8th Project Access Call – Awarded Projects

Find below the allocations of the PRACE 8th Project Access Call.

SolitonCycLES: Exploring Soliton Cycles exhibiting abnormal combustion in a downsized spark-ignition engine using LES

Project leader: Dr Christian Angelberger, IFP Energies nouvelles, FRANCE
Research field: Engineering
Resource Awarded: 15 360 000 core hours on Curie FN @ GENCI@CEA, France;

Collaborators: Anthony Misdariis, CERFACS- France
Olivier Vermorel, CERFACS- France
Olivier Colin, IFP Energies nouvelles- France
Nicolas Iafrate, IFP Energies nouvelles- France
Lionel Martinez, IFP Energies nouvelles- France
Anthony Robert, IFP Energies nouvelles- France
Karine Truffin, IFP Energies nouvelles- France
Benoit Enaux, PSA Peugeot-Citroën- France
Frédéric Ravet, Renault- France

Abstract: SolitonCycLES proposed a highly innovative research approach for studying the hitherto poorly understood phenomena of rare and extreme (or soliton) 4-stroke cycles in downsized spark-ignited piston engines (SIE), exploiting the unique capacity of Large-Eddy Simulation (LES) and the unprecedented computing power offered in the frame of the PRACE programme.

Downsizing is a major pathway in order to improve the fuel efficiency and thus CO2 emissions from SIE. Its principle is to increase the specific load at which the engine is operated. The resulting higher fresh gases pressures and temperatures inside the cylinder however tend to favour the appearance of abnormal combustion, i.e. uncontrolled and unwanted auto-ignition of fresh gases. Combustion phasing is thus not controlled by spark-ignition, and can result in damages or even the destruction of the engine. In any case, the occurrence of abnormal combustion prevent fully exploiting the theoretical fuel efficiency benefits related to such concepts.One difficulty of studying soliton cycles with abnormal combustion is that their occurrence frequency is subject to important cyclic variations, and can reach values as low as once every hundred or even thousand cycles. They are therefore very complex to study experimentally, and even more so as a single occurrence of such an extreme cycle can destroy the engine.

In this context the objective of SolitonCycLES is to develop and validate a LES methodology based on the AVBP 3D CFD code, and able to quantify the probability of the appearance of soliton cycles with regard to different engine parameters as spark advance, trapped mass, intake air temperature and wall temperatures for 4 different baseline engine operation cases. In order to yield statistically relevant results, the effect of varying the 4 studied parameters for a number of cycles for the 4 baseline cases will be explored by LES, which requires perfuming more than one thousand LES. Each of these complex simulations requires an important number of cores and computing time in order to be accomplished in a time compatible with practical requirements. This is only possible if efficiently exploiting the unique computing resources offered by PRACE.

Work will consist of performing the important number of LES, of post-processing them in terms of probability for the appearance of soliton cycles with abnormal combustion and its sensitivity to the studied parameters, and of validating the achieved LES methodology by comparing its predictions with experimental findings. The acquired understanding will be exploited by supporting the formulation of a reduced model for abnormal combustion integrated into a system simulation code, and by calibrating it for the specific engine studied in the present project .

The LES methodology and reduced model developed within SolitonCycLES will be readily available to the project’s industrial partner for supporting the development of design and control strategies aimed at mastering abnormal soliton cycles. This will contribute reducing their negative impact, and thus contribute better exploiting the theoretical fuel efficiency benefits offered by downsized spark-ignition engines.

PentaGate-L – Tracing lipid-mediated gating and permeation of pentameric ligand-gated ion channels in atomic detail

Project leader: Dr Marc Baaden, Centre National de la Recherche Scientifique (CNRS) UMR5247 Institut des Biomolécules Max Mousseron, France
Research field: Biochemistry, Bioinformatics and Life sciences
Resource Awarded: 21 472 000 core hours on CURIE TN @ GENCI@CEA, France;

Collaborators: Jessica Jonquet, CNRS- France
Benoist Laurent, CNRS- France
Samuel Murail, CNRS- France

Abstract: PentaGate concerns the important family of pentameric ligand-gated ion channels (pLGICs), which mediates fast-millisecond synaptic transmission and modulation in the brain with prominent members such as the nicotinic acetylcholine receptor. pLGICs are responsive to allosteric modulation by a wide range of endogenous substances, such as lipids, neurosteroids, and endocanabinoids. Our main objective is to understand, at the highest possible structural and timescale resolution, the mechanisms of gating and permeation occurring in pLGICs using a combination of molecular dynamics (MD) simulations with experimental data. We focus on the bacterial homologue GLIC that provides several advantages : a very high resolution structure at 2.4 (EMBO J 2013), identified allosteric binding sites for anesthetics and alcohol and very good structural and functional characterization as well as unambiguous preliminary data on this system.

Using MD we were the first to carry out extensive, micro-second length simulations of GLICs gating transition from open to closed state. Very recently we elucidated a generic mechanism for ion permeation with novel features and our predictions were confirmed by experiment. The aim of the present PRACE project is to fully and quantitatively understand receptor gating, permeation and lipid modulation.Experimental validation will be carried out through a long-standing collaboration with two groups at the Pasteur Institute in Paris (cristallography, membrane reconstitution, electrophysiology, mutagenesis).

Task 1 : Membrane-modulated gating transition : how lipids stabilize pLGICs.As pointed out in the literature, pentameric channels are sensitive to their lipid environment. Here we will investigate such effects at the molecular level.

Task 2 : Dewetting : quantitative description of this key event in gating.Channel gating goes hand in hand with its dewetting. We will investigate dewetting by using Potential of Mean Force (PMF) calculations along two previously identified relevant reaction coordinates.

Task 3 : Permeation : quantitative characterization of pLGIC ion conductance.In preliminary work, we have been able to observe a close link between in silico predictions and single-channel conductance (EMBO 2013).Zachariae’s novel computational electrophysiology method can be used to predict IV curves and ion flux in the 60 to 400 mV range. We have successfully implemented this double bilayer simulation setup for GLIC and ran initial test simulations. The method requires PRACE resources.

Direct numerical simulation of partially premixed combustion in internal combustion engine relevant conditions

Project leader: Prof Xue-Song Bai, Lund University- Sweden
search field:
Resource Awarded: 26 000 000 core hours on SuperMUC @ GCS@LRZ, Germany;

Collaborators: Henning Carlsson, Lund University- Sweden
Vivianne Holmen, Lund University- Sweden
Siyuan Hu, Lund University- Sweden
Rickard Solsjo, Lund University- Sweden
Rixin Yu, Lund University- Sweden

Abstract: In the past decade, the European and world engine industry and research community have spent a great effort in developing clean combustion engines using the concept of fuel-lean mixture and low temperature combustion which offers great potential in reducing NOx (due to low temperature), soot and unburned hydrocarbon (due to excessive air), and meanwhile achieving high engine efficiency. One example is the well-known homogeneous charge compression ignition (HCCI) combustion engine, which operates with excessive air in the cylinder, and produces simultaneously low soot and NOx. However, HCCI combustion is found to be very sensitive to the flow and mixture conditions prior to the onset of auto-ignition. As a result, HCCI engine is rather difficult to control. At high load (with high temperature and high pressure) engine knock may occur with pressure waves in the cylinder interacting with the reaction fronts, leading to excessive noise and even damage on the cylinder and piston surface. At low load (with lower temperature and pressure) high level emissions of CO and unburned hydrocarbons may occur, which lowers the fuel efficiency and pollutes the environment. Recently, it has been demonstrated experimentally that with partially premixed charge compression ignition, also known as partially premixed combustion (PPC), smoother combustion can be achieved by managing the local fuel/air ratio (thereby the ignition delay time) in an overall lean charge.

There are several technical barriers in applying the PPC concept to practical engines running with overall fuel-lean mixture, low temperature combustion. For example, it is not known what the optimized partially premixed charge is for a desirable ignition, while at the same time maintaining low emissions. The main difficulty lies in the non-linear behavior of the dominating phenomena and the interaction among them (e.g. chemistry and turbulence). To develop an applicable PPC technology for IC-engine industry, improved understanding of the multiple scale physical and chemical process is necessary. Further, there is a need to develop computational models for simulating the process for the design where a large number of control parameters are to be investigated.

The goals of this project are to achieve improved understanding of the physical and chemical processes in overall fuel-lean PPC processes, and to generate reliable database for validating simulation models for analysis of the class of combustion problems. This shall lead to development of new strategies to achieve controllable low temperature combustion IC-engines, while maintaining high efficiency and low levels of emissions (soot, NOx, CO and unburned hydrocarbons). Direct numerical simulation (DNS) approach that employs detailed chemistry and transport properties will be used to study the mechanisms responsible for the onset of auto-ignition, and the structures and dynamics of the reaction front propagation in PPC conditions.

LEAC – Laser-plasma Electron Acceleration for CILEX

Project leader: Dr Arnaud Beck, CNRS, France
Research field: Fundamental Constituents of Matter
Resource Awarded: 80 000 000 core hours on FERMI @ CINECA, Italy; 2 500 000 core hours on Curie TN @ GENCI@CEA, France;

Collaborators: Jacob Trier Fredriksen, University of Copenhagen- Denmark
Arnd Specka, CNRS- France

Abstract: CILEX (Centre Interdisciplinaire Lumiere Extreme) is the Interdisciplinary Center on EXtreme Light. This facility located in the Paris area will host the APOLLON-10P laser, which will deliver pulses eventually as short as 15 fs at a still unreached instantaneous power of eventually 10PW. CILEX also hosts the associated experimental infrastructures which will offer the possibility to achieve scientific breakthroughs in various domains (electron and ion acceleration, X ray generation,high-field science). In particular, Laboratoire Leprince-Ringuet at Ecole Polytechnique, France, and Niels Bohr institute at the university of Copenhagen, Denmark, are collaborating on the laser-plasma acceleration of electrons. In this domain, the scientific goal of CILEX is twofold: explore laser wakefield acceleration at the highest possible powers, and realize a two stages acceleration experiment. CILEX operation will start in 2015 and be open to the international community shortly afterwards.

The first objective of the simulation campaign is to finalize the design of the dedicated experimental area of the CILEX facility and identify optimal laser and plasma parameters in terms of electron beams and X-ray sources quality for the initial campaign of acceleration experiments. Numerical simulations will provide guidance in the choice of several crucial design parameters.

The second objective is to evaluate the spectrum and phase-space of the generated relativistic electrons and photons so that instruments with proper detection ranges can be designed before the first laser shots. A well targeted detection range allows for a higher resolution so the overall quality of a large amount of experimental data will be impacted.

Finally, the numerical simulations will help to gain insight into the physics at play in these new regimes and will prove invaluable to help prepare the analysis of the experimental data produced by this future world-leader laser facility. These results will be published in international journals, and will be relevant to the scientific program of existing or upcoming short-pulse, petawatt-class laser facilities such as Texas Petawatt Laser (USA) or Berkley Lab Laser Accelerator (USA). It is also an important step towards even more ambitious future experiments on the ELI-Beamlines project in Europe.

To do so, two PIC codes are going to be used. First, Calder-Circ, a robust and well demonstrated reduced code that will allow a quick exploration. And then, Photon-Plasma, a very modern, fully 3D, code running in a hybrid openMP – MPI mode, to get accurate, quantitative results in the regions of interest previously highlighted by Calder-Circ.

Photon-Plasma has already proven an excellent scalability on IBM Blue Gene/Q using the JUQUEEN system. It uses an original high order explicit scheme with filtering, specifically designed for laser propagation in a plasma. The heaviest 3D runs of our campaign will be run on the same architecture provided by the Fermi system.

Calder-Circ is MPI only, has a small memory footprint and is optimized for intel processors. CILEX cases have already been run with it on Curie’s thin nodes successfully for one million cpu hours previously granted by GENCI.

JOREK_ITERNon-linear simulations of MHD instabilities and methods of their control in realistic tokamaks plasmas and in ITER.

