link to homepage

Institute for Advanced Simulation (IAS)

Navigation and service

Forschungszentrum Jülich - Workshop HYBRID2013, List of Abstracts

List of Abstracts

Invited Talks

Quantum-continuum coupling for chemo-mechanics problems
W. A. Curtin

Multiscale coupling of quantum mechanical (QM) domains to domains having a coarser-scale material description is necessary for non-periodic problems that may also involve long-range deformation fields such as caused by dislocations or crack tips. The goal of a multiscale method is to compute the interactions of the quantum domain with the surrounding domain with the same accuracy as would be obtained if the surrounding domain were fully quantum mechanical. The QM domain inevitably requires some type of cluster calculation, and the major errors then stem from electronic effects at the cluster surface extending into the cluster interior and generating spurious forces and incorrect physical configurations. Here, we discuss two recent methods to achieve robust coupling using full Kohn-Sham DFT methods. The first method uses a thick buffer region of quantum ions and electrons in which the ionic displacements are determined by elasticity or atomistic methods. The second method uses the concept of constrained DFT to force the electronic configuration near the outer boundary of the cluster to be identical to an approximate bulk electronic charge density, with the ion positions again controlled by elasticity or atomistic methods. The success of the two methods is demonstrated through application to several simple test problems.

Inertial coupling: A minimal model to resolve inertial effects in particle hydrodynamics
R. Delgado-Buscalioni and F. Balboa Usabiaga

We present an overview of a novel method for hydrodynamics of small particles in fluid solvent. The method consistently solves the fluid and particle inertia and accounts for thermal fluctuations in the fluid momentum equation. The coupling between the fluid and the blob is based on a no-slip constraint equating the particle velocity with the local average of the fluid velocity, and conserves momentum and energy. Owing to the non-dissipative nature of the no-slip coupling, the fluctuation-dissipation balance is possible without addition of extra particle noise. The local averaging and spreading operations are accomplished using compact kernels commonly used in immersed boundary methods. These kernels make the discrete blob a particle with surprisingly physically-consistent volume, mass, and hydrodynamic properties. The present inertial coupling method can model particulate flows in a wide range of time-scales ranging from Brownian to convection-driven motion, using a minimal cost. It can be naturally extended to polymeric fluids and other types of physico/chemical phenomena.

Creating a "chemistry" for colloids in a liquid crystal matrix
C. Denniston

Colloids in a liquid crystal matrix exhibit very anisotropic interactions. Further, these interactions can be altered by both properties of the colloid and of the liquid crystal. This gives a potential for creating specific colloidal aggregates and crystals by manipulating the interactions between colloids. However, modelling these interacting colloids in a liquid crystal is very challenging. We use a hybrid particle-lattice Boltzmann scheme that incorporates not just hydrodynamic forces but also forces from the liquid crystal field. The algorithm and some of the resulting structures will be described.

On constructing a Hamiltonian multiscale molecular dynamics method
S. Nielsen, A. Laio, and B. Ensing

In this lecture, we discuss two different approaches to treat an adaptive hybrid atomistic/coarse- grain simulation. Such a multiscale algorithm treats the most interesting part of the system in accurate fine-grained (atomistic) detail, whereas the environment is modeled at a less detailed and less computationally demanding resolution. Our interest is in open boundaries: molecules are allowed to diffuse between the different resolution regions and adapt their resolution on the fly. Here we compare our existing approach with a new approach based on a Lagrangian formulation which includes both resolution-switching forces and a constraint on the number of particles in each resolution region.

Atoms, dislocation, continua - how does it go together?
A. Hartmaier

First-principles and tight-binding quantum chemical molecular dynamics simulations on chemical mechanical polishing processes
Y. Higuchi and M. Kubo

Chemical mechanical polishing (CMP) consists of multi-physics processes including friction, fluid, heat and chemical reaction and then the detailed mechanism of the CMP has not been clarified. We have successfully developed a multi-physics process simulator based on our tight-binding quantum chemical molecular dynamics (TB-QCMD) method. This simulator is very effective to clarify the multi-physics process including chemical reaction, friction, impact, fluid, heat etc. In the present study we applied our TB-QCMD code to the CMP processes of Cu surface by SiO2 particles. Under H2O2 and H2O solution, the Cu surface was oxidized by H2O2 and SiO2 particles. Then Cu surface was softened and easily polished. We clarified that the friction force of SiO2 particle accelerates the chemical oxidation of Cu surface. Activation barrier for the oxidation reaction obtained by the first-principles calculation supports the above results. Then we confirmed the effectiveness of our TB-QCMD code.

Multiscale flow simulations using particles
P. Koumoutsakos

Particle methods provide a powerful framework to describe a wide range of physical phenomena across multiple scales. The simplicity of the method, as a set of interacting particles, allows for a unifying formulation encompassing discrete atomistic/mesoscale models and approximations of conservation laws. In this talk I will present work on developing wavelet adapted particle methods for the simulation of vortical flows. I will discuss the mapping of these algorithms to multi/many-core architectures and the coupling of atomistic, mesoscale and continuum particle based flow solvers. I demonstrate the computational capabilities and the versatility of the proposed particle formulation by presenting applications ranging from fish swimming, to cancer induced angiogenesis and nanofluidics.

Atomically smooth cleavage of hydrogen-implanted silicon crystals: a hybrid quantum/classical MD study
G. Moras, L. Colombi Ciacchi, Ch. Elsässer, P. Gumbsch, and A. De Vita

We present a quantum-accurate multiscale study of how hydrogen-filled discoidal “platelet” defects grow inside a silicon crystal. Hybrid quantum/classical molecular dynamics simulations of a 10-nm-diameter platelet reveal that H2 molecules form at its internal surfaces, diffuse, and dissociate at its perimeter, where they both induce and stabilize the breaking up of highly stressed silicon bonds. A buildup of H2 internal pressure is neither needed for nor allowed by this stress-corrosion growth mechanism, at odds with previous models. A mesoscale model based on our atomistic results predicts slow stress-corrosion platelet growth up to micrometric sizes, making atomically smooth crystal cleavage possible in implantation experiments.

Linear scaling electronic structure calculations in liquids: many body expansion and Coulomb coupling
L. Pastewka, T. Järvi, and M. Moseler

We present a linear-scaling method based on self-consistent charge non-orthogonal tight-binding. Linear scaling is achieved using a many-body expansion, which is adjusted dynamically to the instantaneous molecular configuration of a liquid. Electrostatic embedding of the subsystem related to one-body and two-body terms of the expansion is taken into account in an energy functional. Variation of the functional with respect to the wave functions of the subsystems lead to a hierarchy of secular equations and corresponding atomic forces that obey the Hellmann-Feynman theorem. The method is capable of simulating liquids over large length and time scales, and also handles reactions correctly. Benchmarking on typical carbonate electrolytes used in Li-ion batteries displays excellent agreement with results from full tight-binding calculations. The new method should be useful for a wide range of applications e.g. from electrochemistry or tribology.

