7th Project Access Call – Awarded Projects

Find below the allocations of the PRACE 7th Project Access Call (in alphabetical order of the name of the project leader).

The nature of the CdTe/CdS interface: atomic diffusion from large-scale ab initio simulations

Project leader: Andreoni, Wanda; Ecole Polytechinque Federale de Lausanne(EPFL, Switzerland

Collaborators: Xiaoliang Hu, Ecole Polytechinque Federale de Lausanne (EPFL), Switzerland;Fabio Pietrucci, Ecole Polytechinque Federale de Lausanne (EPFL), Switzerland

Abstract: Solar cells based on cadmium telluride (CdTe) offer the cheapest thin-film technology amongst those so far proposed as alternative to crystalline silicon (c-Si). CdTe has characteristics that are ideal for photovoltaic conversion, namely a direct energy-gap of 1.45eV at room temperature and a very high absorption coefficient (104-105 cm-1) in the visible, combined with a relative ease of film formation. Very recently, a cell efficiency of 18.7% was attained. Recent remarkable progress in the fabrication of CdTe/CdS solar cells was grounded on empirical development. The understanding of the microscopic mechanisms affecting performance is very limited and the need is felt for an innovative knowledge-driven approach. This strategy could ultimately integrate experiments with simulations. The proposed project is meant to be a first step in this direction. It tackes th key component of any such device: the CdTe/CdS interface whose composition and structure influence crucially the performance. Our simulations will characterize the key inter-diffusion processes leading to the formation of the interface and the diffusion of impurities (including dopants) through it. Changes in the interface configuration and the electronic properties of the system will be monitored. Our calculations will benefit from advanced methodologies for atomistic simulations (ab initio molecular dynamics empowered by accelerated sampling techinques) and advanced software (CPMD) that exploits at best the power of BG/Q architectures.

Resource awarded: 26 405 600 core hours on FERMI @ CINECA, Italy

CHROMATOLOGY – Understanding the color optical properties of natural dyes using quantum mechanics and digital computers

Project leader: Baroni, Stefano; SISSA – Scuola Internazionale Superiore di Studi Avanzati, Italy

Collaborators: Arrigo Calzolari, CNR-Nano Istituto di Nanoscienze, Italy; Alessandro Biancardi, SISSA – Scuola Internazionale Superiore di Studi Avanzati, Italy; Xiaochuan Ge, SISSA – Scuola Internazionale Superiore di Studi Avanzati, Italy; Iurii Timrov, SISSA – Scuola Internazionale Superiore di Studi Avanzati, Italy

Abstract: The purpose of this project is to clarify the molecular mechanisms that determine the color optical properties of anthocyanins, a class of natural dyes responsible for the characteristic coloration of many fruits, vegetables, and flowers, as well as of the red and purple shades in the leaves of plants. The multitude of colors that these dyes express results from a complex range of factors, including the substituents of the aromatic rings, the acidity of the solution, its temperature, as well as the presence of metal cations and/or co-pigmentations. The relative importance of some of these factors will be addressed by a combination of ab-initio molecular dynamics, Monte Carlo Sampling, and implicit and explicit solvent models, to simulate thermal vibrations and the interaction with the water solvent, and a newly developed implementation of time-dependent density-functional (perturbation) theory, capable to compute the optical properties of molecular models of up to many hundred atoms in a wide frequency range. Building on our previous work on the optical properties of cyanin, we will address three main issues: i) the robustness of the color optical properties of anthocyanins with respect to the choice of the XC functional; ii) the dependence of the color optical properties on temperature; iii) as a grand-challenge application, the effect of co-pigmentation, to be simulated through an explicitly solvated model of a complexed anthocyanin, resulting from the pi-pi stacking between the dye backbone and other complex organic molecules covalently bonded to it.

Resource awarded: 26 000 000 core hours on FERMI @ CINECA, Italy

Thousands of trees for 4 billion years of life evolution on Earth

Project leader: Boussau, Bastien; CNRS, France

Collaborators: Vincent Daubin, CNRS, France; Vincent Miele, CNRS, France; Gergely Szollosi, CNRS, France; Eric Tannier, INRIA, France

Abstract: In the last fifteen years, thousands of genomes have been sequenced from species sampling the entire tree of life. Genomes contain a huge amount of information about how the diversity of life appeared and changed through time, and about how living systems work, from protein functions to ecological communities. However, revealing this information depends on a better understanding of the complex molecular and evolutionary processes through which it was recorded. Although genomic data holds a great promise for understanding the history of life and the principles that govern biological evolution, its complexity has so far impeded this quest. A key set of evolutionary mechanisms that need to be accounted for is how genetic information is duplicated, lost and sometimes transferred among organisms. One powerful approach to deciphering this information is to compare the genomes of different species in an evolutionary framework: each gene of a genome brings information on the evolution of species, and in return the evolution of species brings the key to understanding gene evolution and function. Because they contain thousands of genes, each of them having a different history, the analysis of genomes requires dedicated algorithms and vast amounts of computation, but methods that are tailored to use all the information included in the genomes are just starting to appear. In particular the reconstruction of gene trees and species trees on a genomic scale is a critical step for evolutionary studies, but is a very complex and costly task. We have developed a program to jointly infer gene and species trees by modeling events of gene duplication, loss, and transfer. Simulations and analyses on real data show that our method is accurate. It can also infer other key features of the history of the genomes under study, such as ancestral gene contents and speciation times, more accurately than commonly-used methods. This program can run on several processors simultaneously. Its scaling properties have been studied and improved on the supercomputer CURIE in 2012-2013 thanks to the PRACE 4th call. We have started using it on CURIE on a data set of 102 genomes from the three domains of life. We will reconstruct a species tree based on all the genes in these genomes, and infer gene trees and events of gene duplication, transfer and loss. In the process, we will obtain precise dates for speciation events billions of years old, where the lack of interpretable fossils renders other dating methods inaccurate. We will also reconstruct ancestral gene contents, thus illuminating metabolism evolution and consequent
ly the history of geochemical conditions on earth for the past 4 billion years. The results of these computations performed at an unprecedented scale will be the core data for a group of leading scientists in Evolution gathered in the context of the French ANR project ANCESTROME (funded from the 2011 call Bioinformatique, Investissements davenir). We will study the evolution of genome structure, metabolism, ecological communities, and we will create user-friendly databases to make the result of our computations available to the international scientific community.