Project leader: Dr Marina Becoulet, CEA/IRFM, France
Research field: Fundamental Constituents of Matter
Resource Awarded: 12 200 000 core hours on Curie TN @ GENCI@CEA, France;

Collaborators: Alexandre Fil, CEA/IRFM- France
Eric Nardon, CEA/IRFM- France
Virginie Grandgirard, CEA/IRFM- France
Jorge Morales, CEA/IRFM- France
Shimpei Futatani, Ecole Centrale de Lyon- France
Guillaume Latu, IRFM/CEA- France
Chantal Passeron, IRFM/CEA- France
François Orain, IRFM/CEA- France
Guilhem Dif-Pradalier, IRFM/SCCP- France
Guido Huismans, ITER- France
Feng Liu, ITER- France
Stanislas Pamela, CCFE- United Kingdom

Abstract: The project aims to do a significant progress in understanding of full dynamics of relevant for fusion devises (tokamaks) MHD instabilities such as Edge Localised Modes (ELMs) and plasma disruptions with the goal to develop, predict and optimise the methods of active control/mitigate the impact of MHD instabilities such as Resonant Magnetic Perturbations (RMPs) and pellets for ELMs control and Massive Gas Injection (MGI) for disruptions mitigation [Becoulet2003], [PIPB2007].These fast transient MHD events can strongly limit fusion plasma performance, moreover, if not controlled, fast (10-4-10-3s) transients represent a danger for Plasma Facing Components (PFC) due to large heat and particle loads [Loarte2006]. The goals of the project are to improve understanding of the related physics and propose possible new strategies to improve effectiveness of ELMs and disruptions control techniques foreseen for future fusuin reactors. The related intensive experimental and theoretical study of ELMs and disruptions and some promising methods of their control have a great importance for International Thermonuclear Experimental Reactor (ITER), large international project involving partnership of EU, China, Russia, USA, Japan, India and Korea, and which is already under construction in Cadarache, France. Our computer project is based on the non-linear MHD theory and modelling using large parallelised numerical code JOREK [Huysmans2007-2013], [Czarny2008] demanding large memory (essentially needed for a matrix solver) and time resources to treat different time scales of the problem ( from Alfven time 10-7s to a plasma energy confinement time, few seconds, and current diffusion – resistive time >100s ). The present project is proposed in order to progress in urgently needed solutions for ITER.

The CURIE computer we aim here is particularly adapted for our code, it was actively used for production on similar computer HELIOS (Japan) and started already to be used on CURIE in 2013. The recent JOREK results were largely presented on the International Conferences and published in related papers. All subjects we propose continue in many aspects our previous studies but on the highest level of physics implemented in the code JOREK also new subjects will be treated. At the same time numerical optimisation of the code continues and in particular the further optimisation of the memory distribution and computational time especially needed in in multi-harmonics non-linear simulations with fine space and time resolution. The project is a part of our much larger collaboration on MHD transients modelling based on JOREK code which includes also IPP Garching, Germany, IPP Prague, Check Republic and DIFFER-Netherland, INRIA of Nice and Bordeaux Universities, France.

The proposed work scope is organized in five main tasks:

  1. Modelling of ELMs in existing machines and extrapolation to ITER, estimation of heat and particle loads due to ELMs, impact on PFC.
  2. Nonlinear MHD simulations of QH mode plasmas and possible applicability in ITER.
  3. Modelling of interaction of RMPs with rotating plasma and ELMs, find the regimes with ELM suppression in ITER.
  4. Non-linear MHD simulation of ELM triggering by pellet injection for mitigation and control
  5. Disruptions dynamics and disruptions mitigation by MGI.

EGOIST – Endogenic oil synthesis in the deep Earth interior: ab initio molecular dynamic simulation

Project leader: Prof Anatoly Belonoshko, The Royal Institute of Technology (KTH), Sweden
Research field: Earth System Sciences
Resource Awarded: 50 000 000 core hours on MareNostrum @ BSC, Spain;

Collaborators: Pavel Gavryushkin, V.S. Sobolev Institute of Geology and Mineralogy, Siberian Branch, RAS- Russia
Konstantin Litasov, V.S. Sobolev Institute of Geology and Mineralogy, Siberian Branch, RAS- Russia
Tymofiy Lukinov, The Royal Institute of Technology (KTH)- Sweden

Abstract: Physics and chemistry of C-O-H fluids at high pressures and temperatures of Earth interior is important in several applications. First, the thermodynamics of these fluids is needed to describe properties of the Earth interior which, in turn, might be important for predicting seismic events. Second, estimating the balance of CO2 between atmosphere and Earth interior is impossible without detailed knowledge of thermodynamics of C-O-H fluids. Third, there are indications from experiment that chemical reactions in C-O-H fluids at high P and T might lead to a synthesis of hydrocarbons and heavy alkanes, providing a possibility for formation of oil deposits at the relevant depth. Experimental difficulties in studying C-O-H fluids at high PT are numerous – for example, diffusion of H2 is one of them. Therefore, a theoretical approach is a valuable asset in these studies. Presently, we can compute phase and chemical equilibrium using density functional theory and molecular dynamics. When combined together, they represent a powerful tool. We shall study various components in C-O-H system, systematically collecting data on their equations of state to use for computing, in turn, their Gibbs free energy. Minimization of the Gibbs free energy allows to determine chemical composition at equilibrium as soon as thermodynamics of all possible components is available. DFT based MD is similar in a way to a high PT experiment, yet without experimental problems. While theoretical approach has its own limitations, they are well known and understood. Thus, we expect that the acquired knowledge of thermodynamics, phase and chemical equilibrium in C-O-H system will be highly reliable. In a way, our simulations are similar to a real experiment – we shall place a certain composition into an experimental box and apply certain pressure and temperature. That will allow us to observe the chemical composition that forms in the ’experimental’ chamber. We expect to observe chemical reactions that lead to formation of hydrocarbons and alkanes and describe the range of pressures, temperatures and compositions where these reactions occur. This, in turn, might enable an educated search for the regions in the Earth interior that might contain the products of this reaction. As a spin-off, the acquired knowledge will help us to solve the problem of excessive CO2 as well as to understand the interior of icy planets and sattelites of giant planets.

Amino-acid anions in organic compounds: charting the boundary of room temperature ionic liquids.

Project leader: Dr Enrico Bodo, University of Rome “Sapienza”, Italy
Research field: Chemistry
Resource Awarded: 8 166 667 core hours on Curie TN @ GENCI@CEA, France;

Collaborators: Antonio Benedetto, University College Dublin (UCD)- Ireland
Lorenzo Gontrani, CNR- Italy
Pietro Ballone, NTNU Norwegian University of Science and Technology- Norway

Abstract: Ionic liquids (ILs) are salts made by complex, sterically mismatched molecular ions which possess a low melting point owing to the fact that the electrostatic interactions are weakened and lattice formation frustrated by geometric effects. In contrast to traditional organic solvents, ILs possess negligible flammability and volatility and represent a new class of “green” solvents that are inherently safer and more environmentally friendly than conventional solvents. Their use as solvents in chemical processes is in line with the topics covered in the Horizon 2020, as the development of green and sustainable procedures is one of its objectives.

The combination of amino acids in their deprotonated and thus anionic form with choline and phosphocholine cations gives origin to a novel and potentially important class of organic ionic liquids. Preliminary results have revealed intriguing structural motives as well as regular patterns in the properties that could guide the experimental investigation of these compounds. Analysis of the ab-initio data highlights that proton trasfer in these liquids may be possible to such an extent that was not possible with other ionic liquids (including protic ones).

The main focus of this project is a theoretical description of a series of liquids composed by amino-acid anions (alanine [Ala], serine [Ser], cysteine [Cys], valine [Val], phenylalanine [Phe], aspartic acid [Asp], asparagine [Asn], and histidine [His]) combined with choline and phosphocoline cations. The description of the bulk phase will be performed using Car-Parrinello molecular dynamics (CPMD) in order to unravel their nanoscopic structure and the mechanisms of possible proton transfer processes that can affect their macroscopic properties and their use a new solvents with tunable ionicity. The transfer of a proton between the Choline cation to the amino-acid anion has the net result of neutralizing the ionic couple involved. Ideally, we can imagine that a collective, complete transfer of protons from the cations to the anions in the liquid phase could lead to a transition from an ionic liquid to a traditional polar liquid. This kind of chemical transition may be triggered by temperature pressure changes or by mixing different compounds. This study has the potential of opening an unexpected and advantageous route to the synthesis, characterization and design of new, unconventional materials whose chemical/solvation properties can be tuned upon changes of their molecular structure. In particular one could be able to control the degree of autoionization and ultimately the solvation properties of a fluid by engineering the right molecular structure.

Given the time limitation of ab-initio approaches, a variety of constrained simulations will be used to investigate free energy barriers opposing proton transfer events, entering the definition of kinetic models used to extend the simulation results to time scales inaccessible to ab-initio methods. The first quality check of the computation will be provided by a comparison of the liquid structure with X-ray diffraction experiments that are currently under development at the PI institution. Additional checks will be provided by comparison with the results of conductivity measurements, dielectric spectroscopy and neutron scattering experiments that are being planned.

Rare structures in the Lyman-alpha forest: bridging the gap between small and large scales.

Project leader: Dr James Bolton, University of Nottingham, United Kingdom
Research field: Universe Sciences
Resource Awarded: 15 000 000 core hours on Curie TN @ GENCI@CEA, France;

Collaborators: John Regan, University of Helsinki- Finland
Matteo Viel, INAF- Italy
Martin Haehnelt, University of Cambridge- United Kingdom
Ewald Puchwein, University of Cambridge- United Kingdom
Debora Sijacki, University of Cambridge- United Kingdom
Avery Meiksin, University of Edinburgh- United Kingdom
Frazer Pearce, University of Nottingham- United Kingdom

Abstract: The intergalactic medium (IGM) is the rarefied material which spans the vast distances between galaxies in the Universe. The IGM therefore straddles the interface between studies of galaxy formation and the evolution of large scale structure, and its observable properties are closely intertwined with both processes. One of the key observational probes of the IGM is the Lyman-alpha forest of hydrogen absorption lines observed in the spectra of distant quasars. Careful comparison between detailed hydrodynamical simulations of the Lyman-alpha forest and high resolution, high signal-to-noise echellete spectra have yielded valuable insights into how cold dark matter is, the epoch of reionisation and the interplay between galaxies and gas in the early Universe. A key limitation of the existing numerical models, however, is their narrow dynamic range. This translates into rather small simulation volumes due to the requirement of resolving the Jeans scale in the IGM. Highly resolved simulations are essential for quantitative comparison to the available high-quality, high-resolution observational data. Large scale variations and rare objects, such as massive dark matter haloes and deep voids, are therefore not well captured in existing Lyman-alpha forest simulations. This significantly limits the utility of these models when confronted with observational data, and requires large corrections to be applied to the simulation results. This PRACE project aims to alleviate these issues, by bridging the important gap between small and large scales with a suite of the highest resolution Lyman-alpha forest simulations performed to date within large volumes.

FREEZEOUT – Heavy ion phenomenology from lattice simulations

Project leader: Dr Szabolcs Borsanyi, Bergische Universität Wuppertal, Germany
Research field: Fundamental Constituents of Matter
Resource Awarded: 73 200 000 core hours on FERMI @ CINECA, Italy; 18 000 000 core hours on JUQUEEN @ GCS@Jülich, Germany;

Collaborators: Zoltan Fodor, Bergische Universität Wuppertal- Germany
Stefan Krieg, Forschungszentrum Juelich- Germany
Kalmán Szabó, Unversität Regensburg- Germany
Sándor Katz, Eötvos University- Hungary

Abstract: We are building theoretical background to the interpretation ofheavy ion experiments at the Relativistic Heavy Ion Collider (Brookhaven,USA). We perform lattice simulations to Quantum Chromodynamics (QCD) atfinite temperature. We calculate higher moments of the net charge andbaryon distribution at the instant of last inelastic scattering (chemicalfreeze-out). We generalize earlier lattice results towards finitedensities and study the sensitivity of the distribution of net conservedcharges to the collision energy. To circumvent the sign-problem thatprohibits simulations at using finite chemical potentials we introduceimaginary chemical potentials. Through analytical continuation a freeze-outcurve can be determined from the matching of lattice data to the experiment.This method bypa
sses the model assumptions of today’s estimates.

STiMulUs – Lagrangian Space-Time Methods for Multi-Fluid Problems on Unstructured Meshes

Project leader: Ing Walter Boscheri, University of Trento, Italy
Research field: Engineering
Resource Awarded: 8 000 000 core hours on SuperMUC @ GCS@LRZ, Germany;

Collaborators: Prof Michael Dumbser, University of Trento

Abstract: This project is inserted in the framework of the STiMulUs project, which has begun in 2011, when the PI (Prof. Dr.-Ing. Dumbser) won an ERC Starting Grant of 60 months duration. STiMulUs main task is the development of new robust, efficient and high order accurate numerical algorithms for the solution of time dependent partial differential equations (PDE) in the context of non-ideal magnetized multi-fluid plasma flows with thermal radiation. It will consider both, high order unstructured Eulerian methods on fixed grids as well as high order unstructured Lagrangian schemes on moving meshes, to reduce numerical diffusion at material interfaces. This project and our request of computational resources will focus on the Lagrangian part of STiMulUs, whose growth started last year when the PI et al. were the first who developed a one-dimensional Lagrangian arbitrary high-order one step WENO finite volume scheme for stiff hyperbolic balance laws. The work is currently in progress and we have already ex-tended the previous work to two space dimensions, using unstructured triangular meshes . A very challenging field of application for non-ideal multi-fluid plasma flows with thermal radiation is nuclear fusion, in particular inertial confinement fusion (ICF).). The science is very interested in this topic because of the shortage of energy on the Earth. In fact the world human population is rapidly growing and as a natural consequence also its need for energy. For these reasons and due to their better instabilities resolution, our new Lagrangian schemes would be suitable to study those phenomena occurring during ICF. In this research project we therefore want to carry out very important basic research on that topic and develop completely new, very high order accurate numerical algorithms in space and time for the solution of time dependent partial differential equations (PDE) with stiff source terms on general unstructured meshes that govern non-ideal magnetized multi-fluid plasma flows with thermal radiation, occurring before the onset of the nuclear fusion process. We rely on high order schemes to resolve very well and with only little numerical diffusion also the fine details of the flow that are crucial for this kind of applications. The highly accurate next-generation mathematical tools emerging from the STiMulUs project may lead to completely new fluid-mechanical key insights in ICF flows that can subsequently be used by physicists and engineers to succeed with the next ICF experiments, thus providing modern civilization with clean energy in the future. The importance of our research topic for our society is underlined by the recent international discussions on the accelerating global climate change, mainly caused by modern civilization and its increasing need for energy.

