binding and catalytic reactions. These processes range in scales from nanoseconds to seconds in time, and from few nanometers to several micrometers in space. It is usually challenging to efficiently simulate reactive systems over different scales. However, several methods have been developed to successfully deal with multi-scale systems, these methods mainly use two different approaches. In concentration-based approaches the volume is divided into subvolumes of constant particle density, whereas particle-based approaches explicitly account for every particle in the system. This minisymposium is aimed at presenting a selection of these methods.