(Bio)molecule solvation in aqueous mixtures: From an "effective" to a "truly" open boundary
D. Mukherji and K. Kremer

(Bio)physical properties of macromolecules in aqueous mixtures are dictated by the preferential interactions of co-solvents with the solute. The numerical studies using small-sized closed boundary setups, however, suffer from severe system size effects. Especially, when the concentration fluctuations are large or intimately linked to the physical properties. Therefore, we will present two related efficient approaches that makes use of the adaptive resolution scheme (AdResS), where an all-atom region is coupled to a coarse-grained reservoir. In one case, we use the conventional AdResS scheme, and in another, we combine the AdResS scheme with a metropolis particle exchange criterion. We calculate well converged Kirkwood-Buff integrals, which connect the solvation energies to the molecular distributions, within the small all-atom region that are impossible from an all-atom MD of similar system sizes. Results will be presented for the solvation of a tri-glycine in aqueous urea and coil-globule-coil transition of a poly(N-isopropylacrylamide) in aqueous methanol.

Studying slow collective phenomena by concurrently coupling particle-based and continuum descriptions
M. Müller

The simulation of collective phenomena by particle-based simulations poses a computational challenge because of (i) the wide spread of time scales or (ii) the presence of free-energy barriers along the transformation path. A prototypical example of the former difficulty of multiple disparate time scales is the simultaneous presence of stiff bonded interactions, defining the molecular architecture of polymer systems and the weak non-bonded interactions, giving rise to phase separation or structure formation in dense multicomponent systems. A characteristic illustration of the latter problem are nucleation barriers or metastable intermediate states in the course of macrophase separation or self-assembly. Continuum models, in turn, describe the system by a collective order-parameter field, e.g., the composition, rather than particle coordinates and often do not suffer from these limitations because (i) the stiff molecular degrees of freedom have been integrated out or (ii) advanced numerical techniques like the string method exist that identify free-energy barriers and most probable transition paths. Using field-theoretic umbrella sampling, we determine an approximation of the continuum free-energy functional for a specific particle-based model. We illustrate how (i) the so determined free-energy functional can be used in conjunction with a heterogeneous multiscale method to speed-up the simulation of Lifshitz-Slyozov coarsening in a binary polymer blend by two orders of magnitude [1] and (ii) the on-the-fly string method can identify the minimal free-energy path for the formation of an hourglass-shaped passage (stalk) between two apposing bilayer membranes.

[1] Speeding up intrinsically slow collective processes in particle simulations by concurrent coupling to a continuum description, M. Müller and K.Ch. Daoulas, Phys. Rev. Lett. 107, 227801 (2011)
[2] Transition path from two apposed membranes to a stalk obtained by a combination of particle simulations and string method, M. Müller, Y.G. Smirnova, G. Marelli, M. Fuhrmans, and A.C. Shi, Phys. Rev. Lett. 108, 228103 (2012)

Modeling the dielectric response of atomistic and continuous media with the split-charge method
M. Müser

Many processes involving ions, polar molecules, or polar moieties take place in an external medium with heterogeneous dielectric properties. Examples range from protein folding in a polarizable solvent to contact electrification induced by the rubbing of two dislike solids. When simulating such processes, it is not appropriate to decompose the electrostatic forces between the central atomistic degrees of freedom into (effective) two-body contributions. Instead, one needs to consider the dielectric response of the external medium, which one may want to represent as a continuum. In this contribution, we show that the split-charge equilibration (SQE) method can be used to describe continua with well-defined dielectric properties, although it was originally designed to assign atomic charges on the fly. As such, SQE bears much potential for hybrid particle-continuum simulations. The comparison of dielectric response functions as obtained by SQE and point dipole methods reveals many advantages for SQE. The main points are: SQE requires fewer floating-point operations, non-local dielectric properties are more easily embedded, and the leading-order corrections to the continuum limit are isotropic on the simple cubic lattice in contrast to point dipole models.

Numerical simulations of the optical response of atomic clusters
M. Sukharev and A. Nitzan

The interaction of electromagnetic radiation of arbitrary polarization with a system of multi-level atoms in is computed by combining the Maxwell equations for the radiation field with the Liouville-Bloch equations for the atoms in a self consistent manner. We examine the linear optical properties of nanoscale atomic clusters demonstrating significant role played by collective effects and dephasing. The optics of core-shell nanostructures, with a metal core and a shell composed of resonant atoms, is studied as well.

Theory and practice of adaptive resolution simulations
R. Potestio

In the last few decades computer simulations have become a fundamental tool in the field of soft matter science, allowing researchers to investigate the properties of a large variety of systems. Nonetheless, even the most powerful computational resources presently available are, in general, only sufficient to simulate complex biomolecules for a few nanoseconds. An important limitation, hampering the achievement of larger length and time scales, is represented by the need to simulate with fine-grained detail a consistent fraction of the system, such as the solvent far from the solute's surface, which is eventually removed from the subsequent analysis. In order to overcome this limitation, several schemes have been developed, where a subregion of the system is described with a higher resolution -typically at the atomistic level- with respect to a surrounding thermal bath containing solvent molecules in a coarser representation; open boundaries between these regions and a position-dependent resolution switch guarantee that the correct thermodynamics is preserved in the high-resolution region of interest. During this lecture an overview of the available hybrid resolution simulation schemes will be given; the focus will be on their theoretical background. Applications of these methods will be discussed, as well as future directions of research.

Concurrent atomistic/continuum modeling of fluids: Transport of solvent, heat and ions
L. Guo, J. Liu, M. Wang, X. Nie, S. Chen, and M. O. Robbins

Hybrid atomistic/continuum methods have been developed that allow full atomic detail to be retained at interfaces while more efficient continuum models are used in other regions. The two descriptions are coupled in an overlap region where the fluxes from one description provide boundary conditions for the other. Atoms and heat can flow seamlessly from one description to the other. The most recent version treats charged systems. A discrete description of ions is retained in the continuum region because of the low density of ions and the long-range of electrostatic interactions. The solvent is treated implicitly with ions following Langevin dynamics relative to the local solvent velocity. A multi-grid particle-particle particle-mesh (PPPM) method is used throughout all regions to calculate Coulomb interactions. Results from this approach will be presented for slip at rough walls, singular corner flows in cavities, contact line motion, heat transport and electroosmotic flow. In each case, pure MD results are used to validate the method.