The Global Turbulent Sun

Project leader: Dr Allan Sacha Brun, CEA-Saclay, France
Research field: Universe Sciences
Resource Awarded: 25 000 000 core hours on Curie TN @ GENCI@CEA, France;

Collaborators: Lucie Alvan, CEA-Saclay- France
Antoine Strugarek, CEA-Saclay- France

Abstract: The one-year Global Turbulent Sun project aims at modeling on massively parallel computers in a self-consistent and 3-D spherical (global) geometry the complex time dependent and nonlinear processes operating in the Sun. We wish to characterize global solar turbulent convection and magnetic field generation via dynamo action by coupling the convective envelope to a subadiabatic layer below. This dynamical, thermal and magnetic coupling is key to organize at the interface between the convection and radiation zones large-scale non-axisymmetric magnetic wreaths that can yield cyclic activity and become unstable and launch magnetic loops to the surface. Having simultaneously very high resolution, improve boundary conditions and low diffusivities is likely to result in the first self-consistent solar global ’spot’ dynamo.

PlasTitZir – Plasticity in Titanium and Zirconium

Project leader: Dr Emmanuel Clouet, CEA, France
Research field: Chemistry
Resource Awarded: 12 000 000 core hours on CURIE TN @ GENCI@CEA, France;

Collaborators: Nermine Chaari, CEA- France

Abstract: This proposal aims to build a physical sound model of plasticity in zirconium and titanium, by looking more precisely at the influence of oxygen solute atoms on the plastic behavior. Both metals have similar properties arising from their hexagonal compact (hcp) crystallography and from their alike electronic structure. In particular, their plastic behavior is strongly influenced by the interaction of screw dislocations with oxygen impurities. At a macroscopic scale, a small O addition to either Zr or Ti leads to an important hardening, with a plastic flow controlled by thermal activation instead of an athermal regimes in pure metals. This hardening cannot be associated with a simple elastic interaction between dislocations and impurities, but seems to be induced by a change of dislocation properties trough a modification of their core structure by O atoms. The purpose of this project is therefore to understand how O impurities modify dislocation properties in pure Ti and Zr, so as to model then the hardening induced by O addition.

There is no direct experiment able to image a modification of the screw dislocation core induced by O atoms. On the other hand, atomistic simulations are the good tool to study dislocation cores. In Ti and Zr, dislocations glide in the prism planes of the hcp crystal. The relative ease of prismatic compared to basal glide has been shown to be linked to the ratio of the corresponding stacking fault energies, which in turn is controlled by the filling of the valence d band which induces an angular dependence of the atomic bonding that cannot be neglected. An important consequence for the modeling at an atomic scale of plasticity in Ti and Zr is that one cannot rely on simple empirical potentials, like EAM potentials, but one has to take full account of the electronic structure. We therefore propose to use ab initio calculations based on the density functional theory to study the interaction of screw dislocations with O impurities in both Zr and Ti. This project will provide quantitative data for the modeling then of dislocation mobility in presence of impurities. Ab initio calculations with both an O interstitial and a screw dislocation will be performed to see how the impurity modifies the core structure of the dislocation. A result of these calculations will also be the interaction energies,
which will be used then to build a thermodynamic modeling of the impurity segregation on the dislocation.

A deep understanding of the interaction between screw dislocations and O atoms will be gained from these atomistic simulations. This will allow us to understand how O atoms modify plasticity in both Ti and Zr, and to build a physical model of their plastic behavior. Such a model could be then extended to industrial alloys by incorporating contributions of other alloying elements.

AMOGLASS – Amorphous order in glassy silica

Project leader: Dr Daniele Coslovich, Laboratoire Charles Coulomb, France
Research field: Chemistry
Resource Awarded: 150 000 core hours on MareNostrum @ BSC, Spain;

Collaborators: Walter Kob, Laboratoire Charles Coulomb-France

Abstract: Glasses are omnipresent in our daily life: Oxide glasses are found in window panels, vessels, bottles and other containers. Most polymeric materials are glasses and also other materials such as metals and a large variety of complex fluids (emulsions, pastes, foams) can be produced in a glassy state. Despite this ubiquity, there is at present no fundamental understanding of the glass state. This fact can be illustrated easily by considering a glass-forming liquid, e.g. silica: At high temperatures it flows easily, but if it is cooled its viscosity raises by 10-15 decades and becomes so large that the liquid remains trapped in a dynamically arrested state and becomes a glass. Despite a large number of experimental, theoretical, and simulation studies, the physical origins for this dramatic slowing down of the dynamics remain poorly understood. Note that this freezing in seems to have little to do with a change of the structure, i.e. from a structural point of view a glass and a liquid are practically identical.

Motivated by recent theoretical developments, we want to use a novel approach to test the existence of “amorphous order”. This more exotic type of structural order can be probed by measuring how the static structure of glass-forming liquids reacts when freezing the positions of a set of particles in the system. This novel approach will allow us to investigate whether there is an equilibrium phase transition associated to the development long range amorphous order, as predicted theoretically. The presence of such order may constitute the fundamental difference between liquids and glasses and provide a rationale for the dramatic increase of viscosity.

The liquid we will consider is a simple but quite realistic model for silica, one of the most important known glass-formers. The simulation technique is a combination of molecular dynamics and parallel tempering. This latter Monte Carlo method has been shown to be highly efficient to study the equilibrium properties of glass-forming systems even at low temperatures, hence it will allow us to probe the putative phase transition. To tackle these ambitious issues, we will resort to state-of-art molecular dynamics simulations running entirely on GPUs, which are particularly well suited for this kind of investigations.

Our simulations will allow to make an important step forward in our understanding what a glass really is and also motivate researchers in the field to carry out analogous experiments in the laboratory.

CoLGIC: Conformational changes and allosteric regulation inligand-gated ion channels

Project leader: Prof Francesco Luigi Gervasio, University College London, United Kingdom
Research field: Biochemistry, Bioinformatics and Life sciences
Resource Awarded: 18 340 000 core hours on Curie TN @ GENCI@CEA, France;

Collaborators: Giorgio Saladino, University College London-United Kingdom
Lucia Sivilotti, University College London-United Kingdom

Abstract: Ligand-gated ion channels mediate communications in both the central and peripheral nervous systems, converting a rapid increase in a neurotransmitter concentration into an electric signal. Anion-selective channels, are connected to fast inhibitory neurotransmission and are crucial to maintain the basal neuronal activity.

The inhibition of ligand-gated ion channels is used for the treatment of several disorders, ranging from nicotine addiction to Alzheimer’s disease, and these channels are themain target of very successful and common drugs, including benzodiazepines and barbiturates. Unfortunately, the lack of structural information about anion-selective channels has limited the understanding of the complex mechanism behind ion permeation and gating, slowing down the development of new therapies. Here we’ll use state of the art molecular dynamics simulations and free energy calculations to understand in great details these mechanisms, potentially opening the avenue to a new generation of more potent and less toxic drugs.

STARZOOM – Zooming in on Star Formation

Project leader: Dr Troels Haugboelle, University of Copenhagen, Denmark
Research field: Universe Sciences
Resource Awarded: 4 000 000 core hours on Curie FN @ GENCI@CEA, France; 6 000 000 core hours on Curie TN @ GENCI@CEA, France;

Collaborators: Christian Brinch, University of Copenhagen-Denmark
Soeren Frimann, University of Copenhagen-Denmark
Jes Kristian Joergensen, University of Copenhagen-Denmark
Aake Nordlund, University of Copenhagen-Denmark
Oliver Lothar Gressel, University of Copenhagen-Denmark
Colin McNally, University of Copenhagen-Denmark
Gareth Murphy, University of Copenhagen-Denmark
Martin Pessah, University of Copenhagen-Denmark
Paolo Padoan, University of Barcelona- Spain

Abstract: Newborn stars, surrounded by centrifugally supported discs of gas and dust, reside in the central regions of hot cores, which are embedded in colder, extended envelopes. Above the discs and close to the star, outflows are launched in the form of winds and jets. These systems exist for a few million years during and after the birth of the star. In the early stages of this process, the envelope collapses under its own gravity, and a disc is quickly formed. By removing excess angular momentum and potential energy, this accretion disc functions as a conduit, which allows most of the gas and dust to either accrete onto the star or be lost in the outflows, while leaving a small fraction of the mass in the form of planets. Detailed modelling of the properties of proto-stellar systems is thus a prerequisite for understanding planet formation. Because of the small scale of the inner proto-planetary disc, relative to the distance of even the nearest star forming regions, very few detailed – spatially resolved – observations have been available in the past. The Herschel satellite and the new sub-millimeter facility ALMA are revolutionizing observational star formation. In particular, ALMA is currently providing many new observations of proto-stellar systems, and a more complex picture is emerging with e.g. warp
ed discs containing intricate structures, such as large-scale dust devils, and many signs of newborn planets perturbing the gas and dust.

Unravelling the physical processes at play is crucially important and requires modelling, and understanding, an intricate interplay between large-scale environmental factors, which regulate the supply of mass, angular momentum, and magnetic flux to the forming stars, and small-scale processes close to the star, which control the evolution and dynamics in the inner part of the envelope and proto-planetary disc. This proposal will encompass these disparate scales using extremely deep adaptive mesh refinement simulations, reaching a factor of a billion in linear resolution compared to the outer scales of the simulated domain. These unprecedented capabilities are complemented by a unique approach that we have developed to provide boundary conditions by embedding our model in to some of the largest models ever made of star forming regions.

Using the CURIE supercomputer we will perform, for the first time ever, a systematic study that bridges the gap between molecular cloud scales, where magnetic fields are anchored and the initial and boundary conditions for proto-stellar accretion are set, and the much smaller disc and jet scales, which determine the ultimate evolution of proto-stellar systems. Because of the unique combination of methods and resources, this study will have a major impact in our global understanding of how stars and planets form.

FENICS-HPC – High performance adaptive finite element methods for turbulent flow and multiphysics with applications to aerodynamics, aeroacoustics, biomedicine and geophysics

Project leader: Dr Johan Hoffman, KTH Royal Institute of Technology, Sweden
Research field: Mathematics and Computer Sciences
Resource Awarded: 10 000 000 core hours on Hermit @ GCS@HLRS, Germany; 10 000 000 core hours on SuperMUC @ GCS@LRZ, Germany;

Collaborators: Cem Degirmenci, KTH Royal Institute of Technology-Sweden
Johan Jansson, KTH Royal Institute of Technology-Sweden
Niclas Jansson, KTH Royal Institute of Technology-Sweden
Aurelien Larcher, KTH Royal Institute of Technology-Sweden

Abstract: This project concerns the development of parallel computationalmethods for solving turbulent fluid flow problems with focus onindustrial applications, such as the aerodynamics of a full aircraftat realistic flight conditions, the sound generated by the turbulentflow past the aircraft during landing and takeoff, the blood flowinside a human heart and geophysical flows. The massive computationalcost for resolving all turbulent scales in such problems makes DirectNumerical Simulation of the underlying Navier-Stokes equationsimpossible. Instead, various approaches based on partial resolution ofthe flow have been developed, such as Reynolds Averaged Navier-Stokesequations or Large-Eddy simulation (LES). For these methods newquestions arise: what is the accuracy of the approximation, how finescales have to be resolved, and what are the proper boundaryconditions? To answer such questions, a number of challenges have tobe addressed simultaneously in the fields of fluid mechanics,mathematics, numerical analysis and HPC.

The main focus of the research at The Computational TechnologyLaboratory (CTL) is the development of high performance, parallel,adaptive algorithms for FEM modeling of turbulent flows andmultiphysics, including fluid-structure interaction andaeroacoustics. The adaptive finite element method G2 has beendeveloped over the past 10 years for time-resolved simulations ofturbulent flows and it works as an implicit LES method with a residualbased subgrid model that accounts for the unresolved scales. With meshadaptivity based on “a posteriori” error estimates, efficientparallelization, and the use of unstructured meshes, G2 constitutes apowerful tool in Computational Fluid Dynamics, which can be used tosolve time dependent problems efficiently. Of particular interest isthe error estimation framework of G2 and, currently, we work to extendthe framework to include uncertainty quantification of data andmodeling parameters. Within our group, there are a number of projectsin various applications areas, where the new adaptive algorithms arebeing used and developed.