Resource awarded: 34 000 000 core hours on Curie TN @ GENCI@CEA, France

SMOC – Submesoscale ocean MOdeling for Climate

Project leader: Capet, Xavier; CNRS, France

Collaborators: Christian Ethe, CNRS, France;Julien Jouanno, CNRS, France;Gurvan Madec, CNRS, France;Marina Levy, CNRS, France;Guillaume Roullet, Universite de Bretagne Occidentale, France

Abstract: Climate modeling has reached a point where biases in the ocean state and circulation are a primary concern. Some of its current deficiencies may only be resolved by explicit resolution of or improved parameterizations for meso- and submeso-scale (MSS) ocean turbulence. MSS processes are related to the ubiquitous presence of coherent mesoscale vortices (typical radii 30-100 km) and of cohorts of submesoscale fronts in their vicinity (typical transverse scale in the range 1-10 km). Because its typical time scales match those of atmospheric synoptic variability (hours to days) MSS is thought to impact ocean-atmosphere interactions and the mediation of atmospheric energy inputs into the ocean interior, with implications in terms of mixing and overall ocean functioning.

The Southern Ocean is a region where MSS processes are important at leading order, eg. for tracer budgets and momentum balances. Yet, there have been no studies where they are fully resolved so that their long-term effects can be accurately quantified. Our project proposes to remedy this by computing and analyzing state-of-the-art Southern Ocean numerical solutions at kilometer-scale horizontal resolution. Some simplifications of the domain geometry will allow us to take the most advantage of Tier-0 computer resources and compute decade-long solutions that will provide robust statistics on MSS turbulence and their impacts on the ocean functioning.

Resource awarded: 19 000 000 core hours on Curie TN @ GENCI@CEA, France

Physics of the Solar Chromosphere

Project leader: Carlsson, Mats; University of Oslo, Norway

Collaborators: Boris Gudiksen, University of Oslo, Norway; Viggo Hansteen, University of Oslo, Norway

Abstract: This project aims at a breakthrough in our understanding of the solar chromosphere by developing sophisticated radiation-magnetohydrodynamic simulations in order to interpret observations from the upcoming NASA SMEX mission Interface Region Imaging Spectrograph (IRIS).

The enigmatic chromosphere is the transition between the solar surface and the eruptive outer solar atmosphere. The chromosphere harbours and constrains the mass and energy loading processes that define the heating of the corona, the acceleration and the composition of the solar wind, and the energetics and triggering of solar outbursts (filament eruptions, flares, coronal mass ejections) that govern near-Earth space weather and affect mankind”s technological environment.

Small-scale MHD processes play a a pivotal role in defining the intricate fine structure and enormous dynamics of the chromosphere, controlling a reservoir of mass and energy much in excess of what is sent up into the corona. This project targets the intrinsic physics of the chromosphere in order to understand its mass and energy budgets and transfer mechanisms. Elucidating these is a principal quest of solar physics, a necessary step towards better space-weather prediction, and of interest to general astrophysics using the Sun as a close-up Rosetta-Stone star and to plasma physics using the Sun and heliosphere as a nearby laboratory.

Our group is world-leading in modelling the solar atmosphere as one system; from the convection zone where the motions feed energy into the magnetic field and all the way to the corona where the release of magnetic energy is more or less violent. The computational challenge is both in simplifying the complex physics without loosing the main properties and in treating a large enough volume to encompass the large chromospheric structures with enough resolution to capture the dynamics of the system. We have developed a massively parallel code, called Bifrost, to tackle this challenge. The resulting simulations are very time-consuming but crucial for the understanding of the magnetic outer atmosphere of the Sun.

Resource awarded: 34 560 000 core hours on SuperMUC @ GCS@LRZ, Germany

WETMD – Unravelling the Salvinia paradox: towards a new generation of superhydrophobic surfaces

Project leader: Casciola, Carlo Massimo; Sapienza University of Rome, Italy

Collaborators: Mauro Chinappi, Istituto Italiano di Tecnologia, Italy;Daniele Gentili, Sapienza University of Rome, Italy;Alberto Giacomello, Sapienza University of Rome, Italy

Abstract: Wetting on rough surfaces is a conundrum that challenges both the scientific and the technological communities. Remarkable macroscopic properties, collectively known as superhydrophobicity, e.g. high contact angle of liquid drops and self-cleaning features, emerge as a result of the combination of roughness and hydrophobic surface chemistry. The superhydrophobic Cassie state, for instance, is induced by air or vapor bubbles entrapped in the asperities. However, since the mechanism of wetting remains elusive, it is still difficult, if not impossible, to predict the loss of stability of such state (transition to the fully wet Wenzel state) and to design synthetic surfaces with the desired properties. Nonetheless surfaces exhibiting extraordinary features exist in nature. A remarkable example is the Salvinia molesta [Barthlott et al., Adv. Mater. 2010], a water fern that, due the presence of small hydrophilic patches on top of rough hydrophobic surfaces, is able to stabilize the air pockets of the Cassie state against pressure fluctuations. This example indicates that more fundamental investigations of wetting are needed to open the way to a second generation of advanced bio-inspired surfaces able to overcome the limitations of surfaces based on the so-called Lotus effect.

Molecular dynamics (MD), which can treat chemically complex systems and is based on minimal assumptions at atomistic level, is the tool of choice for unravelling this puzzle. However the Cassie-Wenzel transition (CWT) is a rare event that cannot be captured on the time scale amenable to a straightforward MD simulation. In WETMD project we propose to apply dedicated state-of-the-art statistical mechanics machinery for rare events to rigorously estimate the free-energy barriers associated to the
metastabilities of the different wetting states on realistic superhydrophobic surfaces at varying temperature and pressure. Our group recently demonstrated that restrained molecular dynamics is indeed capable to describe the CWT on a simple model system [Giacomello et al., Langmuir 2012]. However, going beyond this proof-of-concept simulations, with the goal of reproducing the chemical nature of real hydrophobic coatings and the complex morphology of natural surfaces, requires a boost in computational resources. Massively parallel simulations on top-notch machines are mandatory for mimicking the Salvinia leaves and revealing its strategies for air-trapping. This research has the potential to inspire radically innovative biomimetic surfaces, as well as to provide benchmarks for continuum models of wetting. Summing up, appealing to the full potentialities of Tier-0 computer architectures is the only possible way to understand the delicate effects at the basis of wetting metastabilities, thus inspiring next generation of superhydrophobic surfaces.