Modeling atomic scale structure and dynamics at interfaces on diffusive timescales
J. Rottler

Materials phenomena at interfaces such as solidification, solid-solid phase transformations and thin film growth involve complex structural changes that couple atomic scale elastic and plastic effects with diffusional processes. Conventional molecular dynamics is unable to address these problems on experimentally relevant timescales. Highly coarse-grained order parameter (phase field) theories based on Ginzburg-Landau effective Hamiltonians are in widespread use, but are devoid of any atomic level features. Free energy functionals that are minimized by periodic ground states instead (phase field crystals) reflect the discrete structure of matter, and elastic interactions between local deviations from the ground state in the form of defects are naturally incorporated without being constrained to the phononic timescales of MD. Despite their simplicity, these functionals have been shown to capture a wide range of phenomena on a semiquantitative (i.e. scaling) level, including dislocation dynamics, grain boundary energetics, grain coarsening and crystal plasticity. Here we present an overview of this approach including a new formulation that permits the selection of ground states with different symmetries (sc, bcc, fcc, etc) [1] and determine the phase diagram as well as elastic properties [2]. We also extend the methodology to binary alloys [3] and quasicrystals [4]. The versatility of the technique is illustrated with three examples: nucleation of structurally different daughter phases after a temperature quench [1], lamellae growth in eutectic alloys [3] and monolayer pseudomorphic growth on quasicrystalline surfaces [4].

[1] M. Greenwood, N. Provatas, J. Rottler, Phys. Rev. Lett. 105, 045702 (2010).
[2] M. Greenwood, J. Rottler and N. Provatas Phys. Rev. E 83, 031601 (2011) .
[3] M. Greenwood, N. Ofori-Opoku, J. Rottler, and N. Provatas, Phys. Rev. B 84, 064104 (2011).
[4] J. Rottler, M. Greenwood, and B. Ziebarth , J. Phys.: Condens. Matter 24, 135002 (2012).

Parallel simulation of Brownian dynamics with MPI, OpenMP and UPC
C. Teijeiro, G. Sutmann, G. L. Taboada, and J. Touriño

This work presents the design and implementation of a parallel simulation code for the Brownian motion of particles in a fluid. Three different parallelization approaches have been followed: (1) using traditional distributed memory message-passing programming with MPI, (2) using a directive-based approach on shared memory with OpenMP, and (2) using the Partitioned Global Address Space (PGAS) programming model, oriented towards hybrid shared/distributed memory systems, with the Unified Parallel C (UPC) language. According to the selected environment, different domain decompositions and work distributions are studied in terms of efficiency and programmability in order to select the most suitable strategy. Performance results on different testbeds and using a large number of threads are presented in order to assess the performance and scalability of the parallel solutions.

AA: a super coarse-grained model for disordered proteins
A. Ghavami, P.R. Onck, and E. van der Giessen

Recent studies have revealed the key role of natively-unfolded proteins in many important biological processes. In order to study the conformational changes of these proteins, a one-bead-per-amino-acid (AA) coarse grained model is developed and a method is proposed to extract the potential functions for the local interactions between beads. Experimentally obtained Ramachandran data for the coil regions of proteins are converted into distributions of pseudo-bond and pseudo-dihedral angles between neighboring alpha-carbons in the polypeptide chain. These are then used to derive bending and torsion potentials, which are residue and sequence specific.

Simulations of colloids and self-propelled particles with fully resolved hydrodynamics
R. Yamamoto and J. Molina

Using the smoothed profile method (SPM) method [1,2] developed for direct numerical simulations (DNS) of colloidal dispersions, we studied several dynamical problems of particle dispersions, including their rheological behaviors under steady and oscillatory shear flows. Recently, the SPM is extended for dispersions of self-propelled particles by replacing the non-slip boundary condition (usual for colloids) with an actively slip boundary condition (to model squirmers) at the fluid/particle interface. Several dynamical behaviors of the hydrodynamically interacting self-propelled particles will be discussed [4].

[1] Y. Nakayama, K. Kim and RY, EPJE, 26, 361-368 (2008).
[2] KAPSEL website,
[3] R. Tatsumi and RY, PRE, 85, 066704 (2012).
[4] J. Molina, Y. Nakayama, and RY, preprint.

Contributed Talks

Local elastic fields in granular solids
J. Boberski, L. Brendel, and D. E. Wolf

The modeling of elastic properties of disordered or granular solids requires a theory of elasticity that takes non-affine deformations into account. Using a linearized force law, the non-affine elastic deformations are calculated. Based on the microscopically exact expressions for the local strain and stress fields a way to calculate maps of the local linear elastic constants for frictional granular packings is presented. The elastic constants are found to be scale and system size independent within an appropriate parameter range.

On the atomistic and continuum modeling: Theoretical link and numerical examples
D. Davydov, A. Javili, and P. Steinmann

In order to bridge the gap between particle-based models and continuum approaches, as well as to enhance the MD-FE simulation scheme, based on the fundamental principles of classical mechanics and statistical physics the basic framework to link the atomistic and the continuum world is introduced. The approach allows a comparison of the MD and surface-enhanced FE solutions. Several benchmark examples (both molecular statics and molecular dynamics) are considered. We compared atomistic fields obtained from the averaging procedure to their counterpart, obtained from numerical approximations to the surface-enhanced continuum theory, whereby the surface is equipped with its own constitutive structure. The ability of the continuum formulation enhanced with a surface energy to model size effects, as observed in the atomistic simulations, is studied. The local fields evaluated using both the continuum and discrete approach are compared. Possible usage in FE-MD coupling schemes is discussed.

Parallel multiscale simulations of a brain aneurysm
D. A. Fedosov, L. Grinberg, and G. Em Karniadakis

Cardiovascular pathologies, such as a brain aneurysm, are affected by the global blood flow as well as by the local microrheology. Computational models for such problems require the coupling of disparate spatio-temporal scales often governed by diverse mathematical descriptions (e.g., atomistic, mesoscopic, continuum). However, coupling of particle-based with continuum-based methods is a challenging problem that requires both mathematical and computational advances. We will present a hybrid methodology that enables multiscale simulations of platelet deposition in a brain aneurysm. The large scale flow features are resolved by using the high-order spectral element Navier-Stokes solver. The blood rheology inside the aneurysm is modeled using a mesoscopic approach based on the dissipative particle dynamics method. The continuum and particle domains overlap with interface conditions provided by effective forces to ensure continuity of states across the interface boundary. We will present simulation results on clot formation inside the aneurysm and discuss the computational challenges involved.

Modelling the oral bacterial ecosystem and other biofilms
D.A. Head, P.D. Marsh, and D. Devine

Biofilms are sessile microbial communities that arise frequently in nature, and form an integral part of our own microbiome. Modelling biofilms is challenging as it couples biology (microbe growth and division) to chemistry (reaction, diffusion and advection of nutrients and metabolites) to physics (biofilm elasticity in the presence of flow). Natural biofilms exhibit chemical gradients and architectural structures on cellular length scales, thus a representative model must treat the biofilm as particulate, while retaining a continuum description for small dissolved molecules. Here I will present work developing a software platform coupling all of the key features mentioned above, and highlight two early applications. (1) Modulating dental plaque to be in its healthy state by subjecting the system to low doses of fluoride; (2) The rapid growth of surface roughness and how it is smoothed by shear flow.