These areas include aerodynamics, aeroacoustics, biomedicine,geophysics and FSI. In the past 3 years, we have obtained significant results in the development of G2. These include: the implementation of a hybrid MPI+PGAS linear algebra backend, which enhanced the performance of the code for larger core counts as compared to the previous MPI implementation; the successful computation of the flow past an extremely complex noselanding gear geometry and the flow past a high-lift device, both ascontributions to the second workshop on Benchmark problems forAirframe Noise Computations, BANC-II, proposed and developed byNASA. For the following year, we plan to follow-up these contributions with more detailed, larger computations. In 2013 we were granted the EU FP7 project: “Extensive UNIfied-domain SimulatiON of the human voice” (EUNISON)for the simulation of the human voice based on our framework. We have also successfully participated in a NASA/Boeing challenge/workshop on simulation of a full aircraft (HiLiftPW-2), and we are invited to submit a paper to the AIAA SciTech 2014 conference based on our results. Our adaptive results were specifically highlighted in the summary by the organizers.

Time-resolved evolution of vorticity and momentum cascades in statistically stationary homogeneous shear turbulence

Project leader: Prof Javier Jimenez, Universidad Politecnica Madrid, Spain
Research field: Engineering
Resource Awarded: 20 000 000 core hours on JUQUEEN @ GCS@Jülich, Germany;

Collaborators: Siwei Dong, Universidad Politecnica Madrid-Spain
Atsushi Sekimoto, Universidad Politecnica Madrid-Spain

Abstract: The project seeks to answer two questions about shear-dominated turbulence: a) do all shear-induced turbulent flows with linearly stable mean velocity profiles share common mechanisms for momentum transfer and energy generation? (Linearly unstable profiles are known to behave differently). b) Do the coherent structures in turbulence undergo a cascade as proposed by Richardson (1920), not only as a conceptual model but as an actual physical process? Similar questions have been studied by our group in recent years for wall-bounded turbulence (WT), mainly in turbulent channels. That work involved simulating the flow, isolating individual vortex clusters [3] and structures of the momentum transfer (ejections and sweeps) [6], storing them often enough to be considered as temporally resolved, and organizing their evolution and interactions in time and space [5]. We have paid especial attention to the logarithmic layer of wall-bounded turbulence [7], which is a region of particular theoretical and practical interest. We have been able to carry that work to Reynolds numbers high enough for a cascade to be present.

It turns out that the channel dynamics is controlled by structures that are attached to the wall, but which scale well with the local shear, and it is unclear which of the two effects is more important. We intend to separate the effect of the shear from that of the wall
by applying the same techniques to the canonical case of statistically-stationary homogeneous sheared turbulence in a finite simulation box (HST), which is known to burst intermittently [1,2] in a very similar manner to WT [4,7,8]. Preliminary work at low Reynolds numbers has allowed us to determine the ideal geometry for an HST simulation box to mimic the statistical properties of the logarithmic layer, and to confirm that structures broadly similar to those of WT exist in HST. The proposal is to extend the HST work to a Reynolds number similar to that available for channels (Re_lambda=200), and to compare the results in both flows.

If successful, the results would simplify considerably our understanding of the dynamics of shear flows in general, but they would also point the way to simpler boundary conditions for large-eddy simulations of wall-bounded flows. The practical application of those simulations is limited at present by the resolution requirements near the wall, but if it could be confirmed that the details of the wall are unimportant, simpler boundary conditions could be designed. In fact, large-eddy simulations of wall-less channels have been recently performed in our group with promising results [9].

TWISTER – Large-Eddy Simulation of tornado formation by a supercell.

Project leader: Prof  dr Harmen Jonker, Delft University of Technology, The Netherlands
Research field: Earth System Sciences
Resource Awarded: 500 000 core hours on Curie H @ GENCI@CEA, France;

Collaborators: Jerome Schalkwijk, Delft University of Technology-The Netherlands

Abstract: This project proposes to perform high resolution simulations of the formation process of a tornado in close conjunction to the supercell (thunderstorm) that causes the tornado to form. The huge difference in scales between tornado and supercell makes simultaneous simulation of both phenomena extremely challenging, but using the vast computational acceleration offered by a multi-GPU system, such simulations have now come within reach.

The goal of the proposed extreme computations is

  1. to learn more about the inner (thermo-)dynamical structure of tornadoes;
  2. to increase our fundamental understanding of tornado formation in relation to the supercell that breeds it;
  3. to demonstrate to the atmospheric community that it is possible to study by detailed numerical simulation the full phenomenon, i.e. tornado formation in close conjunction with the thunderstorm; this will have an important impact on people’s view of future weather prediction systems and will influence strategic decisions on how these models can be developed;
  4. to show the general public a spectacular visual image of the opportunities offered by modern supercomputing.

Deconfinement of charm and strangeness in the Quark Gluon Plasma

Project leader: Dr Olaf Kaczmarek, University of Bielefeld, Germany
Research field: Fundamental Constituents of Matter
Resource Awarded: 58 350 000 core hours on FERMI @ CINECA, Italy; 21 650 000 core hours on JUQUEEN @ GCS@Jülich, Germany;

Collaborators: Heng Tong Ding, Central China Normal University- China
Laermann Edwin, University of Bielefeld-Germany
Frithjof Karsch, University of Bielefeld,-Germany
Florian Meyer, University of Bielefeld-Germany
Maezawa Yu, University of Bielefeld-Germany
Swagato Mukherjee, Brookhaven National Laboratory- United States
Hiroshi Ohno, Brookhaven National Laboratory- United States

Abstract: We propose to perform the first physical calculation of hadronic spectral functions at finite temperature for mesons with open charm and open strangeness. The calculations shall be performed using large thermal lattices of size 1283 x48 generated with 2+1 flavors of Highly Improved Staggered Quarks (HISQ) having physical masses. By applying Maximum Entropy Method spectral functions will be extracted from thermal hadron correlation functions calculated with appropriately tuned Wilson-clover fermions.

SOLDYN: Simulations of SOLar DYNamo cycle and differential rotation

Project leader: Dr Petri Kapyla, University of Helsinki, Finland
Research field: Universe Sciences
Resource Awarded: 32 352 064 core hours on Hermit @ GCS@HLRS, Germany;

Collaborators: Joern Warnecke, Max Planck Institute for Solar System Research-Germany
Maarit Mantere, Aalto University- Finland
Elizabeth Cole, University of Helsinki- Finland
Axel Brandenburg, NORDITA- Sweden

Abstract: The Sun exhibits magnetic activity at various spatial and temporalscales. The best known example is the 11-year sunspot cycle which isrelated to the 22-year periodicity of the Sun’s magnetic field. Thesunspots, and thus solar magnetic activity, have some robustsystematic features: in the beginning of the cycle sunspots appear atlatitudes around 40 degrees. As the cycle progresses these belts ofactivity move towards the equator. The sign of the magnetic fieldchanges from one cycle to the next and the large-scale field remainsapproximately antisymmetric with respect to the equator. This cyclehas been studied using direct observations for four centuries.Furthermore, proxy data from tree rings and Greenland ice cores hasrevealed that the cycle has persisted through millennia. The periodand amplitude of activity change from cycle to cycle and there areeven periods of several decades in the modern era when the activityhas been very low. Since it is unlikely that the primordial field ofthe hydrogen gas that formed the Sun billions of years ago could havesurvived to the present day, the solar magnetic field is considered tobe continuously replenished by some dynamo mechanism. The cycle alsomanifests itself in the occurrence of space weather events, where ahuge amount of energy is released in violent eruptions on theSun. These events can have huge impacts on human kind due to powergrid failures, high radiation doses in particular on polar flights,and high risk on spacecraft outside the Earth’s magnetosphere.

We study the solar dynamo and magnetic eruptions by performing twosets of high-resolution global numerical simulations of the turbulentconvection zone with or without a simplified corona above. The resultsof the study are likely to help in understanding what is causing solaractivity and how magnetic eruptions are initiated.

Multiscale Atomistic Simulation of the Mechanical Behaviour of Nickel-based Superalloys

Project leader: Dr James Kermode, King’s College London, United Kingdom
Research field: Chemistry
Resource Awarded: 15 000 000 core hours on FERMI @ CINECA, Italy; 5 000 000 core hours on JUQUEEN @ GCS@Jülich, Germany;

Collaborators: Federico Bianchini, King’s College London- Untied Kingdom
Alessio Comisso, King’s College London- Untied Kingdom
Alessandro De Vita, King’s College London- Untied Kingdom

Abstract: In this project we will study the mechanical properties of nickel-based superalloys, a class of materials that exhibit excellent strength and creep-resistance at high temperatures, even in chemically aggressive environments, making them suitable for the construction of efficient turbines for energy generation and aerospace applications. These ’superalloys’ are usually manufactured as single crystals, with very high strength resulting from the precipitation of an LI2-ordered structure known as γ within the face centred cubic matrix (γ phase). Dislocations are known to be pinned at γ/γ interfaces, strengthening the alloy, while defects such as vacancies and impurities are also thought to affect dislocation motion. However, there is currently no detailed understanding of the atomic-scale mechanisms underlying these processes. Here, we will address this, with the ultimate goal of providing fundamental insight that may allow the future design of materials with improved properties, for example raising the operating temperatures (and, thus, the efficiency) of turbines and, at the same time, their lifetime. In a world where high energy efficiency is essential for the success of new technologies, there is great scope for new and more accurate studies of the mechanical properties of this class of materials.

The problem is formidable if approached at the atomistic level: quantum precision is needed in order to accurately describe atoms close to dislocation cores or chemically complex γ/γ interfaces, but the system must be large enough to accommodate the strain gradients typical of long-range elastic interactions. For these reasons, no QM-accurate atomistic study of dislocation motion in superalloys has been carried out to date. We are addressing these issues using the ’Learn on the Fly’ (LOTF) technique, a multiscale quantum mechanical/molecular mechanical (QM/MM) method augmented by learning algorithms that allows us to perform quantum calculations only where and when they are necessary, minimising the overall computational cost. Massively parallel supercomputing platforms are an essential ingredient of the work, providing the necessary computational power to achieve converged results.

Linear Scaling and Massively Parallel Coupled-Cluster Calculations on the Leucine Transporter within a DEC framework

Project leader: Dr Thomas Kjaergaard, Aarhus University, Denmark
Research field: Biochemistry, Bioinformatics and Life sciences
Resource Awarded: 40 000 000 core hours on Curie TN @ GENCI@CEA, France;

Collaborators: Janus Eriksen, Aarhus University-Denmark
Patrick Ettenhuber, Aarhus University-Denmark
Ida Hoeyvik, Aarhus University-Denmark
Frank Jensen, Aarhus University-Denmark
Poul Joergense, Aarhus University-Denmark
Kasper Kristensen, Aarhus University-Denmark
Birgit Schiøtt, Aarhus University-Denmark

Abstract: The project aims at using the Divide-Expand-Consolidate (DEC) strategy to extend the application range of accurate quantum chemical calculations by applying state of the art coupled-cluster methods, CCSD (coupled-cluster with single and double excitations) and CCSD(T) (CCSD with an approximate treatment of triple excitations ), to molecular systems with sizes beyond the scope of standard implementations, thereby impacting real-life applications.

For this purpose we will consider neurotransmitter sodium symporters, which are membrane proteins responsible for regulating the level of neurotransmitter signaling. They are the main target for psycho-stimulants, antidepressants, and medications against e.g., anxiety, obesity, and addictive drugs. No high-resolution structures of these mammalian proteins are yet available, but the structure of a bacterial homologue, the leucine transporter (LeuT), was published in 2005. It has been speculated that the binding of two ligand molecules are required to activate the transport mechanism, and two possible binding sites have indeed been identified in LeuT by crystallography, binding assays, and molecular dynamics simulations. We propose to use this system as a target for the DEC-CCSD and DEC-CCSD(T) methods, where we aim at calculating benchmark binding energies for the binding of one and two leucine ligands to a model of the LeuT protein.

The DEC-CCSD and DEC-CCSD(T) methods have been implemented using the technology we previously have used for DEC second order Møller-Plesset perturbation theory (MP2) calculations. For DEC-MP2 we have shown that our implementation exhibits both strong and weak scaling. We want to show similar scalings for DEC-CCSD and DEC-CCSD(T) implementations by investigating increasingly larger stacks of benzene rings and cytosine nucleotides. These calculations will allow us to probe if pi-stacking can be described by pairwise interactions between units, or whether cooperative binding effects are important. The DEC-CCSD(T) interaction energies will also be compared to those from commonly used force fields that have been parameterized using pairwise interactions, and thereby probe the accuracy of force field methods, which are currently used routinely to estimate non-covalent binding affinities. Our investigation may in this way provide information for improving these empirical models.