Resource awarded: 31 000 000 core hours on FERMI @ CINECA, Italy

PNPBind – Unraveling the mechanism of binding to Purine Nucleoside Phosphorylase of potent, transition-state analog inhibitors

Project leader: Cavalli, Andrea; Italian Institute of Technology, Italy

Collaborators: Andrea Spitaleri, Fondazione Centro San Raffaele, Italy;Sergio Decherchi, Italian Institute of Technology, Italy;Walter Rocchia, Italian Institute of Technology, Italy;Syeda Rehana Zia, Italian Institute of Technology, Italy

Abstract: In this project proposal, we aim at investigating the binding and the unbinding of two potent transition state analog inhibitors of the Purine Nucleoside Phosphorylase enzyme, a target of pharmaceutical interest. We use first long (in the microsecond time frame) unbiased molecular dynamics simulations to generate a sufficient number of binding events, which are expected to occur, in our simulative framework, in a few hundred of nanoseconds. Then, to simulate the unbinding event, which occurs on a timescale out of reach for unbiased simulation, we utilize enhanced sampling approaches. In particular, we will utilize the path collective variable coupled with metadynamics and/or umbrella sampling. Finally, to properly analyze the enormous amount of data we have developed an ad hoc clustering algorithm, which will allow us to extract relevant information in a user-independent manner. We are confident that our protocol could be of general applicability and can represent a further step towards the systematic exploitation of molecular dynamics approaches to drug design and discovery.

Resource awarded: 39,518,208 core hours on FERMI @ CINECA, Italy

GIANT-MOFS – Ab initio modelling of the adsorption in giant Metal-Organic Frameworks: from small molecules to drugs

Project leader: Civalleri, Bartolomeo; University of Torino, Italy

Collaborators: Guillaume Maurin, Universite” de Montpellier 2, FRANCE;Elisa Albanese, University of Torino, Italy;Jacopo Baima, University of Torino, Italy;Silvia Casassa, University of Torino, Italy;Massimo Delle Piane, University of Torino, Italy;Matteo Ferrabone, University of Torino, Italy;Migen Halo, University of Torino, Italy;Piero Ugliengo, University of Torino, Italy

Abstract: Metal-Organic Frameworks (MOFs) are a new class of materials which is expected to play a huge impact in the development of next-generation technologies. They consist of inorganic nodes connected through organic linkers to form a porous three-dimensional framework. The combination of different nodes and linkers makes MOFs very versatile materials with interesting and promising applications for gas capture, storage and separation. So far, the target of ab-initio modeling of the adsorption in MOFs has been small-to-medium size frameworks and their interaction with small molecules. Here, we aim at making a step forward by investigating the adsorptive capacity of the so-called giant MOFs, also denoted as mesoporous MOFs. Among giant MOFs, the most representative ones are denoted as MIL-100 and MIL-101. They are comprised of trimeric units of chromium(III) octahedral connected with 1,3,5-BTC (benzene-1,3,5-tricarboxylic acid) and 1,4-BDC (benzene-1,4-dicarboxylic acid), respectively. Because of the huge size of their unit cells, these giant MOFs represent a tremendous challenge for current ab-initio calculations. In this project, we will focus on MIL-100. MIL-100 has cages of different dimensions from micro- (6.5 ) to meso-pores (2530 ) and 2720 atoms in the primitive cell. It has a surface areas of 3340 m^2/g, which is 3 times larger than the values measured for the MCM-41 inorganic mesoporous materials.

Among the hottest topics in MOFs research it has been their application in clean energy and environmental protection, as for CO2 capture. An important consideration in maximizing the uptake of gases within porous MOF crystals is to increase the number of adsorptive sites within a given material. MIL-100 is characterized by the presence of a large number (i.e. more than 200) of coordinatively unsaturated (CUS) metal atoms which are exposed at the inner surface of the pores. Here, the adsorption of CO2 on the open metal site (primary site) and on the linker (secondary site) of isoreticular MIL-100 with different metals (e.g. Al and Fe) will be studied.

Beyond the fundamental interest to investigate this giant MOF in interaction with small molecules, there is a considerable interest in the adsorption of drugs for application in drug delivery [19]. Several drugs suffer from important drawbacks such as poor solubility and/or stability in the biological aqueous media. MOFs nanoparticles have been recently proposed to circumvent these drawbacks [20]. Among them, MIL-100(Fe), because of its nontoxicity, has been shown to be a very promising nanovector. Here, we intend to investigate the specific interaction of Busulfan, an antitumoral widely used in chemotherapy regimes for leukaemias, with the open metal sites in MIL-100. The atomistic details provided by these ab-initio calculations would allow us, for the first time, to understand the interaction forces acting between key drugs in the pharmaceutical field and an innovative drug carrier such as MIL-100 to an unprecedented level of accuracy.

As an important by-product, results from ab-initio calculations will be used to refine existing force fields, for MIL-100 and the adsorbed molecules, to be employed in classical molecular dynamics and Monte-Carlo (GCMC) simulations.