Hydrodynamic correlations in multi-particle collision dynamics fluids
C.-C. Huang, G. Gompper, and R. G. Winkler

We analyze, both theoretically and numerically, the hydrodynamic correlations of a multiparticle collision dynamics (MPC) fluid, a particles based mesoscale simulation method. The fluid is characterized by its longitudinal and transverse velocity correlation functions in Fourier space and velocity autocorrelation functions in real space. Particular attention is paid to the role of sound, which leads to piecewise negative correlation functions. The minimal length scale in MPC for hydrodynamics is investigated. In addition, dilute polymer solutions are considered. We investigated the center-of-mass velocity-autocorrelation function of a polymer, unravelling the influence of sound. At long times, the correlation function exhibits a long-time tail decaying algebraically as t-1.5, which is independent of any polymer property and solely depends on the solvent viscosity.

Massively parallel molecular-continuum simulations with the macro-micro-coupling tool
Ph. Neumann and J. Harting

Efficient implementations of hybrid molecular-continuum flow solvers are required to allow for fast and massively parallel simulations of large complex systems. Several coupling strategies have been proposed over the last years for 2D/ 3D, time-dependent/ steady-state or compressible/ incompressible scenarios. Despite their different application areas, most of these schemes comprise the same or similar building blocks. Still, to the author's knowledge, no common implementation of these building blocks is available yet. In this contribution, the Macro-Micro-Coupling tool is presented which is meant to support developers in coupling mesh-based methods with molecular dynamics. It is written in C++ and supports two- and three-dimensional scenarios. Its design is reviewed and aspects for massively parallel coupled scenarios are addressed. Afterwards, scaling results are presented for a hybrid simulation which couples a molecular dynamics code to the Lattice Boltzmann application of the Peano framework.

Molecular dynamics meets finite elements: an approach for coupled simulations of nanocomposites
S. Pfaller, G. Possart, P. Steinmann, M. Rahimi, M. C. Böhm, and F. Müller-Plathe

In contrast to field based continuum mechanics, particle based methods can take into account the specific atomistic structure of the material under consideration. However, in engineering approaches, they are often computationally prohibitive due to the huge number of particles to be considered. In our approach the system consists of a particle region that is coupled to a continuum by introducing a bridging domain where both regions overlap. The particle domain is computed by Molecular Dynamics (MD) at finite temperature, while the continuum is discretized and solved using the Finite Element Method (FEM). In addition to existing coupling schemes, the particles are tethered to anchor points which transfer displacements and forces between the different domains. The work to be presented is a results of a collaboration with the Theoretical Physical Chemistry Group at TU Darmstadt and also part of the DFG-priority programme 1369 "Polymer-Solid Contacts: Interfaces and Interphases''.

Hybrid particle-field representation simulations in soft condensed matter systems
Q. Shuanhu

A multiscale hybrid model combining the particle-representation method and field-representation method is developed for the simulations in soft condensed matter systems. The hybrid model treats part of the system as particles, while the other part as field, and particles in different representation regions can switch and migrate on the fly. The switch and migration of particles from different resolution regions are controlled by the inhomogeneous chemical potential difference. The hybrid model is tested in a colloid diblock copolymer systems with the comparison to that of the pure particle representation method, and good agreements are obtained.

A hybrid particle-field description for complex fluid dynamics
G.J.A. Sevink, K.M. Langner, M. Charlaganov, and J.G.E.M. Fraaije

We developed a hybrid method that combines discrete particle-based models such as Brownian Dynamics (BD) with continuous field descriptions like Dynamic Density Functional Theory (DDFT). The approach allows for a natural separation of sparse and abundant constituents and enables simulation of sufficiently large systems with (coarse-grained) molecular detail. The particle- and field-based subsystems evolve simultaneously; both constituents are not spatially restricted and the boundary between the subsystems does not require any explicit treatment. To account for excluded volume effects and interactions , we introduce a coupling term that contributes to both the conservative force acting on the particles and the intrinsic chemical potential that governs the dynamics of the fields. The coupling parameter can be determined for pure systems based on direct thermodynamic mapping. We illustrate the potential of the new method by showing results for liposomes and nanocomposites.

Connecting continuous and discrete system using control volume
E.R Smith, D.M Heyes, D Dini, and T.A. Zaki

All coupling schemes share a common element - exchange of properties between the molecular and continuum regions. However, the treatment of both regions appears incompatible, consisting of N-discrete molecules on one hand and continuous fluid flow through fixed volumes on the other. In order to address this problem, we apply the mathematical framework from our recent paper [E.R Smith, D.M Heyes, D Dini, T.A. Zaki Phys. Rev. E 85, 056705 (2012)] expressing the dynamics of the discrete system in a control volume formulation. This yields the relationship between molecular time evolution of mass, momentum and energy and the surface fluxes and forces. With both systems in an equivalent mathematical framework, it is then possible to derive consistent coupling schemes between the two regions. Using this Control Volume approach, we derive coupling algorithms for large-scale simulation of continuum-MD systems on HPC facilities.


The posters will be mounted on movable walls provided by the organizers. The maximum size of a single poster should not exceed 90cm widths and 145cm height (portrait format).

The number in brackets in front of the title is the number of the movable wall where to place the poster for poster session.

[9] Adaptively restrained particle simulations
S. Artemova and S. Redon

Particle simulations are widely used not only in molecular dynamics, but also in fluid dynamics (dynamics of liquid, gas and plasma), celestial mechanics, and even in computer graphics. Numerous algorithms have been developed to accelerate these simulations, but many problems still remain challenging, e.g molecular docking, protein folding, diffusion across bio-membranes, fracture in metals, ion implantation, etc. We believe that faster simulations may result in progress on these important problems. We have introduced a novel general approach to speed up particle simulations that we call ARPS: Adaptively Restrained Particle Simulations. This approach is based on Hamiltonian dynamics of particle systems, and the inverse inertia matrix in the Hamiltonian is made a general function of phase-space coordinates. Due to this modified inverse inertia, positional degrees of freedom in the system are repeatedly switched on and off during the simulation. As a result, under frequently-used assumptions on the interaction potential, less forces may be computed at each time step. This allows us, when performing constant-energy simulations, to finely and continuously trade between precision and computational cost, and rapidly obtain approximate trajectories. Moreover, when performing AR simulations in the canonical (NVT) ensemble, correct static equilibrium properties can be computed. This poster will present the main ideas of the new approach, as well as some of the numerical examples illustrating the advantages of ARPS.