Magnetic Reconnection in Three-dimensional Turbulent Configurations

Project leader: Prof Giovanni Lapenta, KU Leuven, Belgium
Research field: Universe Sciences
Resource Awarded: 25 000 000 core hours on SuperMUC @ GCS@LRZ, Germany;

Collaborators: Jorge Amaya, KU Leuven-Belgium
Emanuele Cazzola, KU Leuven-Belgium
Jan Deca, KU Leuven-Belgium
Maria Elena Innocenti, KU Leuven-Belgium
Evan Alexander Johnson, KU Leuven-Belgium
Koen Kemel, KU Leuven-Belgium
Vyacheslav Olshevsky, KU Leuven-Belgium
Lorenzo Siddi, KU Leuven-Belgium
Andrey Divin, Uppsala University- Sweden

Abstract: We aim at modeling of the most striking process in astrophysical plasma: magnetic reconnection. Magnetic reconnection is believed to be the only way the magnetic field energy is released to particles. In this fundamental process, magnetic field lines of opposite polarity are brought together and fused into a new magnetic configuration. The end effect of reconnection is conversion of magnetic energy to kinetic energy of the particles and a corresponding increase in thermal energy and flow velocity of the plasma. Because of its kinetic nature, magnetic reconnection in space plasmas could be thoroughly described and understood only by means of computer modeling. Of particular interest to the physicists now is the three-dimensional (3D) volumetric, random reconnection that might be happening in numerous null-points and small-scale current sheets in the highly turbulent interstellar environment.

We will conduct numerous three-dimensional simulations of such reconnection, which will allow us to
answer the three important questions:

  1. What agents are important for energy dissipation in the complex null-point configurations that are luckily present in interstellar plasma: the X-type null-points, or null-lines?
  2. What are the signatures of such reconnection (waves, particle jets, non-thermal particle spectra, etc.), that could be determined by spacecraft and in laboratory devices?
  3. What are the properties of turbulent reconnection and how those compare to magnetohydrodynamic (MHD) turbulence?

Leading hadronic contribution to the anomalous magnetic moment of the muon

Project leader: Dr Laurent Lellouch, CNRS (INP) and Aix-Marseille U., France
Research field: Fundamental Constituents of Matter
Resource Awarded: 35 350 000 core hours on FERMI @ CINECA, Italy; 35 350 000 core hours on JUQUEEN @ GCS@Jülich, Germany;

Collaborators: Stefan Krieg, Forschungszentrum Jülich-Germany
Eric Gregory, Wuppertal University-Germany
Christian Hoelbling, Wuppertal University-Germany
Kalman Szabo, Wuppertal University-Germany
Rehan Malak, CEA/CNRS (INP)-France
Christian Torrero, Centre de Physique theorique de Marseille- France

Abstract: We will compute from first principles and with small and controlled errors the leading hadronic contribution to the anomalous magnetic moment of the muon. This intrinsic property of a cousin of the well known electron is the most prominent of a very small number of experimentally observable quantities that show a significant discrepancy between the measured value and the prediction from the standard model of particle physics. Moreover, new experiments being built plan to reduce the measurement uncertainties by a factor of four. If theory can follow, these experiments may yield an unambiguous sign of new fundamental physics. Indeed, this apparent discrepancy has direct bearing on the search for such physics. The muon magnetic moment is the only particle physics observable that provides an upper bound on masses of speculative new particles in supersymmetric theories and other standard model extensions. The caveat of the current situation is that the dominant uncertainties in the theory prediction come from strong interaction, hadronic corrections which are exceedingly challenging to compute. Of the two hadronic contributions, the leading one is obtained via dispersion relations from experiment while the second one is obtained with incomplete model calculations. Since the leading contribution also contributes dominantly to the total uncertainty, this is the one that we propose to investigate here using ab-initio lattice QCD computations. Our result will provide an important crosscheck of the phenomenological determinations of this contribution and will help validate the lattice approach for such computations, in view of calculating the even more challenging, sub-leading hadronic light-by-light contribution. Once this is done, theory and experiment will hopefully be on an equal footing, allowing physicists to find or strongly constrain new fundamental physics scenarios.

Towards ultimate turbulence – Taylor-Couette and Rayleigh-Bénard flow

Project leader: Prof Detlef Lohse, University of Twente, The Netherlands
Research field: Engineering
Resource Awarded: 15 500 000 core hours on Curie TN @ GENCI@CEA, France; 16 000 000 core hours on Hermit @ GCS@HLRS, Germany;

Collaborators: Roberto Verzicco, Univ. of Tor Vergata-Italy
Rodolfo Ostilla Monico, University of Twente-The Netherlands
Vamsi Arza Spandan, University of Twente-The Netherlands
Erwin van der Poel, University of Twente-The Netherlands
Yantao Yang, University of Twente-The Netherlands

Abstract: Turbulent flow is abundant in nature and technology. In contrast to a decade-old paradigm, even highly turbulent flow is strongly influenced by the boundaries. There is increasing evidence that there are different turbulent states, with sharp transitions in between them. The strength of turbulence is characterized by the Reynolds number Re, which gives the ratio between inertial and viscous forces. Reynolds numbers in typical process technology applications are of the order of 10^8. In the atmosphere Re=10^10 is achieved, and in an open ocean one has Re=10^11 and in stars even much higher values.

We will focus our study on two paradigmatic systems in fluid dynamics, namely Rayleigh-Benard (RB) convection [1-2] (the flow in a closed box heated from below and cooled from above) and Taylor-Couette (TC) turbulence [3] (the flow in between two independently rotating coaxial cylinders). The reasons why these systems are so popular are: (i) These systems are mathematically well-defined by the (extended) Navier-Stokes equations with their respective boundary conditions; (ii) for these closed system exact global balance relations between the respective driving and the dissipation can be derived; and (iii) they are experimentally accessible with high precision, thanks to the simple geometries and high symmetries. An addition benefit is that within our collaboration we have access to state of the art experimental RB [4-5] and TC setups [6-9], which will allow us to compare the simulation results with experimental findings.

In laboratory experiments and direct numerical simulations it is impossible to achieve the high Re numbers mentioned above. So one has to somehow extrapolate the experimental and numerical results at much lower Reynolds numbers to these high values. However, such an extrapolation becomes meaningless once there is a transition from one turbulent state at lower Re to another turbulent state at higher Re, or once different turbulent states coexist at the same Re. However, recent high Reynolds number RB and TC experiments strongly suggest that there are different turbulent states in this high Reynolds number regime [10-16]. However, it remains elusive why one or the other state is realized, and it is speculated that the existence of multiple turbulent states may be the origin of the observed behaviour. The difference between the different turbulent states can be huge, for example in RB convection the difference in heat transport between the two turbulent states is a factor of 3 in the Ra number range that is relevant for various geophysical, astrophysical, and process-technological situations a better prediction of the heat transfer is necessary. Therefore, the objective of this project is to explore when there are different states of turbulence and how transitions between these different states occur. What determines in what state the turbulent flow is? What are the roles of the boundary layers and how do boundary layers and bulk flow interact? What are the most appropriate observables to characterize the different turbulent states? And finally: Can one trigger such a transition?

Atomistic simulations of advanced band-to-band tunneling transistors

Project leader: Prof Mathieu Luisier, ETH Zurich, Switzerland
Research field: Chemistry
Resource Awarded: 13 300 000 core hours on Hermit @ GCS@HLRS, Germany;


Abstract: To circumvent the continuous power increase of conventional metal-oxide-semiconductor field-effect transistors (MOSFETs), new device architectures are needed. Band-to-band tunneling field-effect transistors (TFETs) represent an energy-efficient alternative to MOSFETs and could become the next generation logic switches thanks to the injection of cold instead of hot electrons from source to drain. TFETs are expected to deliver the same performance as MOSFETs, but at a lower supply voltage, and therefore at a lower power consumption. Unfortunately, so far, there have been very few successful experimental demonstrations of TFETs exhibiting the desired properties, the most important shortcoming being a very low ON-state current. In this research effort, which is part of a recently funded European project called E2SWITCH, two directions will be pursued to resolve this issue: (i) the usage of low-band gap materials to enhance the tunneling rate between source and channel and (ii) the investigation of novel device geometries to increase the tunneling area and the resulting drive current. With that respect, a new concept has been recently proposed, the electron-hole bilayer TFET (EHBTFET), which exploits the formation of a 2-D, bias-induced electron-hole gas below the gate contact. With a state-of-the-art quantum transport solver, EHBTFETs composed of a GaSb-InAs broken gap hetero-junctions will be simulated and their performance optimized so that they can challenge MOSFETs, but at a lower power consumption.

MICRODYN – The microscopic dynamics of functional oxides: non-collinear polarization and domain-wall mobility

Project leader: Prof Nicola Marzari, EPFL, Switzerland
Research field: Chemistry
Resource Awarded: 50 000 000 core hours on FERMI @ CINECA, Italy;

Collaborators: Paolo Umari, University of Padova-Italy

Abstract: Functional oxides are a materials’ class that combines some of the most exciting fundamental physics and possible technological applications. Key phenomena include ferroelectricity, magnetoresistance, charge ordering, and superconductivity, to name only a few, and applications span e.g. solar energy harvesting in metal-organic perovskites, water splitting in double perovskites, superconductivity in cuprates, piezoelectric coupling in titanates, superionic proton and Li conduction in solid-state electrolytes, or even post-perovskite transitions in the planet’s interior. In our study of ferroelectric response in perovskites we have uncovered novel microscopic mechanisms that we believe are key to understanding phase stability in this class of materials, and we want to launch a massive project to elucidate the microscopic dynamics of ferroelectric perovskites as they go across their phase transitions in temperature – in the process clarifying once and for all the order-disorder vs. displacive nature of the transitions, debated for decades in the literature. We believe only large-scale first-principles (Car-Parrinello) molecular dynamics are able to provide the correct description of these phenomena, something for which we have an extensive track record and optimal codes for massively-parallel architectures (the CP code in the Quantum-ESPRESSO distribution). We also want to directly apply this fundamental insight on an applied technological problem, i.e. the dielectric response of most common functional ferroelectrics (PbTiO3, PbZrO3 and their PZT alloys), by understanding the dynamics of the domain walls between different ferroelectric regions, that is key in classifying ’hard’ and ’soft’ ferroelectrics, and where, very recently, the existence of 2D electron gas is being uncovered at charged domain wall – these static calculations require unit cells of the order of 1000-5000 atoms, and such unprecedented sizes are made possible by the application of linear-scaling density-functional theory, using the recent development of the ONETEP code for BlueGene architectures. Finally, we believe many of these findings will also be relevant to many other of the (multi)functional perovskites, beyond the present focus on ferroelectric oxides.

ENS4OCEANENSemble-based approach for global OCEAN forecasting

Project leader: Dr Simona Masina, Centro Euro-Mediterraneo sui Cambiamenti Climatici-Italy
Research field: Earth System Sciences
Resource Awarded: 13 000 000 core hours on MareNostrum @ BSC, Spain;

Collaborators: Giovanni Aloisio, Centro Euro-Mediterraneo sui Cambiamenti Climatici-Italy
Italo Epicoco, Centro Euro-Mediterraneo sui Cambiamenti Climatici-Italy
Pier Giuseppe Fogli, Centro Euro-Mediterraneo sui Cambiamenti Climatici-Italy
Doroteaciro Iovino, Centro Euro-Mediterraneo sui Cambiamenti Climatici-Italy
Silvia Mocavero, Centro Euro-Mediterraneo sui Cambiamenti Climatici-Italy
Andrea Storto, Centro Euro-Mediterraneo sui Cambiamenti Climatici-Italy

Abstract: The exploitation and protection of marine coastal and polar regions require predictive capability at space scales that so far have been forbidden due to the numerical and computational challenges involved. Improved ocean forecasts at space and time scales higher than those presently available will support ocean related industrial activities (such as offshore activities for wind and tidal energy production and fisheries activities), facilitate the implementation of European maritime and environmental policies (e.g MSFD) and cope with hazards and scarceness of natural resources.

In the attempt to fulfill these needs at CMCC-ANS Division we have recently implemented a global ocean data assimilation system based on a model at 1/16° horizontal resolution and 100 vertical levels with the purpose to develop in the near future a real-time forecasting system able to provide on daily basis forecasts of global oceanographic parameters for the following 10 days. The predictability skills of such a system are highly dependent on the model performance and therefore it is crucial to validate the model and test its capability to represent in an accurate way physical processes both for open ocean and coastal domains such as for example transport through straits and overflows at sills, mixing processes and daily frequency variability.