Resource awarded: 40 000 000 core hours on SuperMUC @ GCS@LRZ, Germany

Response of the Atlantic Ocean Circulation to Greenland Ice Sheet Melting

Project leader: Dijkstra, Henk; Utrecht University, The Netherlands

Collaborators: Henri Bal, Free University, The Netherlands;Frank Seinstra, Netherlands eScience Center, The Netherlands;Nicole Gregoire, SURFnet, The Netherlands;Walter Lioen, SURFsara, The Netherlands;Sandra Brunnabend, Utrecht University,
The Netherlands;Michael Kliphuis, Utrecht University, The Netherlands

Abstract: The Meridional Overturning Circulation (MOC) in the Atlantic Ocean is considered an important component of climate, because of the large northward heat transport associated with it. It is, however, far from being understood how climate would respond to a disruption of the MOC. What would be the impact on regional climate of such event? And how would the change of atmospheric conditions feed back to the circulation in the ocean?In this project, we want to compute the response of the MOC to meltwater input from the Greenland Ice Sheet in a high-resolution ocean-atmosphere model (the Community Earth System Model). In one of our recent studies (Weijer et al., 2012), a strong dependence of the response on the location of freshwater (e.g., near Greenland versus over a broad area in the North Atlantic) release was found in a strongly eddying ocean-only model, in contrast with the results from the equivalent low-resolution (non-eddying) model. We will compare results from the high-resolution CESM model to those from a low-resolution version of the same model and in particular focus on the role of ocean-atmosphere feedbacks on the MOC behavior and the impact of the MOC decrease on Western European climate. A major result from this project would be to establish (and understand) that the real (strongly eddying) MOC is much more sensitive to Greenland Ice Sheet freshwater anomalies than current (non-eddying) climate models indicate.

Resource awarded: 23 000 000 core hours on Hermit @ GCS@HLRS, Germany

HiResClim: High Resolution Ensemble Climate Modeling

Project leader: Doblas-Reyes, Francisco; Institut Catal de Cincies del Clima, Spain

Collaborators: Muhammad Asif, Institut Catal de Cincies del Clima, Spain;Domingo Manubens, Institut Catal de Cincies del Clima, Spain;Eric Maisonnave, CERFACS, France;Laurent Terray, CERFACS, France;Sophie Valcke, CERFACS, France;Wilco Hazeleger, Royal Netherlands Meteorological Institute (KNMI), The Netherlands;Camiel Severijns, Royal Netherlands Meteorological Institute (KNMI), The Netherlands;Uwe Fladrich, Swedish Meterological and Hydrological Institute (SMHI), Sweden;Colin Jones, Swedish Meterological and Hydrological Institute (SMHI), Sweden;Klaus Wyser, Swedish Meterological and Hydrological Institute (SMHI), Sweden

Abstract: HiResClim aims to make major advances in the science of estimating climate change and formulating climate predictions . This will be achieved by addressing the dual requirements of increased climate model resolution and increased number of ensemble realizations of future climate conditions over a range of time scales and for a set of plausible socio-economic development pathways. Increased model resolution aims to deliver a significant improvement in our ability to simulate key modes of climate and weather variability and, thereby, provide reliable estimates of future changes in this variability. The multi-model ensemble approach acknowledges the inherent uncertainty in estimating changes in climate over seasonal to centennial time scales, particularly in phenomena that are highly variable and, of which, changes in the occurrence of the rare but intense events are those impacting society and nature most strongly. To provide credible risk assessment in phenomena such as extra-tropical and tropical cyclones, heat waves, droughts and flood events to inform climate adaptation and climate services, the combination of high climate model resolution and a multi-model ensemble approach is unavoidable. In HiResClim we attack both of these requirements using a seamless multi-model climate modelling approach, which, as well as being the most efficient way to utilise the most advanced HPC systems of today and improve the realism of climate simulations, is also the only path to providing robust and actionable estimates of climate changes. The requirements of the project were defined considering the two state-of-the-art coupled model used, EC-Earth and ARPEGE-NEMO HR.

Resource awarded: 50 440 000 core hours on MareNostrum @ BSC, Spain


Project leader: Dormy, Emmanuel; CNRS, France

Abstract: The origin of the Earth”s magnetic field as well as the ones of planets, stars, galaxies remains an important and open physical problem. Current numerical modelling of this problem is severly limited in terms of number of CPU used, as all the numerical codes written for this problem rely on spectral expansion (which are not easily parallelized on a large number of nodes).

For this reason we have developped a new scheme based on a finite volume discretization and a semi-lagrangian transport algorithm. This code has been fully validated on the Curie Hybrid nodes thanks to a development account (10000h).

The new algorithm scales remarquably well on GPUs (with a factor 30 compared to CPUs). We now want to use this tool to physically understand the behavior of non-linear dynamo action in the limit of very large magnetic Reynolds numbers.

Resource awarded: 500 000 core hours on Curie H @ GENCI@CEA, France


MF-DISK: Protoplanetary disk dynamics: the multifluid magneto-rotational instability, gaps and jets

Project leader: Downes, Turlough; Dublin City University, Ireland

Collaborators: Wayne O”Keeffe, Dublin City University, Ireland;Donna Rogers, Dublin Institute for Advanced Studies, Ireland;Aleks Scholz, Dublin Institute for Advanced Studies, Ireland;Gilles Civario, Irish Centre for High End Computing, Ireland

Abstract: Virtually all low and intermediate mass stars form by accreting material from a surrounding molecular cloud through an orbiting accretion disk and onto the surface of the forming star, or young stellar object (YSO). Observations indicate that many of these disks are in Keplerian rotation, implying a balance between the centrifugal and gravitational forces. There is an, as yet unsolved, mystery surrounding the issue of how material can move inward through an accretion disk and onto the YSO. As the material moves inward it must lose angular momentum rapidly – if this did not happen the material could not move in toward the forming star at the observed rates and star formation would be extremely difficult. How, then, does the accreting material get rid of its angular momentum?

One possibility is that an instability such as the magnetorotational instability (MRI) could produce turbulence in the accretion disk which would, itself, create a highly effective viscosity. This viscosity would then act to transfer angular momentum from material at a particular point in the disk to material further out in the disk, thereby enhancing the rateof accretion. Accretion disks, though, are complex systems comprised of many chemical species and dust grains some of which are charged (and interact with magnetic fields), and some of which are neutral: in short, it is a multifluid system made up of various charged and neutral fluids. To gain a proper understanding of how turbulence in accretion disks behaves and how it can be generated we must move towards full multifluid modeling. With such modeling we will be able to self-consistently determine the regions of dust concentrations (of
relevance to planet formation) and the creation of disk “gaps” by the action of MRI.