[1] SPH simulation of particulate suspension under Couette flow
Xin Bian and Marco Ellero

We present simulation results of hard particles suspended in a Newtonian solvent by using the smoothed particle hydrodynamics (SPH). SPH is a mesh-free method, which is able to discretize the Navier-Stokes equations on a set of Lagrangian particles. Rigid structures are modeled by restricting a set of SPH boundary particles to translate/rotate together. In a low-Reynolds number regime, a particulate suspension confined in a channel undergoing Couette flow will be studied and its viscosity dependence on solid concentration, confinement and shear rate presented. The microstructure and effective viscosity of the suspension in each case are tightly connected. In particular, particle radial distribution function in the shear-thickening regime will be analyzed in detail, where anisotropy is expected to occur. Further on, the hydro-cluster hypothesis will be validated by analyzing cluster-size distribution in different system sizes, where we found that the theory of hydro-cluster alone does not explain the complete shear-thickening regime.

[2] A multiscale method applied to nano/micro fluidic channel networks
M. Borg, D. Lockerby, and J. Reese

We present a new hybrid molecular-continuum methodology for resolving multiscale flows emergent in nano-/micro-scale networks, in particular for NEMS/MEMS applications. The method models junction and channel components of the network using independent MD micro elements. Long channels with uniform or gradually varying nano-scale sections along the direction of flow, contribute the most towards the highest computational savings, by replacing them with much smaller MD simulations. Junction components, however, do not exhibit any length-scale separation and are modelled in their entirety. All micro elements are coupled together in one hybrid simulation using standard continuum fluid-dynamics equations, that dictate the overall macroscopic flow in the network. In the case of isothermal, incompressible, low-speed flows we use the conservative continuity and momentum equations. An iterative algorithm is presented that computes at each iteration the new constraints on the pressure differences applied to individual micro elements, in addition to enforcing overall continuity in the network. We show that the hybrid simulation of various small network cases converge quickly to the result of a full MD simulation over just a few iterations, with significant computational savings.

[27] Atom-based simulations of the discharge of a battery
W. Dapp and M. Müser

Batteries are pivotal components in overcoming some of today’s greatest technological challenges. Yet to date there is no self-consistent atomistic description of a complete battery. We take first steps towards modeling of a battery as a whole microscopically. Our focus lies on phenomema occuring at the electrode-electrolyte interface which are not easily studied with other methods. We use the redox split-charge equilibration (redoxSQE) method that assigns a discrete ionization state to each atom. Along with exchanging partial charges, atoms can swap integer charges across bonds. With redoxSQE, we study the discharge behavior of a nano battery, and demonstrate that this reproduces the generic properties of a macroscopic battery qualitatively. Examples are dependence of the battery’s capacity on temperature and discharge rate, as well as performance degradation upon recharge.

[16] Unraveling and eliminating dissipation mechanisms in contacts of polymer-bearing surfaces
S. de Beer and M. H. Müser

Polymer brushes are well known to lubricate high-pressure contacts, because they can sustain a high normal load while maintaining low friction at the interface. Depending on the contact-geometry, direction of motion and brush characteristics, different dissipation mechanisms dominate the friction forces. For example, in a parallel plate geometry the interdigitation of the opposing polymers determines the lubricity [1], while for spherical star polymers in relative motion, viscoelastic deformation governs the energy dissipation [2]. We discuss the relative importance of the dissipation channels for real contacts and show via molecular dynamics simulations that, by using an asymmetric contact of a hydrophobic and hydrophilic polymer-bearing surface, the important dissipation-mechanisms – interdigitation and capillary break-up – can be eliminated. This can reduce friction by a few orders of magnitude compared to a symmetric contact. Our proposed system therefore holds great potential for applications in industry.

[10] Chain deformation in translocation phenomena
F. Farahpour, F. Varnik, and M. Reza Ejtehadi

Deformation of single stranded DNA in translocation process before reaching the pore is investigated. By solving the Laplace equation in a suitable coordinate system and with appropriate boundary conditions, an approximate solution for the electric field inside and outside of a narrow pore is obtained. With an analysis based on “electrohydrodynamic equivalence” we determine the possibility of extension of a charged polymer due to the presence of an electric field gradient in the vicinity of the pore entrance. Such deformation is shown to have a great contribution to the capturing process, the first stage of any translocation phenomenon, especially in the diffusion-limited regime. With a multi-scale hybrid simulation (LB-MD) it is shown that an effective deformation before reaching the pore occurs which facilitates the process of finding the entrance for the end monomers.

[17] Local density dependent potential for compressible mesoparticles
G. Faure, J.B. Maillet, and G. Stoltz

Coarse-graining methods are a popular way of replacing an expensive atomistic system by larger particles interacting with an effective potential. When a detailed description of the material is not needed, it is interesting to coarse-grain at a larger scale by including several molecules into a mesoparticle in order to save some CPU time. While, at a molecular scale, it is possible to consider rigid molecules even at high pressure, it is necessary to take into account the compressibility of the mesoparticles. We explore the possibility of doing so by introducing a dependence on the local environment of the particle in the potential either through a local density (computed on the particles within a certain radius of the mesoparticle) or through the volume of the Voronoi cell of the mesoparticle.

[11] Modeling the mesoscale morphology of polymeric semiconductors using soft models
Patrick Gemünden, Carl Pölking, Kurt Kremer, Denis Andrienko, and Kostas Ch. Daoulas

We will explore a combined coarse-graining approach for modeling the mesoscale morphology of polymeric semiconductors focussing on P3HT systems. Bonded interactions are obtained from systematic coarse-graining of atomistic P3HT configurations. The non-bonded interactions are captured phenomenologically by soft-directional potentials designed to mimic local stacking effects. Their form is inspired by a functional of collective variables. The deep connection between particle and density-functional based description simplifies model parameterization and comparison with mean-field results. However, since particles are explicitly present, standard simulation techniques (e.g. MC simulations) can be applied. We will show that we reproduce reasonable conformational, thermodynamic (phase transitions), and, despite the softness of the interactions, material properties (Frank elastic constants). Subsequently we present first results on charge transport calculations based on the coarse-grained morphologies.

[24] Diffusion based adaptive load-balancing for domain decomposition in particle simulations
Rene Halver, Martin Reissel, and Godehard Sutmann

In the present work we consider a load-balancing scheme for domain decomposition schemes in particle simulations, based on local work-diffusion. The scheme is based on local work exchange between (i) neighbored domains in x-direction, (ii) neighbored domain columns (formed by domains in x-direction) and (iii) neighbored domain surfaces (formed by domains in x-y-layers). The problem corresponds to a time dependent global optimization problem, which is solved in the present approach iteratively. The assumption which is made here is a quasi-static scenario, i.e. load-imbalances develop on a slower time scale than the convergence of our scheme. It is demonstrated that this assumption is justified for typical density distributions in inhomogeneous particle systems under study. A mathematical analysis is provided which shows convergence of the scheme and provides domain size changes within a range of stability.