The project will focus on error modelling estimation using an ensemble spread-based approach to provide meaningful estimations of uncertainties. Ensemble simulations are proved to optimally mimic the error propagation of a forecast system in both global and regional applications, provided that all the sources of errors are adequately perturbed. Thus, the ensemble spread is a measure of the forecast error. The construction of an ensemble system in this project has the twofold objective to use the results to build the background-error covariance matrix of the assimilation scheme used in the forecasting system and quantify the model uncertainties. While the former is a crucial step in the construction of any data assimilation system, either variational or ensemble, the latter will allow us to study the absolute predictability of main ocean processes (e.g. mixing, transports in the main ocean
straits, etc.) as a function of the forecast range.

The ensemble members will be generated by perturbing the initial conditions, the surface forcings and the ocean model parameterizations. The experimental design will consider a baseline experiment, i.e. a deterministic run of the model, from which we will draw some ocean state realizations that will be perturbed. Each ensemble simulation will start three months apart (one per season), and the ensemble size will consist in 8 members. The simulation will be 10 years long, covering the 2003-2012 period, i.e. the period of full deployment of Argo floats. Each ensemble member will run for 40 days in a forecast mode. The expected achievements of this project will therefore be i) a 1/16 global ocean simulation for process studies; ii) monthly time scale predictability analysis for the 1/16 degree forecast system; iii) re-estimation of background-error covariances for further use in the operational context.

QUASINO – QUAntum SImulation of ultimate NanO-devices

Project leader: Mr Yann-Michel Niquet, CEA, France
Research field: Chemistry
Resource Awarded: 123 000 core hours on Curie H @ GENCI@CEA, France; 6 505 000 core hours on Curie TN @ GENCI@CEA, France;

Collaborators: Ivan Duchemin, CEA- France
Jing Li, CEA- France
François Triozon, CEA- France
Christophe Delerue, IEMN- France
Denis Rideau, STMicroelectronics- France

Abstract: The characteristic dimensions of the transistors on a processor chip have steadily decreased over the last fifty years, allowing for ever more performances and functionalities. The International Technology Roadmap for Semiconductors (ITRS), which sets targets for the microelectronics community, is now discussing options beyond the “16 nm node”. The physics of such devices is, however, very complex and goes beyond semi-classical understanding. Modeling and simulation are therefore expected to play an increasing role in the exploration of original and innovative designs.

This calls for the development of advanced simulators able to account for quantum effects, such as tunneling and confinement, prevailing at the nanometer scale. The Non-Equilibrium Green’s Functions (NEGF) method is one of the most versatile approaches to quantum transport. It describes all important scattering mechanisms (scattering of electrons by impurities, lattice vibrations, etc…) in an unified framework. We have recently developed a new NEGF solver based on effective mass and atomistic tight-binding models of the electronic structure of the materials, carefully optimized for high-performance computing infrastructures and hybrid CPU/GPU machines. This code is able to simulate devices with realistic geometries and sizes in the 5-30 nm range, representative of the next generation of transistors being developed at present.The objective of this project is to model the latest “Fully-Depeleted Silicon-on-Insulator” technologies developed at STMicroelectronics in Europe, and the next “Trigate” transistors being prepared at CEA/LETI. We will, in particular, try to answer pending issues about the strength of the interaction between electrons and lattice vibrations in these devices, and about the mechanisms of interaction between the electrons and the so-called “gate stack”, a complex structure of oxides controlling the behavior of the transistors. As the reproducibility of the characteristics of ultra-scaled device is a real challenge, we will also put a particular emphasis on the modeling of the variability in short channel devices. Our aim is to introduce quantum simulation in the design of the next generation of transistors, in order to reduce the number of costly development batches.

MACS – MARSIS Clutter Simulator

Project leader: Dr Roberto Orosei, Istituto Nazionale di Astrofisica, Italy
Research field: Universe Sciences
Resource Awarded: 7 000 000 core hours on SuperMUC @ GCS@LRZ, Germany;

Collaborators: Wlodek Kofman, Observatoire des Sciences de l’Univers de Grenoble- France
Marco Cartacci, Istituto Nazionale di Astrofisica- Italy
Alessandro Frigeri, Istituto Nazionale di Astrofisica- Italy
Federico Cantini, Universit di Roma “La Sapienza”- Italy

Abstract: Mars is today a cold, dry and sterile world with a thin atmosphere made of CO2. The geologic and compositional record of the surface reveals however that in the past Mars had a thicker atmosphere and liquid water flowing on its surface. For this reason, it has been postulated that life could have developed and that some primitive life forms may be existing even today. To discover the possibilities for life on Mars – past or present – the NASA Mars Exploration Program has developed an exploration strategy known as “Follow the Water”. Following the water requires searching for water or ice reservoirs on the surface or in the subsurface of Mars.

Ground Penetrating Radar (GPR) is a well-established geophysical technique based on the transmission of radar pulses at frequencies in the MF, HF and VHF portions of the electromagnetic spectrum into the surface, to detect reflected signals from subsurface structures. Orbiting GPR, often called subsurface radar sounders, have been successfully employed in planetary exploration, and are the only remote sensing instruments capable of studying the subsurface of a planet from orbit.

MARSIS is a synthetic-aperture, orbital sounding radar carried by the European Space Agency spacecraft Mars Express. MARSIS transmits through a dipole, which has negligible directivity, with the consequence that the radar pulse illuminates the entire surface beneath the spacecraft and not only the near-nadir portion from which subsurface echoes are expected. The electromagnetic wave can then be scattered by any roughness of the surface.

If the surface of the body being sounded is not smooth, then part of the incident radiation will be scattered in directions different from the specular one: areas of the surface that are not directly beneath the radar will scatter part of the incident radiation back towards it, and thus produce surface echoes that will reach the radar after the echo coming from nadir, and that can be mistaken for subsurface echoes.

To validate the detection of subsurface interfaces, numerical electromagnetic models of surface scattering have been used to produce simulations of surface echoes, which are then compared to real echoes.

The goal of this proposal is to use a parallelized and numerically optimized version of a code for the simulation of radar surface scattering to simulate the entire MARSIS dataset, consisting of several thousand observations, and to produce a dataset of simulated observations that will be made publicly available to the international scientific community for MARSIS data analysis.

Such dataset will be a unique contribution without precedents in the interpretation of subsurface sounding data. No other planetary radar sounding experiment has provided a similar resource to data users. Simulations will make it possible, for example, to use automated procedures to detect subsurface interfaces, or to study the correlation between real and simulated surface echo strengths to detect unusual surface properties.

crowding effect on protein landscape

Project leader: Prof Modesto Orozco, IRB Barcelona, Spain
Research field: Biochemistry, Bioinformatics and Life sciences
Resource Awarded: 33 050 000 core hours on MareNostrum @ BSC, Spain;

Collaborators: Federica Battistini, IRB Barcelona-Spain
Michela Candotti, IRB Barcelona-Spain
Pablo Dans, IRB Barcelona-Spain

Abstract: Macromolecular crowding is an important factor and a not yet understood effect, which can influence the behavior of proteins in cellular environments. However the effect of crowding has not been tackled yet, due to the intrinsic problems of studying complex and large systems both theoretically and computationally. In particular the scenario gets more complex for IDP, intrinsically disordered proteins, in determining the role of the environment on protein structural properties and function. Therefore we propose to investigate in silico the crowding effect of an IDP, the NCBD (nuclear co-activator binding domain) protein, using molecular dynamics (MD) simulations, on systems with different protein concentrations. Our goal is to answer and provide atomic-level evidence to a variety of questions that include:

  1. Effect of crowding on protein stability, analysis of universal and specific effects of protein crowding using different protein concentrations of NCBD in solution.
  2. Differential effects of protein crowders versus non-protein crowders, comparison with PEG (polyethylene glycol) control simulation.
  3. Complex formation of NCBD with its multiple partners (proteins IRF and ACTR), using different concentrations in water solution.
  4. Comparison of computational results with experimental data on heterogeneous and homogeneous protein crowded systems.

MEGAPV — Multiple Exciton Generation: Application to PhotoVoltaic

Project leader: Prof Stefano Ossicini, Università degli Studi di Modena e Reggio Emilia, Italy
Research field: Chemistry
Resource Awarded: 32 000 000 core hours on FERMI @ CINECA, Italy;

Collaborators: Olmes Bisi, Università degli Studi di Modena e Reggio Emilia-Italy
Ivan Marri, Università degli Studi di Modena e Reggio Emilia-Italy
Marco Govoni, University of California, Davis- United States

Abstract: In this proposal density functional perturbation theory (DFPT) will be used to study Carrier Multiplication (CM) dynamics in oxygen passivated Silicon nanocrystals (Si-NCs). CM is a Coulomb driven nonradiative recombination mechanism that consists in the generation of multiexcitons after absorption of a single photon. If efficient, CM can be as fast as or even faster than photon emission, extending thus the portion of the solar spectrum that can be converted into photocurrent. Effects induced by CM on the excited carried dynamics were observed in different nanostructured (PbSE and PbS [1-2], CdSe [3], PbTe [4], InAs [5] and Si [6]) materials. Moreover, thanks to the pioneering work of Semonin et al. [7], a relevant photocurrent enhancement arising from CM was proven in a PbSe based quantum dot solar cell. Recently, new CM dynamics were observed by Timmerman et al. and Trinh et al. [8-10] in a dense array of Si-NCs. This effect, called space separated quantum cutting (SSQC), differs from the standard CM (one-site CM), because the generation of two e- h pairs after absorption of a single photon occurs in two different (space separated) NCs. CM via SSQC stems from the NC-NC interaction and represents one of the most suitable routes for solar cell loss factor minimization. In this project we will analyze, by ab-initio techniques, CM dynamics in oxygen passivated Si-NCs of different size and with different surface conformations. Due to the complexity of the problem that requires massive simulations, CM lifetimes have been never calculated in oxygen terminated Si-NCs, despite these systems are often closer to the ones investigated experimentally. Using a procedure never proposed before, we will investigate CM dynamics in three different Si-NCs passivated with hydroxyls functional groups (-OH). Results obtained for the largest system (the Si147(OH)100) will be then compared with the one obtained for a Si-NCs (with the same number of Si atoms) extracted from a SiO2 dielectric matrix. Effects induced by oxygen surface conformation on CM dynamics will be investigated in details. As a final step, we will analyze effects induced by NC-NC interaction on CM efficiency and CM threshold (two-site CM effects). This analysis will be performed by placing two different oxygen-passivated Si-NCs in the same simulation box, thus accurately simulating the systems investigated by Timmerman et al. and Trinh et al. [8-10]. Effects induced on CM dynamics by both energy and charge transfer processes will be investigated in details. A direct comparison with results obtained by our group in the study of CM dynamics for isolated and interacting hydrogenated Si-NCs will be performed [11]. At the end, starting from the calculated one-site and two-site CM lifetimes, we will solve a set of rate equations in order to simulate excited carriers dynamics in a dense array of Si-NCs when high energy photons are absorbed.

Spatially adaptive radiation-hydrodynamical simulations of reionization

Project leader: Dr Andreas Pawlik, Max Planck Society, Germany
Research field: Universe Sciences
Resource Awarded: 33 800 000 core hours on SuperMUC @ GCS@LRZ, Germany;

Collaborators: Claudio Dalla Vecchia, Max Planck Society-Germany
Alireza Rahmati, Max Planck Society-Germany
Joop Schaye, Leiden University-Germany

Abstract: The first sources of ionizing radiation in our universe transformed the cold and neutral cosmic gas that was present shortly after the Big Bang into the hot and ionized plasma that we observe in the intergalactic medium (IGM) today. Research into this transformation – the Epoch of Reionization – is crucial for understanding the formation and evolution of galaxies, including our own galaxy, the Milky Way. A new generation of telescopes is therefore underway to unravel the astrophysics of these early times, including flagships such as the Low Frequency Array and the James Webb Space Telescope. Our project aims to quantify the potential of observations with these upcoming telescopes to constrain physical models of galaxy formation and reionization and further to guide the interpretation of the observations once the data arrives.

The complex and nonlinear nature of reionization requires the use of cosmological numerical simulations for an ab initio theoretical treatment of the physical processes involved, primarily the transport of ionizing photons and their interaction with the cosmic gas. While radiative transfer simulations of reionization have become feasible in the last decade thanks to advances in numerical techniques and the advent of high-performance supercomputing facilities, simulating reionization remains a computationally demanding
task. A principle challenge is the required dynamic range. Simulations need to cover large representative volumes of the universe and resolve them down to very small scales to follow both the progress of reionization as well as the formation of the galaxies driving it. Treating this vast range in scales requires the use of spatially adaptive simulation techniques. Such techniques are now routinely employed in modern cosmological hydrodynamical simulations. However, most of the accurate radiative transfer methods have not yet arrived at this level of algorithmical sophistication. These methods typically also require a computational cost that increases linearly with the number of sources, which greatly exacerbates the numerical challenge.