We will use the state-of-the-art, massively parallel multifluid magnetohydrodynamic code HYDRA, coupled with power of the JUQUEEN system, to perform full 3D modeling of the dynamics of stratified accretion disks around YSOs. We will characterise the dynamics both in terms of the accretion rates achieved for various initial magnetic field configurations, and also in terms of the observational signatures produced by the simulations. The observational signatures will be derived a posteriori from the simulations using a dust radiative transfer code such as the open-source HYPERION code. The team for this project includes an accomplished astronomical observer with highly relevant expertise and we are optimistic about the conclusions of this work having high impact in both the theoretical and observational communities. HYDRA is well-tested and proven both from the point of view of the published physical results it has generated and in terms of its scalability: it has been shown to perform remarkably well in strong scaling tests up to 300,000 cores on the BlueGene/P architecture and up to at least 65536 cores on the JUQUEEN BlueGene/Q system.

Resource awarded: 16,300,000 core hours on JUQUEEN @ GCS@Jülich, Germany

KinOncoMut: The effect of oncogenic mutations on the conformational free energy landscape of kinases.

Project leader: Gervasio, Francesco Luigi; University College London, United Kingdom

Collaborators: Ludovico Sutto, University College London, United Kingdom

Abstract: We will investigate the role that a specific and widespread oncogenic mutation plays in reshaping the active/inactive equilibrium of four mutant kinases found in cancers such as melanoma, lung cancer and chronic myeloid leukemia. By comparing the free energy surfaces of wild-type and mutant kinases and characterising the most stable structures, we will shed light on the structural and chemico-physical principles that lead to a constitutively activated oncogenic kinases. This knowledge could greatly contribute to the development of more effective and selective drugs and to counter the emergence of resistance to drugs currently in clinical use.

Resource awarded: 18 000 000 core hours on MareNostrum @ BSC, Spain

Direct numerical Simulation of a spatially developing mixing layer with temperature gradient.

Project leader: Grasso, Francesco; Conservatoire National des Arts et Mtiers, France

Collaborators: Simon Marie, Conservatoire National des Arts et Mtiers, France;Matteo Bernardini, University of Rome la Sapienza, Italy;Sergio Pirozzoli, University of Rome la Sapienza, Italy

Abstract: This project aims at assessing the influence of the interaction between the supersonic hot jet (issuing at a nozzle exit) and the external subsonic cold flow. This subject is of great interest for the aerospace community, in particular for the understanding of the external unsteady loads that can be responsible for structural damage of the nozzle.This project will rely on direct numerical simulations that constitute a powerful tool to improve the physical knowledge of this type of turbulent flows. In order to evaluate the effects driven by the presence of temperature fluctuations, we will perform several computations with different values of temperature ratio between the upper and lower stream.

Resource awarded: 35,000,000 core hours on JUQUEEN @ GCS@Jülich, Germany

PHOTOSYSTEM2 – Water oxidation by photosynthesis: dynamics and reactivity of the manganese cluster in the Photosystem II complex explored by Quantum Mechanics / Molecular Mechanics simulations

Project leader: Guidoni, Leonardo; University of L”Aquila, Italy

Collaborators: Daniele Bovi, La Sapienza, Universit di Roma, Italy;Daniele Narzi, La Sapienza, Universit di Roma, Italy;Andrea Zen, La Sapienza, Universit di Roma, Italy;Matteo Barborini, University of L”Aquila, Italy;Emanuele Coccia, University of L”Aquila, ItalyMaria Montagna, University of L”Aquila, Italy;Fabio Pitari, University of L”Aquila, Italy

Abstract: The most difficult step in photosynthetic solar energy conversion performed by plants, algae and cyanobacteria is the splitting of water into molecular oxygen and hydrogen equivalents. To achieve this challenging catalytic step photosynthetic organisms used a special protein complex: the Photosystem II (PSII). The light induced oxidation of the water in PSII is catalysed by its Mn4Ca catalytic core proceeding by the accumulation of four oxidizing equivalents through five (S0 – S4) oxidation states known as Kok’s cycle. Recently, a new crystallographic structure of the PSII [Umena et al. Nature, vol. 473, p. 55, 2011] has revealed for the first time the three-dimensional molecular arrangement of the complex with high-precision atomistic details, opening the way to first-principles computational studies of its catalytic mechanisms. The deep understanding of the way Nature has chosen to perform efficiently this difficult task has a great relevance not only for Biology but also for inspiring the development of biomimetic artifcial systems that can be used to store solar energy in an environmentally-friendly way. In the present project we use ab initio molecular dynamics simulations (AIMD) based on broken-symmetry DFT+U approach combined with a Quantum Mechanics / Molecular Mechanics scheme to investigate the properties and reactivity of the manganese cluster of PSII immersed into its complex protein environment. Using unconstrained AIMD ans calculations of free-energy barriers and minimum energy paths, we will investigate several aspects of this system: the effects of the protein environment and the dynamics of the Mn cluster in its high-spin and low-spin states of S2; the interconversion free-energy barriers between the latter states and their relative stability; the S3 and S4 states and the formation of the O-O bond; the role of the hydrogen bonding network in the proton transfer around the active site.Thanks to PRACE infrastructure, the fulfilment of the present project will allow us for the first time to unravel with high-level atomistic description the mechanisms of working of Photosystem II, the protein complex that is changing the earth atmosphere and sustaining the life on our planet since 2.5 billion years.