[12] Direct numerical simulation of non-Brownian particles: Smooth profile method
A. Hamid and R. Yamamoto

We performed the direct numerical simulations of non-Brownian sedimenting particles using a smooth profile method over a wide range of volume fraction from 0.01 to 0.4. We observed three distinct regimes with respect to volume fraction for velocity fluctuations, their relaxation times and diffusion anisotropy. Hydrodynamic velocity fluctuations in both directions scale as at low volume fraction, saturates at moderate volume fraction regime and decay sharply at high volume fraction. Unlike velocity fluctuations, vertical relaxation time relaxes as for the full range of volume fraction, whereas horizontal relaxation time decays as at low volume fractions, remain unchanged at moderate volume fraction and shows perplexing behavior at high volume fractions. Similarly, horizontal and vertical diffusion coefficients increases as at low volume fraction, whereas at moderate volume fraction vertical diffusion decays as , in contrast; horizontal diffusion remains unchanged. At higher volume fraction both diffusion coefficients decays sharply.

[25] A high order solver for the unbounded Poisson equation.
M. M. Hejlesen, J. T. Rasmussen, Ph. Chatelain, and J. H. Walther

In mesh-free particle methods a high order solution to the unbounded Poisson equation is usually achieved by constructing regularised integration kernels for the Biot-Savart law. Here the singular, point particles are regularised using smoothed particles to obtain an accurate solution with an order of convergence consistent with the moments conserved by the applied smoothing function. In the hybrid particle-mesh method of Hockney and Eastwood (HE) the particles are interpolated onto a regular mesh where the unbounded Poisson equation is solved by a discrete non-cyclic convolution of the mesh values and the integration kernel. In this work we show an implementation of high order regularised integration kernels in the HE algorithm for the unbounded Poisson equation to formally achieve an arbitrary high order convergence. We further present a quantitative study of the convergence rate to give further insight in the convergence of particle methods.

[28] Hydrodynamic correlations in multi-particle collision dynamics fluids
C.-C. Huang, G. Gompper, and R. G. Winkler

We analyze, both theoretically and numerically, the hydrodynamic correlations of a multiparticle collision dynamics (MPC) fluid, a particles based mesoscale simulation method. The fluid is characterized by its longitudinal and transverse velocity correlation functions in Fourier space and velocity autocorrelation functions in real space. Particular attention is paid to the role of sound, which leads to piecewise negative correlation functions. The minimal length scale in MPC for hydrodynamics is investigated. In addition, dilute polymer solutions are considered. We investigated the center-of-mass velocity-autocorrelation function of a polymer, unravelling the influence of sound. At long times, the correlation function exhibits a long-time tail decaying algebraically as t-1.5, which is independent of any polymer property and solely depends on the solvent viscosity.

[26] Shear induced instability of droplets in a colloidal dispersion
Hideki Kobayashi and Hiroshi Morita

We present numerical results for the breakup of a pair of colloidal particles enveloped by a droplet under shear flow. The smoothed profile method is used to accurately account for the hydrodynamic interactions between particles due to the host fluid. We observe that the critical capillary number, Cab, at which droplets breakup depends on a velocity ratio, E, defined as the ratio of the capillary velocity (that restores the droplet shape to a sphere) to the diffusive flux velocity in units of the particle radius, a. For E < 10, Cab is independent of E, as is consistent with the regime studied by Taylor. When E > 10, Cab behaves as Cab= 2/E. As a consequence, droplet breakup will occur when the time scale of droplet deformation is smaller than the diffusive time scale in units of a. We emphasize that the breakup of droplet dispersed particles is not only governed by a balance of forces. We find that time scale competition is one of the important contributing factor.

[3] Mesoscale simulations of viscoelastic fluids: Finite extensible Gaussian dumbbells
Bartosz Kowali and Roland G. Winkler

Complex fluids are often rather viscoelastic than purely viscose. Particular examples are polymer solutions and melts. To capture viscoelastic properties in computer simulations, an efficient and simple description for the fluid is desirable, particular in studies of embedded objects such as colloids, polymers, or cells. To bridge the length- and time-scale gap between particle and fluid degrees of freedom, mesoscale simulation techniques for simple fluids have been developed and shown to be extremely valuable in studies of soft matter systems. Incorporation of viscoelastic effects requires extensions of these methods.
We propose an extension of the multiparticle collision dynamics (MPC) approach [1] - a particle-based hydrodynamic simulation method - to viscoelastic fluids. Compared to the original formulation, two particles are combined into a dumbbell by a harmonic potential of zero bond length [2]. At equilibrium, the bond lengths are Gaussian distributed with zero mean. To reproduce phenomena like shear-thinning, a dumbbell has to be of finite length even under strong shear [3]. This requirement is captured by a shear-rate dependent force coefficient [3] such that the second moment of the bond-length distribution remains constant under shear. The advantage of the harmonic bond potential is that the known analytical solution of the equations of motion can be exploited in the streaming step, which is therefore only slightly more involved than that of the simple fluid.
We will present results of MPC simulations for such finite-extensible Gaussian dumbbells. In particular, we discuss the shear-dependent structural properties of the fluid, e.g., alignment of the dumbbells. Moreover, the dynamical properties, such as the non-equilibrium relaxation times, are determined. In addition, the rheological behavior is characterized, in particular we show that the system exhibits shear thinning. The numerical results are compared with theoretical calculations.

[1] G. Gompper, T. Ihle, D. M. Kroll, R. G. Winkler, Adv. Polym. Sci. 221, 1 (2009).
[2] Y.-G. Tao, I. O. Goetze, G. Gompper, J. Chem. Phys., 128, 144902 (2008).
[3] R. G. Winkler, J. Chem. Phys., 133, 164905 (2008).

[18] Numerical simulation of debris flows using a discrete element method coupled with a Lattice-Boltzmann fluid
A. Leonardi, F.K. Wittel, M. Mendoza, and H. J. Herrmann

Debris flows are dangerous natural hazards that occur in mountainous terrains after heavy rainfall, responsible for casualties and damages reported yearly worldwide. Their heterogeneous composition, with a viscoplastic fluid and the presence of a relevant granular solid phase, determines a complex behavior making them a challenging problem both for the physical description of the phenomenon and the design of effective protection measures. A numerical model is developed, taking into account the interaction between the two phases. A Discrete Element approach is used for the description of the solid phase, with a realistic particle size distribution, while the fluid phase is solved with a Lattice-Boltzmann method. The effect of shape on the rolling mechanism is taken into account with a simplified model. The numerical results provide insight into complex segregation, transportation and sedimentation phenomena, useful to understand and predict the run-out mechanism.