Our project will improve on previous works by applying the novel, accurate and well-tested radiative transfer technique TRAPHIC in radiation-hydrodynamical simulations of reionization. TRAPHIC is a spatially adaptive radiative transfer method and, thanks to its innovative design, has a computational cost independent of the number of sources. TRAPHIC is coupled to the galaxy formation code GADGET, which contains modules for the birth of stars and their explosion in supernovae, the associated chemical enrichment of the universe and the formation of black holes. Our simulations will thus capture both the physics of reionization and galaxy formation, as well as their tight interplay. This will allow us to address some of the most pressing unanswered questions of reionization, and to help provide the theoretical background that is urgently needed for the interpretation of the upcoming observations.

Evolution of the corona above an active region after sunspots formed

Project leader: Prof Hardi Peter, Max Planck Institute for Solar System Research, Germany
Research field: Universe Sciences
Resource Awarded: 17 400 000 core hours on SuperMUC @ GCS@LRZ, Germany;

Collaborators: Sven Bingert, Max Planck Institute for Solar System Research-Germany
Feng Chen, Max Planck Institute for Solar System Research-Germany
Jörn Warnecke, Max Planck Institute for Solar System Research-Germany

Abstract: Cool stars like our Sun are surrounded by a hot outer atmosphere, the corona. One of the open questions in astrophysics is how the million Kelvin hot corona can be sustained by the cool stellar surface which in the case of the Sun is only 6000 K cool. There is a general consensus that the complex magnetic field in the outer atmosphere plays a major role, but the actual processes are largely undetermined. In our numerical simulation we account for the interaction of the magnetic field and the ionized gas in the corona: the plasma is heated by the dissipation of currents that are induced by the coronal magnetic field being shuffled around through the motions of the convection at the surface of the Sun. From our model we synthesize the emission that would be emitted in the extreme UV and X-rays. This allows us to directly compare our simulation results to modern space-based observations obtained with e.g. the Solar Dynamics Observatory, Hinode or IRIS.

The main interest of our numerical experiment is the emergence of a new active region on the Sun and how the corona evolves during this emergence process. When new sunspots form at the surface of the Sun, new magnetic flux is emerging from the interior of the Sun. This is leading to a significant disturbance of the magnetic field and thus induces very strong currents. Consequently large amounts of energy are dumped in the corona where the plasma is heated and starts shining bright in extreme UV and X-rays. In a still ongoing PRACE project we could follow the initial process of the formation of a new active region, in which new coronal loops appear. These thin loops share major features with actual observations of newly formed coronal loops. This indicates that the physics in the numerical model adequately capture the processes in the solar corona.

In this new PRACE project we will further follow the evolution of the active region until the sunspots have fully formed, i.e. until the active region has fully developed. This simulation will allow us to study the transition of the emerging coronal loops into a more evolved configuration. An active region spends most of its lifetime in such a dynamic state, where coronal loops appear and disappear, but the overall appearance changes only slowly. To capture this state adequately in the simulation will be the next step to understand how the corona is structured and heated.

For our numerical studies we employ the Pencil Code, a high-order modular 3D MHD code that we have adapted for coronal problems. Used by a large astrophysical community this code is well suited for massively parallelized applications and has been run by our group at various Tier-0 systems. For the numerical experiment in this project we will need a total amount of about 17 Mhours of CPU time on SuperMUC to be used in the timeframe of one year.

PHHP – Phases of hydrogen at high pressure

Project leader: Prof Carlo Pierleoni, University of L’Aquila, Italy
Research field: Chemistry
Resource Awarded: 30 000 000 core hours on Hermit @ GCS@HLRS, Germany;

Collaborators: Markus Holzmann, CNRS- France
Sara Bonella, University of Rome “Sapienza”- Italy
Giovanni Rillo, University of Rome “Sapienza”- Italy
David Ceperley, University of Illinois at Urbana-Champaign- United States
Jeffrey McMahon, University of Illinois at Urbana-Champaign- United States
Norman Tubman, University of Illinois at Urbana-Champaign- United States

Abstract: The physical behavior of hydrogen under high compression is a fundamental and still unsolved problem of condensed matter and high pressure physics, probably one of the most challenging from the experimental and the theoretical viewpoints [1]. The search for metallization of hydrogen under compression, inspired by a seminal work of Wigner and Huntington in 1935 [2], has been one of the driving forces behind developments in high pressure experimental techniques and numerical simulation methods. Despite almost 75 years of intense research, and recent claims of success [3,4], hydrogen metallization at low temperature remains elusive [5-8]. At higher temperature in the liquid phase, metallization has been observed [9-11] and a first-order transition line between a molecular insulating liquid and an atomic metallic liquid has been predicted [12-14].

From the theoretical perspective, probably the most relevant, still missing, piece of information is which crystalline structure(s) solid hydrogen adopts when increasing pressure and temperature. This is the essential input to determine the location of the various phase lines and their interplay by ab-initio method. Experiments have detected solid insulating hydrogen up to 350 GPa. There is a general consensus though and expectation that above 600 GPa hydrogen will be metallic and monoatomic. In between, there have been a number of theoretical conjectures about the crystalline structure(s) of both the diatomic and the monoatomic phases, including the proposition of a low temperature liquid phase separating these two crystalline states [1].

Modern structure search methods are based on Density Functional Theo
ry (DFT) and available predictions for dense hydrogen rely on the GGA—PBE approximation [15] . We recently found that this approximation is not accurate enough, both around the liquid-liquid transition line and in phase III of solid hydrogen in particular when quantum nuclei are considered [16,17]. We have recently performed an extensive benchmark of various DFT approximations by Quantum Monte Carlo, which is a more fundamental method, and detected a number of approximated functionals that are more appropriate to high pressure hydrogen. We are presently performing structural predictions using those functionals. Once candidates structures will be available our first goal will be to compute their free energy, considering explicitly quantum nuclei, in order to detect the crystalline structures in the pressure range between 200 and 800GPa. The knowledge of the free energy allows to predict also the melting line and the interplay of this transition with the liquid-liquid line described previously and how this dissociation line enters the solid phase.

Beyond 500-600GPa, recent calculations have predicted a stable monoatomic crystal at low temperature [18-20] which, according to Eliashberg theory [21], should be superconducting up to the melting temperature (around room temperature). This prediction relies on the estimated electron-phonon coupling, the key ingredient of Eliashberg theory, from DFT-PBE. One final goal of our proposal is to benchmark these predictions by computing the crystal stability against the liquid and the electron-phonon coupling of this phase by Quantum Monte Carlo methods.

Influence of the Injector Geometry on Primary Breakup Modeling

Project leader: Univ-Prof Dr -Ing Heinz Pitsch, RWTH Aachen University, Germany
Research field: Engineering
Resource Awarded: 30 300 000 core hours on MareNostrum @ BSC, Spain;

Collaborators: Mathis Bode, RWTH Aachen University- Germany

Abstract: A variety of flows encountered in industrial configurations involve liquid and gas. Systems to atomize liquid fuels, such as diesel injection systems, are one example. The performance of a particular technical design depends on a cascade of physical processes, originating from the nozzle internal flow, potential cavitation, turbulence, and the mixing of a coherent liquid stream with a gaseous ambient environment. This mixing stage is critical, and the transfer occurring between liquid and gas is governed by an interface topology. In this regard, the objective of an atomizer design could therefore be to maximize the interface surface density. However, how design parameters influence surface density and ultimately drop size distribution is not clear, and predictive models do not exist. This is partly caused by the complexity of the process and the difficulties of performing experiments characterizing the atomization process.

The most serious gap in understanding and modeling of spray formation is primary atomization, but it is also the first physical process to be modeled. This means that uncertainties in the modeling of primary atomization will influence, for example, the design and performance of atomizers in diesel combustion systems all the way down to emission and pollutant formation. Therefore, this project addresses the understanding and accurate modeling of the effect of the injector geometry on the primary breakup in turbulent spray formation. It is proposed to perform a 2-step analysis based on hybrid simulations for studying this fundamental problem from first principles and creating a dataset which can be used for model development and as reference case for industrial scale simulations.

All simulations are performed using an in-house code called CIAO. CIAO is a structured, second order, semi-implicit finite difference code which enables the coupling of multiple domains and the simultaneous computation. It is parallelized by MPI.

Longitudinal and Transverse Electronic Transport in Atomically Doped Graphene from First Principles

Project leader: Prof Stephan Roche, Catalan Institute of Nanotechnology, Spain
Research field: Chemistry
Resource Awarded: 22 061 184 core hours on Curie TN @ GENCI@CEA, France;

Collaborators: Jean-Christophe Charlier, Universit catholique de Louvain (UCL)- Belgium
Nicolas Leconte, Universit catholique de Louvain (UCL)- Belgium
Frank Ortmann, Dresden University of Technology- Germany
Aron Cumming, Catalan Institute of Nanotechnology- Spain

Abstract: This project addresses the study of the phase diagram of the Quantum Hall effect (QHE) in complex forms of chemically and structurally disordered graphene-based materials. The QHE is one of the fundamental quantum phenomena discovered in the early eighties, defined by a sample independent series of transverse (Hall) conductances given by integer multiples of the fundamental constants e2/ h (elementary charge e and h Planck constant). The anomalous QHE has been revealed in materials exhibiting low-energy Dirac fermions excitations (graphene in 2004 and topological insulators in recent years). Our proposed scientific project aims to study the QHE in experimentally realistic sized samples of chemically functionalized graphene (mainly hydrogenated and models of polycrystalline graphene) by using an order-N method (N being the number of atoms) under weak magnetic fields (from the mT regime) and strong magnetic fields (up to 80T).

HiThrust: High-pressure Turbine Heat Transfer Numerical Simulations

Project leader: Prof Richard Sandberg, University of Southampton- United Kingdom
Research field: Engineering
Resource Awarded: 35 000 000 core hours on Hermit @ GCS@HLRS, Germany;

Collaborators: Vittorio Michelassi, General Electric- Germany
Thomas Ripplinger, General Electric- Germany
Li Wei Chen, University of Southampton- United Kingdom
Roderick Johnstone, University of Southampton- United Kingdom
Richard Pichler, University of Southampton- United Kingdom
Neil Sandham, University of Southampton- United Kingdom
Andrew Wheeler, University of Southampton- United Kingdom

Abstract: We aim to simulate the unsteady gas flow over a turbine blade within an aero-engine. The study will focus on a high-pressure turbine blade closely situated downstream of the combustion chamber where gas temperatures and speeds are very high. This gas flow can only be simulated accurately with large computing power because the gas flow is turbulent and affected by interactions between moving and stationary components which creates a highly unsteady flow. It is necessary to resolve all the temporal and spatial scales in the flow to determine the effects of turbulence and flow unsteadiness on the aerodynamics and the heat transfer from the gas to the metal blade. The blade aerodynamics affects the efficiency of the aero-engine while the heat transfer from the extremely hot gas to the turbine blades is harmful
to the life-span of the turbine. However, experimental measurements have to date not been able to provide data with enough depth to identify all fundamental mechanisms and to explain weaknesses of currently used design tools due to the difficulties of performing engine-scale experiments and acquiring spatially and temporally resolved data. By directly solving for all turbulent scales, Direct Numerical Simulations (DNS) constitutes the only tool for developing an improved understanding of the role of turbulent phenomena on the flow-field, and for determining the validity of current turbulence modelling. Here we want to address three fundamental questions:

  1. Determine to what degree the assumptions of constant turbulent Prandtl and Schmidt number is valid. The large-scale unsteadiness of the incoming flow overlaps with the background turbulence to invalidate the assumption of constant turbulent Prandtl and Schmidt numbers.
  2. Generate better understanding of the strain-stress misalignment in a realistic turbine flow environment. The large-scale unsteady environment causes a continuous rotation of strain and stress tensors that are very seldom aligned. The relative orientation of these two tensors drives entropy and turbulence production rates, and, ultimately losses and diffusion.
  3. Anisotropy is not captured by Boussinesq viscosity models, which is why these models struggle with predicting transition as they are unable to capture anisotropy around solid boundaries and at stagnation points. In a large-scale unsteady environment this situation is even more complicated. For this project, we will use software which has been developed at Southampton, and is proven to be highly efficient for large parallelized computations. In addition to answering the above listed fundamental scientific questions, the results will also provide a valuable benchmark that can be used to validate and improve current and future modelling of turbulence. The accuracy of such turbulence models is crucial to the development of high performance aero-engines. The project is in close collaboration with GE Global Research who are providing technical support and will play an integral part in the analysis of the data. GE will be able to implement the results of this work throughout their international aero-engine business and therefore this work, besides resulting in scientific publications in leading journals, also has the potential to deliver a significant change to global aviation.

MDCellSignal – Simulating the effect of membrane curvature on the clustering of cell signalling proteins.