Resource awarded: 6 508 160 core hours on Curie FN @ GENCI@CEA, France

MHBSP – Multi-scale heterogeneous bulk-synchronous programming

Project leader: Hains, Gaetan; Universit Paris-Est Crteil (UPEC), France

Collaborators: Frederic Gava, Universit Paris-Est Crteil (UPEC), France;Chong Li, Universit Paris-Est Crteil (UPEC), France;Sovanna Tan, Universit Paris-Est Crteil (UPEC), France;Julien Tesson, Universit Paris-Est Crteil (UPEC), France;Muath Alrammal, Universite d”Orleans, France;Mostafa Bamha, Universite d”Orleans, France;Helene Coulomb, Universite d”Orleans, France;Sylvain Jubertie, Universite d”Orleans, France;Jeoffrey Legaux, Universite d”Orleans, France;Sebastien Limet, Universite d”Orleans, France;Frederic Loulergue, Universite d”Orleans, France

Abstract: We develop since the early 1990s several parallel-programming libraries and parallel execution models allowing high-level scalable application development. They are based on past experience with Bulk-synchronous parallel (BSP) programming and BSP language design. Novel features are motivated by the last decade”s move towards multi-level and heterogeneous parallel architectures invoving multi-core processors, graphics accelerators and hierarchical routing networks in the largest multiprocessing systems. The aim of this project is to test our parallel programming primitives and libraries at a level of hundred thousands of cores and a wide variety of realistic hardware setups, with a very-high level of internal network performance. This should confirm or refine our high-level programming and performance model closer to peta-Flop levels and to experiment with more degrees of architecture decomposition for enhanced performance than with previous “flat” parallel programming where the hardware hierarchy was not modeled.

Resource awarded: 66 000 core hours on Curie FN @ GENCI@CEA, France 6 000 core hours on Curie H @ GENCI@CEA, France 480 000 core hours on Curie TN @ GENCI@CEA, France 678 000 core hours on Hermit @ GCS@HLRS, Germany 2 700 000 core hours on JUQUEEN @ GCS@Jülich, Germany


Project leader: Hakkinen, Hannu; University of Jyvaskyla, Finland

Collaborators: Xi Chen, University of Jyvaskyla, Finland;Lars Gell, University of Jyvaskyla, Finland;Lauri Lehtovaara, University of Jyvaskyla, Finland;Sami Malola, University of Jyvaskyla, Finland

Abstract: In this project, large-scale quantum chemical and molecular dynamics simulations are employed to investigate binding of 1 – 3 nm -diameter functionalized gold nanoclusters to viruses. Information about binding mechanisms and effects of clusters on the dynamics and stability of viruses is of utmost importance in order to understand mechanisms of virus uncoating (release of the viral genome via opening of the virus capsid) and may help development of stabilized viruses to be used in novel vaccines against viral diseases.

Resource awarded: 43 171 840 core hours on Hermit @ GCS@HLRS, Germany

Three-Dimensional Simulations of Core-Collapse Supernova Explosions of Massive Stars Applying Neutrino Hydrodynamics

Project leader: Janka, Hans-Thomas; Max-Planck-Institut fuer Astrophysik, Germany

Collaborators: Florian Hanke, Max-Planck-Institut fuer Astrophysik, Germany;Tobias Melson, Max-Planck-Institut fuer Astrophysik, Germany;Bernhard Mueller, Max-Planck-Institut fuer Astrophysik, Germany;Andreas Marek, Rechenzentrum der Max-Planck-Gesellschaft, Germany

Abstract: Supernovae are gigantic stellar explosions that terminate the lives of massive stars and give birth to neutron stars and black holes. They belong to the most spectacular and brilliant cosmic events, produce potentially measurable neutrino and gravitational-wave signals, and are among the prime candidates for the still mysterious sources of roughly half of the heavy, neutron-rich chemical elements beyond iron. The exact processes that lead to the explosive disruption of a star are not satisfactorily understood yet, but deeper theoretical insight is indispensable to better define the role of supernovae in the evolution of stars and galaxies and as celestial laboratories of nuclear and particle physics at extreme conditions. Neutrinos are thought to play a central role in the solution of this important problem. They are produced in gigantic numbers when the degenerate core of a massive star ultimately reaches its stable mass limit and collapses catastrophically within less than a second to an extremely hot and dense neutron star. The neutrinos carry away the huge gravitational binding energy released in the collapse, but some of them are absorbed in the cooler, infalling stellar material around the nascent neutron star. If this neutrino heating deposits enough energy, a shock wave can be accelerated to expel the outer stellar layer in the supernova blast. But is the energy transfer by neutrinos sufficiently efficient?An answer to this question heavily rests on computer simulations, and the complexity of the involved processes poses a grand challenge for numerical modeling. The transport and reactions of neutrinos in the dense stellar medium are highly complicated and computationally extremely demanding, and hydrodynamical instabilities lead to violent nonradial flows and turbulence that require multi-dimensional modeling. With the most sophisticated computations performed in two dimensions (2D) so far, we could confirm the viability of the neutrino-driven mechanism in a growing number of stellar models over the past years.However, three-dimensional (3D) simulations are needed to overcome the artificial constraint of axisymmetry of the 2D geometry. Thanks to a PRACE 4th Call Tier-0 grant we were recently able to perform the very first 3D simulations with the high sophistication of the neutrino physics applied only in 2D models before. With these spearheading simulations we could demonstrate that violent, large-scale hydrodynamic instabilities, which facilitate explosions in 2D, also occur in the 3D case. Moreover, we discovered the development of a strong spiral-mode instability, whose existence was only a speculative possibility hitherto. These instabilities support the shock expansion and thus provide crucial aid for the neutrino-heating mechanism. But 2D simulations showed that the core-collapse and shock dynamics vary drastically with the structure of the progenitor star. In this follow-up project we therefore intend to take the necessary next step and plan to expand our exploration to a larger variety of stellar progenitor models, whose selection will be guided by existing, successfully exploding 2D runs. This extended set of 3D models shall help clarifying the still barely understood role of the newly established spiral mode.