[4] Multiscale fluid dynamics simulation applied to micellar solution
T. Murashima, M. Toda, and T. Kawakatsu

We simulate wormlike micellar solution with multiscale simulation that consists of macroscopic fluid dynamics simulation and microscopic coarse-grained simulation. The microscopic coarse-grained simulation considers the Helfrich's bending energy that can describe the curvature of the wormlike micelles made of surfactants.

[13] Shear banding of model-stabilized colloidal suspensions in confined Couette flow
Jin Suk Myung and Kyung Hyun Ahn

Dynamics of model-stabilized colloidal suspensions in confined Couette flow was investigated by using the self-consistent particle simulation method (SC). In this method, the fluid-particle interaction is considered by combining the fluid motion analyzed by the finite element method (FEM) with the particle dynamics modeled by the Brownian dynamics (BD). Model-stabilized colloidal suspensions were subjected to shear flow in confined planar Couette geometry, and the flow behavior and microstructure were investigated. At low shear rates, the suspension in the confined Couette showed low slope region in the shear stress versus shear rate, which means that confinement-induced dynamics exist. As shear was applied, flow-induced ordered structure was formed firstly close to the wall, which induced wall slip. As the strain increased, shear banded velocity profile was developed through the gap, which implies that shear banding can be induced by the confinement. The effect of confinement was investigated by varying gap distance and volume fraction. This study clearly shows that the flow behaviors of colloidal suspensions are affected by the formation of flow-induced microstructures.

[19] Massively parallel molecular-continuum simulations with the macro-micro-coupling tool
Ph. Neumann and J. Harting

Efficient implementations of hybrid molecular-continuum flow solvers are required to allow for fast and massively parallel simulations of large complex systems. Several coupling strategies have been proposed over the last years for 2D/ 3D, time-dependent/ steady-state or compressible/ incompressible scenarios. Despite their different application areas, most of these schemes comprise the same or similar building blocks. Still, to the author's knowledge, no common implementation of these building blocks is available yet. In this contribution, the Macro-Micro-Coupling tool is presented which is meant to support developers in coupling mesh-based methods with molecular dynamics. It is written in C++ and supports two- and three-dimensional scenarios. Its design is reviewed and aspects for massively parallel coupled scenarios are addressed. Afterwards, scaling results are presented for a hybrid simulation which couples a molecular dynamics code to the Lattice Boltzmann application of the Peano framework.

[29] Mesoscale simulations of multi-domain protein dynamics
Simón Poblete, Roland G. Winkler, and Gerhard Gompper

The dynamics of the subdomains of a protein can play a fundamental role in its functionality. Such motions can be studied by techniques like neutron spin-echo spectroscopy, which has been shown to be able to resolve the time and length scales in the study of proteins like alcohol dehydrogenase (ADH) and phosphoglycerate kinase (PGK) [1,2]. On the other hand, computer simulations can provide a deeper insight of the protein dynamics, and be of great help for the interpretation of the experimental results.
Our system of interest is the merA protein. This enzyme, present in certain bacteria, is a fundamental part of the operon responsible for their resistance to toxic mercury compounds[3]. Since previous works have shown[4] that the different subdomains of this protein can have a specific role in the reduction of mercury, the study of their dynamics is demanding.
In this context, we have developed a highly coarse-grained model at a mesoscopic scale based on the basic geometry of the merA protein: two end groups attached to the protein core by short polymer chains. The simulations were performed using Multiparticle Collision Dynamics[5,6], a particle-based simulation approach able to capture the hydrodynamic interactions between the domains of our model.
We analyze some basic dynamic features of the protein domains, like diffusion coefficient and velocity autocorrelation functions. The effect of the terminal motion is studied by contrasting two models with flexible and rigid linkers respectively. The effective diffusion coefficient, obtained from the intermediate scattering function, can be directly compared to experimental data.

[1] R. Inoue et al., Biophys J. 99, 2309 (2010).
[2] R. Biehl et al., Soft Matter 7, 1299 (2011) .
[3] Barkay et al., FEMS Microbiology Reviews 27, 335 (2003).
[4] A. Johs et al., J. Mol. Biol. 413, 639 (2011).
[5] A. Malevanets and R. Kapral, J. Chem. Phys. 110, 8605 (1999).
[6] G. Gompper et al., Adv. Polym. Sci. 221, 1 (2009).

[20] Using Green’s function molecular dynamics for contact mechanics simulations
N. Prodanov and M. H. Müser

Numerical simulations of contact mechanics problems usually require fast algorithms and considerable computational resources because spatial features on many length scales must be taken into account. We provide a description of the efficient Green’s function molecular dynamics (GFMD) technique for solving static contact mechanics problems. The efficiency of the method is mainly provided by two factors. The first one is a transformation of a 3D problem into the corresponding 2D case. This allows one to consider a new equivalent model, which is several orders of magnitude smaller than the original one. The second factor is the use of the efficient FFT algorithm, which provides a fast convergence of calculations. The GFMD algorithm and its parallel implementation are described. Examples of applications of the technique to problems of contact mechanics of randomly rough surfaces, such as, percolation of the contact area and finite size effects in the interfacial stiffness are presented.

[5] Hybrid molecular – continuum simulation methods for polymers
M. Rahimi

In many simulations molecular details are required only in small spatial regions such as solid-fluid interfaces, while the continuum descriptions are still accurate in the remaining bulk regions. Therefore, it is desirable to develop a hybrid method to combine the efficiency of continuum mechanics and accuracy of MD simulation. We have developed a new hybrid simulation technique to couple a flexible particle domain to a continuum domain which is modelled by a finite element (FE) approach. The particle based simulations have been performed by molecular dynamics (MD) simulations in coarse grained (CG) representation. A staggered coupling procedure based on the Arlequin method has been chosen. The present MD-FE coupling method approximates the continuum as a static region while the MD particle space is treated as a dynamical ensemble. The information transfer between the MD and FE domain takes place in a coupling region where auxiliary particles, so-called anchor points, have been introduced as transmitter units. The anchor points are harmonically connected to MD particles in the coupling region. Forces and force gradients from the MD domain are transmitted to the FE domain and in turn the FE domain provides new anchor point positions for the MD domain. The present hybrid scheme has to be solved iteratively up to equilibrium. The capability of the new hybrid method has been quantified for an atactic polystyrene as well as polystyrene-silica nanocomposite sample in the linear elastic regime. The reasonable agreements are found between pure FE simulations and hybrid simulations for quantities such as reaction forces and Cauchy stress.