Project leader: Prof Mark Sansom, University of Oxford- United Kingdom
Research field: Biochemistry, Bioinformatics and Life sciences
Resource Awarded: 12 700 000 core hours on Curie TN @ GENCI@CEA, France;

Collaborators: Philip Fowler, University of Oxford- United Kingdom
Heidi Koldsø, University of Oxford- United Kingdom

Abstract: Eukaryotic cells can sense the presence of an external chemical messenger and initiate a series of protein-protein interactions that cumulate in a change in the cell’s behaviour. Many of the important cell signalling proteins are either embedded in the cell membrane (such as receptors) or are associated with the cytoplasmic leaflet (such as G-proteins). Although it is generally accepted that these cell signalling proteins form nano-clusters, enhancing their local concentration, and improving the transduction of the signal, it is not clear how this clustering occurs. We shall use the PRACE infrastructure to investigate, using massive coarse-grained molecular dynamics simulations, the role of intrinsic membrane curvature in the clustering of a series of important cell signalling proteins.

Aeroacoustics of installed complex geometry jets for ultra-high bypass ratio engines

Project leader: Dr James Tyacke, University of Cambridge- United Kingdom
Research field: Engineering
Resource Awarded: 28 876 800 core hours on Hermit @ GCS@HLRS, Germany;

Collaborators: Ann Dowling, University of Cambridge- United Kingdom
Richard Jefferson-Loveday, University of Cambridge- United Kingdom
Iftekhar Naqavi, University of Cambridge- United Kingdom
Paul Tucker, University of Cambridge- United Kingdom

Abstract: The objective of this project is to model turbulence and noise generated by complex ultra-high bypass ratio (UHBR) engine jet nozzle configurations, such as co-axial jets with chevrons and a pylon, wing and flap. The effect of nozzles installed on the wing is of immediate interest to engine manufacturers seeking to use UHBR engines to meet strict noise and pollution emission targets. The close proximity of the wing has an unknown effect on noise generation for both serrated and non-serrated nozzles. The methodology of the project is to compute the turbulent flow field using Large-Eddy Simulation (LES) and use the Ffowcs-Williams Hawkings approach to propagate noise to the far-field, describing how sound emitted by the jet and surrounding geometry will affect perceived ground-level noise. The investigation of complex flow interactions will further current knowledge in flow physics, turbulence modelling and noise generation. In future this knowledge will lead to reduced noise levels. We intend to run LES simulations with mesh densities up to 200 million cells on several thousand processing cores.

WALLGRID – Wall turbulence interacting with grid-generated turbulence

Project leader: Prof J. Christos Vassilicos, Imperial College London- United Kingdom
Research field: Engineering
Resource Awarded: 38 604 800 core hours on SuperMUC @ GCS@LRZ, Germany;

Collaborators: Tom-Robin Teschner, Imperial College London- France
Thibault Dairay, Imperial College London- United Kingdom
Immanuvel Paul, Imperial College London- United Kingdom
Laizet Sylvain, Imperial College London- United Kingdom
Ning Li, Numerical Algorithms Group (NAG)- United Kingdom

Abstract: Cooling of a wall or of a hot fluid by a cold fluid separated by a wall is very common in many strands of industry and engineering. Cooling devices often rely on conduction, i.e. the molecular heat transfer across, say, aluminium fins in car radiators, but often also on turbulent convection, i.e. the passage of a cool turbulent fluid over hot walls. Turbulent convection may in fact be a cooling solution for a broader range of industrial and engineering cooling problems if we knew how to generate bespoke turbulent convection with bespoke properties suited to the problem at hand. The turbulence determines the amount of power required to drive the cooling flow (or the flow to be cooled) as well as the range of sizes of eddies and their distribution in space and, thereby, the heat transfer rate. A good cooling device requires as high a heat transfer rate for as low a driving power input as possible. Similar issues occur in mi
xing devices, equally pervasive throughout industry (breweries, pharmaceuticals, food, water aeration, chemical reactors, pollutant diluters, etc) and the particular type of mixers which this work is directly relevant to are static inline mixers, essentially square ducts/channels with mixing elements in them such as grids or other turbulence-generating obstructions. There, too, as in cooling devices, the goal is to increase mixing rate while at the same time decreasing the amount of power input required to achieve this mixing rate.

The computer calculations proposed here are the first ever of a turbulence generated by a grid interacting with walls calculated from first principles without turbulence modelling of any sort and therefore as reliable as possible. Such extremely advanced computations can only be achieved on a powerful High Performance Computing (HPC) platform. The configurations to be computed have been chosen carefully so as to allow maximum scientific and engineering insight. The bespoke turbulence will be generated with a fractal grid which is able to reduce pressure drop across it yet increase the turbulence intensity where it is needed and redistribute the eddies and their range of influence advantageously for effective stirring and heat transfer. It is yet unknown how power input, mixing rates and heat transfer coefficients shape up in the presence of both a grid and an adjacent wall, and the data obtained from thepresent HPC computations will be the first ever to help answer this question of very far reaching industrial and engineering consequence. This data will in fact do much more that that: it will help us understand the fundamentals of grid-generated turbulence interacting with walls, i.e. provide us with fundamental knowledge and insight on such turbulence. The ultimate goal, of course, is to be able to design bespoke cooling devices and static inline mixers with bespoke turbulence without the need of HPC for calculating performance but on the basis of simple back-of-the-envelope calculations which are possible only if fundamental knowledge is acquired.

CAMEL – CArdiac MechanoELectrophysiology

Project leader: Dr Edward Vigmond, Universite Bordeaux 1-France
Research field: Biochemistry, Bioinformatics and Life sciences
Resource Awarded: 6 764 771 core hours on SuperMUC @ GCS@LRZ, Germany;

Collaborators: Gernot Plank, Medical University of Graz-Austria

Abstract: Advances in science continue to provide ever increasing amounts of data and detail. With respect to cardiac physiology, MRI and CT can provide sub 100 micron resolution for large hearts, providing uprecedented anatomical detail. Models of single cell function now incorporate many subsystems, including ion movement, tension development, energetics, and stochastic calcium regulation. These descriptions continue to grow in size as we unravel the working of the single cell. It is these molecular level events which give rise to organ level behaviour. The most complex human heart simulations use meshes with millions of degrees of freedom, have behaviour at each node described by up to 100 differential equations, and must solve a linear system at each time step. Computational requirements are formidable and have hindered full implementation of all the modelling detail possible. The human heart is the most relevant and useful model, yet its size limits the detail which can be put in. This project will greatly enhance our existing code to decrease runtime of high resolution cadiac electromechanical simulations while increasing the level of complexity and detail used in the model. We will achieve this through several avenues. Strong scalability of our code will be improved through the use of better preconditioners and optimal partitioning. Algorithms which show better stong scalability will be used.

We will use our simulator for two goals. One is to explore the role of the cardiac conduction system in arrhythmogenesis. While the conduction system is vital for properly activating the heart and has been clearly demonstrated to have an effect even on the simplest conduction of electrical waves through the heart tissue, it has been neglected in modelling until quite recently. This project plans to be the first study that systematically elucidates its role during ventricular arrhythmias. We will explore how specific conduction system topology and characteristics affect activation rates, location of reentries, and if the arrhythmia which develops is more prone to degenerate into lethal fibrillation. Modelling the system will require careful integration into the framework since it is a small subsystem solved on fewer cores.

Secondly, the conduction system plays a crucial role in defining the cardiac activation and repolarisation phase of the heart which governs mechanical contraction and relaxation. Advanced studies necessitate to represent both aspects and their bidirectional interaction, referred to as electro-mechanical coupling and mechano-electric feedback. This constitutes a multiphysics problem that is challenging to model. We aim at developing a strongly coupled model of ventricular electromechanics where activation is driven by an anatomically detailed topologically realistic model of the conduction system and the mechanics are solved at the same resolution as the electrical problem..

EndoSel – Lipid degradation in the endocannabinoid system: Deciphering specificity in FAAH catalysis

Project leader: Dr Marco De Vivo, Italian Institute of Technology, Italy
Research field: Biochemistry, Bioinformatics and Life sciences
Resource Awarded: 13 900 000 core hours on Curie FN @ GENCI@CEA, France;

Collaborators: Pablo Campomanes, EPFL- Switzerland
Ursula Roethlisberger, EPFL- Switzerland
Giulia Palermo, Italian Institute of Technology- Italy
laura Riccardi, Italian Institute of Technology- Italy

Abstract: A functional endocannabinoid system is fundamental for human health, and crucial in the regulation of pathophysiological processes such as pain and inflammation. The understanding of the regulatory mechanisms of the endocannabinoid system is therefore of paramount importance in the search of novel drugs and therapeutic strategies against a number of pathologies, including cancer and other inflammatory-related diseases. In the endocannabinoid system, there are two essential hydrolytic enzymes, the fatty acid amide hydrolase (FAAH) and the monoacylglycerol lipase (MGL), that primarily mediate the action of the endocannabinoids anandamide (AEA) and 2-Arachidonoylglycerol (2-AG). How those enzymes, however, selectively hydrolyze the endocannabinoids AEA and 2-AG, better than others similar bioactive lipids, remain unclear. Here, we aim at understanding how FAAH, which hydrolyzes a wide range of fatty acids with different rates of hydrolysis, attains its specificity for AEA. In spite of the relevance of FAAH as a promising target for drug discovery, the structural and kinetic properties for substrate specificity in FAAH catalysis are still largely elusive. To clarify how FAAH attains substrate specificity, here we plan to integrate existing structural data with long-timescale molecular dynamics (MD) simulations of FAAH in complex with fatty acid substrates having different rates of hydrolysis, such as AEA, oleamide (OA) or p
almitoylethanolamide (PEA) – embedded in a realistic membrane/water environment. Building on our recent work on FAAH catalysis, these simulations will highlight the key factors involved in substrate specificity during FAAH catalysis, clarifying the role of structural flexibility in the regulation of endocannabinoid signaling. Importantly, the clarification of structural and kinetic properties of FAAH in regulating its enzymatic activity and selectivity could provide new insights for the rational design of new and more specific drug candidates acting via FAAH inhibition.

Constraining the cause of seismic anisotropy in the Earth’s deep mantle with high-frequency spectral element simulations

Project leader: Dr James Wookey, University of Bristol- United Kingdom
Research field: Earth System Sciences
Resource Awarded: 32 000 000 core hours on FERMI @ CINECA, Italy;

Collaborators: Andy Nowacki, University of Bristol- United Kingdom
Jack Walpole, University of Bristol- United Kingdom
Andrew Walker, University of Leeds- United Kingdom

Abstract: The Earth’s liquid iron outer core and overlying rocky mantle both convect continually to efficiently move heat from the interior to the surface. Mantle material moves on the order of centimetres per year, but the details and directions are unconstrained. Estimates of mantle flow derived from geodetic observations exist but it has been impossible to test whether these are compatible with the most direct probe of the Earth’s interior: seismic waves.

This project enables such a comparison by comparing new seismic observations with flow models. Perhaps the most promising seismic probe which can reveal flow patterns is the observation of seismic anisotropy, in which the seismic wavespeed of a medium varies with direction. In the upper mantle (UM), this is known to be related to the alignment of mineral grains in response to mantle flow, because the dominant UM mineral, olivine, is strongly anisotropic. Whilst most of the lower mantle appears not to show this property, the lowermost few hundred kilometres (known as D’’) does. This is shown by jointly analysing shear waves which pass above the region, and those which travel through it, such as S and ScS respectively. Until recently, it has been a necessary assumption that anisotropy in D’’ shows a high degree of symmetry, such as where seismic wavespeed only varies with wave propagation variation from the radial direction (vertical transverse isotropy, VTI). However, numerous studies show that such assumptions do not accurately represent the true elasticity which must be present in the lowermost mantle. Observations of shear wave splitting unambiguously show the presence of anisotropy in D’’, and provide sufficient information to relax assumptions about anisotropic styles. Hence they can discriminate between different candidate flow models and causative mechanisms behind the anisotropy.

We have recently computed the fully-general, anisotropic elasticity expected in D’’ for a range of candidate mineralogical conditions above the core-mantle boundary (CMB), including the three different ppv plasticity models and mixtures of phases with MgO+ppv. We make use of models of mantle flow derived from geodetic and seismological observations, and we impose no constraints on the symmetry of the elasticity which is generated.

Thus far, we have been able to test our predictions on the global scale using the VTI approximation, by comparison with global VTI models. We have also compared small-scale observations to our predictions with ray theory, allowing arbitrary symmetry of the anisotropy, but making an infinite-frequency approximation. Neither of these methods, however, can accurately capture the true relationship between structure in D’’ and the seismic wavefield measured at the Earth’s surface.

In this project, we compute the predicted seismic wavefield using the spectral element method (SEM) at frequencies sufficiently high to compare directly and accurately with new observations of D’’ shear wave splitting. Using the SPECFEM3D_GLOBE software on tens of thousands of CPUs and terabytes of memory, we aim to finally constrain the cause of anisotropy in D’’ and thus directly infer flow in the deep mantle.