Resource awarded: 97 800 000 core hours on Curie TN @ GENCI@CEA, France 48 900 000 core hours on SuperMUC @ GCS@LRZ, Germany

Aerodynamics and Aeroacoustics of Complex Geometry Hot Jets

Project leader: Jefferson-Loveday, Richard; University of Cambridge, United Kingdom

Collaborators: Ann Dowling, University of Cambridge, United Kingdom;Tom Hynes, University of Cambridge, United Kingdom;Vadlamani Rao, University of Cambridge, United Kingdom;Paul Tucker, University of Cambridge, United Kingdom;James Tyacke, University of Cambridge, United Kingdom

Abstract: The objective of this project is to take a model for turbulent jet noise (previously developed in GR/S43191/01) and extend the model to hot jets and complex nozzles configurations such as chevrons and a co-axial jet with a pylon and flap, which will be of immediate interest to engine manufacturers. The methodology of the project is to compute the propagation of noise to the far field by solving the Linearised Euler Equations (LEE), describing how sound emitted by the jet is modified by propagation through the time-averaged but spatially varying jet flowfield. We intend to characterise the acoustic sources using 4th order correlation coefficients of velocity. These will be informed from the high-order LES simulations in terms of turbulence correlation coefficient shape and the degree of anisotropy of the turbulence field. We intend to run high-order LES simulations with mesh densities up to 1 billion cells on several tens of thousands of processing cores.

Resource awarded: 33 000 000 core hours on Hermit @ GCS@HLRS, Germany

Large eddy simulation of large scale coal and biomass combustion

Project leader: Kempf, Andreas; Universitt Duisburg-Essen, Germany

Collaborators: Miriam Rabacal, Instituto Superior Tecnico, Universidade Tcnica de Lisboa, Portugal

Abstract: In the foreseeable future, coal will continue to be the dominant fuel for electricity production, causing significant amounts of pollution. Some studies have shown that biomass and coal co-firing has the potential to be a near-term, low-risk, low- cost and sustainable energy development to reduce pollutant emissions from existing pulverised coal fired power plants. Co-firing technology, however, faces some technological and operational problems due to interactions between biomass and coal particles during combustion. Reaching high co-firing ratios (in energy basis) is a very challenging task. Only few experimental studies have fully addressed these issues and typically for low co-firing low ratios, below 20% on a thermal basis. The investigation of very high co-firing ratios (up to 50%) is of utmost importance for burner manufacturers and power plant operators, and is the aim of this work. The in-flame environment of large-scale coal and biomass combustion is harsh, making detailed experiments almost impractical so that many questions are still open in the modeling of biomass combustion. Typical numerical studies available in the literature for large-scale furnaces use classical models with low accuracy. Large eddy simulations have been emerging as a more advance technique for coal combustion, but studies often target simple laboratory jets and in the case of large domains, flow-field symmetry is imposed to make the simulation affordable. The present application aims to simulate large-scale coal and biomass flames with a typical power of 100 kW using state of the art models considering the interaction between coal and biomass, which present distinct properties. Such applications possesses a broad range of scales, owing to a very high ratio between the furnace diameter, the injection slot diameter, and the size of the coal particles. The project lies within a cooperation between Instituto Superior Tecnico (IST), Universidade Tecnica de Lisboa, Portugal, and University of Duisburg Essen (UDE), Germany, targeting the co-firing coal and biomass studies within a holistic approach, both experimentally and numerically. The UDE team, applying for this proposal, is in charge of the development and testing of the numerical models. Within the framework of this application, the in-house flow solver PsiPhi will be used for the numerical simulations by a team experienced with many issues of Large Eddy simulation, particle transport, and the mathematical modeling of combustion, devolatilisation, radiation and char burnout. The results will provide important insight on physical and chemical phenomena on locations where reliable and quality measurements are very hard to obtain, if not impossible. The accurate capturing of the physics will allow for a more in-depth understanding of the pollutant formation mechanisms.

Resource awarded: 15 400 000 core hours on SuperMUC @ GCS@LRZ, Germany


Project leader: Lindorff-Larsen, Kresten; University of Copenhagen, Denmark

Collaborators: Kaare Teilum, University of Copenhagen, Denmark;Francesco Luigi Gervasio, University College London, United Kingdom;Michele Vendruscolo, University of Cambridge, United Kingdom

Abstract: Proteins are biological macromolecules that play a central role in biology, biotechnology and medicine. The last 50 years in protein science have provided us with a plethora of atomic resolution structures of proteins that are not only stunningly beautiful, but have also provided crucial insight in to the mechanisms by which proteins function.Proteins are, however, also dynamic molecules and recent years have seen an explosion in our ability to characterize protein motions using both computations and experiments. Importantly, we now know that the way a protein moves can have a large impact of the function of the protein, and that it is not just the structure but also the dynamics of a protein that determines its function. Examples include how motions in enzymes aid in substrate binding and release, or in conformational changes that enable protein molecules to transduce signals in cells.From an experimental point of view, nuclear magnetic resonance (NMR) spectroscopy has emerged as the central most important technique that can be used to study protein dynamics. NMR spectroscopy has the unique ability to provide atomic level data that reports on the structure, dynamics and thermodynamics of the motions in proteins. A relatively recent development in NMR is the ability to study the structure and dynamics of larger proteins, and protein complexes. Another exciting area of development is in experimental studies of so-called intrinsically disordered proteins. These are proteins that display an unusually large amount of dynamics, and which do not attain a single well-defined structure in the cell. 
Simultaneous with these developments in experiments, the last few years have also seen important progress in our ability to use molecular dynamics simulations to study the structure and dynamics of proteins. This progress has been enabled in part by substantial improvements in both the quality and accuracy of the force fields used in simulations, and in methods that allow for increased sampling of the conformations of proteins.We suggest utilizing new computational methods that integrate the strengths of NMR spectroscopy and molecular dynamics simulations to study the structural dynamics of proteins. If successful, the results may provide new insights in many of the areas where protein dynamics affect function, including understanding biological systems and designing new, more selective drug molecules towards proteins or improved enzymes.