[6] Hydrodynamic simulation of bacteria swimming
Shang-Yik Reigh, Roland G. Winkler, and Gerhard Gompper

Bacteria such as Rhizobium meliloti can modulate the rotation speed of an individual motor, which they exploit to change their swimming direction [1]. Starting from a bundled state, where all flagella are synchronized, the decelerated flagellum gets out-of-phase and unbundles. To understand the fundamental mechanisms in this bacteria locomotion, we perform mesoscale hydrodynamic simulations using the multiparticle collision dynamics (MPC) method [2,3], which adequately captures the hydrodynamic interactions between the flagella and bridges the length- and time-scale gap between the fluid and bacteria. A flagellum is constructed by a sequence of mass points interacting by bond, bending, and torsional potentials. Such a model can efficiently be coupled to the MPC fluid. Results are presented for synchronization and bundle formation of flagella. The synchronization and bundling times are analysed in terms of the applied torque, the flagella separation, and the number of flagella [4]. Unbundling is studied in terms of the motor-torque difference between various flagella, and the resulting phase mismatch and tumbling-torque are determined.

[1] R. Scharf, J. Bacteriol. 184, 5979, (2002).
[2] R. Kapral, Adv. Chem. Phys. 140, 89, (2008).
[3] G. Gompper, T. Ihle, D. M. Kroll, and R. G. Winkler, Adv. Polym. Sci. 221, 1, (2009).
[4] S. Y. Reigh, R. G. Winkler, and G. Gompper, Soft Matter 8, 4363, (2012).

[21] Coarse Graining: From Particles to a Continuum
Jens Boberski, Alexander Ries, Lothar Brendel, and Dietrich E. Wolf

After a brief summary of the coarse graining formalism we present two applications with regard to particle continuum hybrid simulations of disordered systems. First we discuss the possibility to calculate local elastic fields and in a second part coarse graining close to interfaces between continuum and discrete particle system is discussed.

[14] Polymer mechanics at interfaces: molecular dynamics simulations and mapping to continuum approaches
M. Solar

The appropriate modeling of matter is becoming an increasingly unavoidable step in the predictive numerical simulations of phenomena like adhesion/adherence taking part of indentation (or scratch) tests and cracking or damage of polymers. In such modeling, the matter is considered from two points of view: a continuum and an atomistic point of view. This poster presentation illustrates former works on the rheology of polymer melts and films in the glassy and near the rubbery domain using two different methods: molecular dynamics (MD) and finite element (FE) simulations. Some comparisons between both methods are proposed and our results provided evidence in favor of using MD simulations to investigate the physics of polymer interfaces, since the volume elements studied were representative and thus contained enough information about the microstructure of the polymer model.

[7] Hydrodynamic interactions in periodic boundary conditions: Error control and parameter optimisation for the Rotne-Prager tensor
Lidia Westphal and Godehard Sutmann

Hydrodynamic interactions between solvated particles are considered on the level of the Rotne-Prager approximation, which is valid in the dilute regime. For the case of periodic boundary conditions a lattice summation was formulated by Beenaker. In the present work we derive expressions for error bounds as function of summation parameters. The prescribed error bound is verified by computing the sums with appropriate parameters for the tolerated errors and comparing with results which were computed close to numerical precision. An analytical model is proposed, which allows to predict the run-time behavior of real- and reciprocal sums as a function of the splitting parameter xi, which allows to determine a set of parameters (Rc,Kmax,xi), which minimizes the CPU time. The model is validated by run-time measurements of an Ewald sum implementation.

[8] Suspended end-functionalized rodlike colloids under shear flow
F. Taslimi, C. C. Huang, G. Gompper, and R. G. Winkler

Functionalized nano-particles are able to self-assemble into complex structures. Solutions of semiflexible rods with mutually attractive ends exhibit intriguing space-spanning scaffold structures at equilibrium within a certain range of densities and attraction strengths. To study the non-equilibrium properties of such structures, we perform non-equilibrium mesoscale hydrodynamic simulations, combining molecular dynamics simulations for the semiflexible rods with and the multiparticle collision dynamics approach for the fluid. We observe different structures depending on the shear rate. At high shear rates, the three-dimensional scaffold structure is destroyed and we obtain nematically aligned rods. For lower shear rates, shear-banding occurs. The scaffold structure breaks up into clusters of shear-aligned rods with a reminiscent scaffold transverse to the shear direction. Results will be presented for structural and rheological properties as function of shear rate and polymer length.

[22] Large time step discontinuous evolution Galerkin methods for multiscale geophysical flows
L. Yelash, M. Lukacova - Medvid’ova, G.Bispen

We present a new semi-implicit discontinuous Galerkin method, in which the flux is obtained by means of genuinely multidimensional evolution Galerkin operator constructed using the theory of bicharacteristics and splitted into a linear part (governing the acoustic and gravity waves) and to the rest nonlinear part that models advection waves. Comparisons with the standard one-dimensional Riemann solvers for the flux integration demonstrate better stability and accuracy for flows with small Mach and Froud numbers (which are of primary interest, e.g., for numerical weather prediction). The present research has been done in cooperation with A. Müller (Monterey, USA), V. Wirth (Mainz, Germany), K. Arun (Trivandrum, India), S. Noelle (Aachen, Germany) and supported by the German Research Foundation DFG under the grant LU 1470/2-2.

[23] Extending the length scale of accurate density-functional calculations
R. Zeller

Multiscale modelling, which connects continuous fields and atoms described quantum-mechanically by density-functional theory, is presently limited to the treatment of a few hundred atoms because the computing effort increases cubically with the number of atoms in the quantum-mechanical region. In my contribution I present a technique that reduces the computational complexity so that without compromises on the accuracy the required computing times increase only quadratically with the number of atoms. I demonstrate that also calculations with linearly scaling effort are possible if total energy errors are tolerated which are of the size of meV per atom. The technique has been implemented in our newly developed, massively parallel density-functional code ''KKRnano'' which is presently applied to calculate the electronic structure of metallic systems which contain several thousand atoms and thus cover length scales of several nanometers. Several examples of such calculations will presented.

[15] Hierarchical modeling of high-molecular-weight polymer melts by aid of soft sphere models
Guojie Zhang, Livia Moreira, Torsten Stühn, Kostas Ch. Daoulas, and Kurt Kremer

We propose a strategy for hierarchical modeling of high-molecular-weight polymer melts based on fluctuating soft-sphere models, for which an efficient grid-based Monte Carlo method has been developed. In the model, each subchain of Nb microscopic segments is represented by a soft sphere corresponding a cloud with Gaussian density distribution. The choice of Nb sets the resolution of the model, and the interactions are defined by Nb-dependent potentials. This intrinsic multiscale feature serves as a framework for a hierarchial modeling strategy. For fast equilibration of polymer melts, we start from a crude model (e.g. Nb =400). The equilibrated configurations obtained with the crude model are subsequently used as a start for simulations with the next degree of resolution (e.g. Nb = 200) after fine-graining. Only a short-time equilibration is required at this stage. The scheme is repeated until equilibrated configurations with high resolution (e.g. Nb = 25) are created. The latter allow reintroduction of microscopic details.