Resource awarded: 11 289 600 core hours on Curie TN @ GENCI@CEA, France

CSSRC – Complete statistical simulation of reverberation chamber

Project leader: Moglie, Franco; Universita` Politecnica delle Marche, Italy

Collaborators: Andrea Cozza, Supelec, France;Valter Mariani Primiani, Universita` Politecnica delle Marche, Italy;Giuseppe Ferrara, Universita` degli Studi di Napoli Parthenope, Italy;Angelo Gifuni, Universita` degli Studi di Napoli Parthenope, Italy;Maurizio Migliaccio, Universita` degli Studi di Napoli Parthenope, Italy;Antonio Sorrentino, Universita` degli Studi di Napoli Parthenope, Italy;Alistair Duffy, De Montfort University, United Kingdom;Gabriele Gradoni, University of Maryland, College Park, United States

Abstract: Nowadays, several international standards allow use of reverberation chambers intended for various electromagnetic compatibility testing. Their electromagnetic use began 30 years ago, and a lot of improvement in the theoretical and experimental understanding has been achieved up to now.The reverberation chamber is a structure for generating chaotic electromagnetic fields with collective properties suitable for performing robust, repeatable and realistic measurements in the radio-frequency and microwave ranges. An electromagnetic cavity is said to reverberate if statistical isotropy, homogeneity and depolarization occur at least over a subspace of the chamber volume.The reverberation chambers are essentially electrically large overmoded cavities with variable geometry, made of highly-reflective metallic walls and excited by a source. It must be pointed out that the internal source is electrically small with respect to the greater dimension of the chamber. A static cavity becomes reverberating when geometry is mechanically changed, causing the mode stirring mechanism.The numerical simulation of a reverberation chamber requires burden computational resources because it has:1) electrically large dimensions that implies a large number of elementary cells to completely discretize the volume;2) high quality factor that implies an investigation of a long time range or a high frequency resolution;3) moving elements that implies a variation of the geometry.

The objective of this project is to analyze and optimize the statistical proprieties of a reverberation chamber as function of its geometry and stirring mechanism. We will use our home made code previously developed and optimized for high-performance parallel computers during other HPC projects.The code is mainly divided in three modules:1) an electromagnetic time domain solver based on finite difference time domain (FDTD) method;2) a fast Fourier transform module to obtain the frequency domain behavior;3) a statistical module to obtain the reverberation chamber proprieties like uncorrelated stirrer positions, field uniformity and statistics.All this modules can run on an unique executable job, avoiding the storage of large data on the hard disks. All the modules was previous optimized for high-performance parallel computers using hybrid method (MPI and OpenMP).Our groups have more than one reverberation chamber and an experimental validation will be done.On the other hand, simulations can produce a larger amount of data with the respect to experiments in a shorter time.

The innovation potential of this research is to better understand the chaotic proprieties of reverberation chamber and their dependence on geometry and stirring mechanism.The availability of an optimized simulation code will give the results in short time avoiding long measurement campaigns.The results will be useful for all the engineering fields that employ reverberation chamber like1) automotive, aeronautic and astronautic industries,2) electromagnetic compatibility of electronic devices (immunity and radiation tests),3) characterization of WiFi propagation and devices in indoor environments,4) analysis of absorbing and shielding proprieties of materials, in particular carbon-based nano-structured materials and devicesIn addition, the reverberation chamber is a chaotic structure and the results of this research will be useful to the wave chaos group of physic community.

Resource awarded: 30 000 000 core hours on FERMI @ CINECA, Italy

DISTRO – DIrect numerical Simulation of Turbulent channel flow at high ReynOlds number

Project leader: Pirozzoli, Sergio; Sapienza, University of Rome, Italy

Collaborators: Matteo Bernardini, Sapienza, University of Rome, Italy;Paolo Orlandi, Sapienza, University of Rome, Italy

Abstract: The present research project aims at improving our knowledge of the physics of wall-bounded turbulent flows through a database obtained by direct numerical simulation (DNS) of the incompressible flow in a rectangular channel at high Reynolds number. In terms of the friction Reynolds number Ret (ratio between the channel-half height, h, and the viscous length-scale, dv), we will perform a simulation corresponding to Ret = 6000. These value represents a substantial increase with respect to those currently available in literature (Ret < 2000), and it will allow to observe physical effects pertaining to the asymptotic high-Re regime. We expect that the project will have a strong impact from the scientific point of view, bringing important contributions to the understanding of unsolved key issues, including: i) characterization of the large turbulent structures of the outer layer and of their interaction mechanisms with the small-scale structures of the near-wall region; ii) numerical evidence of the effective existence of a log region, never clearly observed in a numerical computation. The DNS database will be made available to the fluid dynamics community through the Web. Users will be allowed to access the data, to explore and use them for their own research aims.

Resource awarded: 50 000 000 core hours on FERMI @ CINECA, Italy

NURESAFE – Nuclear Reactor Safety simulation platform

Project leader: Reboux, Sylvain; ASCOMP GmbH, Switzerland

Collaborators: Daniel Caviezel, ASCOMP GmbH, Switzerland;Mathieu Labois, ASCOMP GmbH, Switzerland;Djamel Lakehal, ASCOMP GmbH, Switzerland;Chidambaram Narayanan, ASCOMP GmbH, Switzerland

Abstract: This project is part of the NURESAFE initiative for nuclear safety.

The objective is to develop a global modelling framework for multi-scale core thermal-hydraulics in Pressurized Water Reactors (PWR).

Detailed computational science in this context amounts at using advanced CPU-demanding Computational Fluid Dynamics (CFD) simulations to understand basic heat and fluid flow phenomena and derive practical and useful models to be implemented in (i) CFD codes for component scale, and (ii) system codes for the global reactor behaviour. The numerical models need to be capable of dealing with thermal-hydraulics phenomena acting at several levels: from system-level and component-level scales down to the length scales of the bubble layer adjacent to the hot fuel-rod, and further down to the entrained droplets from ligaments and sheets forming at wall liquid films.

We will separately employ detailed computational techniques/models (namely DNS and LES for turbulence, and Interface Tracking Methods (ITM) and Lagrangian particle tracking for multiphase flow representation) to infer coarse-grained interfacial momentum and heat transfer model
s for phase-averaged Eulerian prediction techniques (Homogeneous Algebraic Slip – ASM, or Two-Fluid). The work will consist in (i) developing flow databases and (ii) explore these databases to infer practical coarse- grained models. This is known as model upscaling or coarse graining.

The numerical models will be validated against two experimental databases: the MIT convective boiling problem and the multiphase flow transition in a vertical pipe of Nottingham University. Both experiments provide valuable data for comparisons, with pertinent physics to the problems in question, while simple enough in terms of geometry to be within reach of simulation using ITM and phase-averaged models. For each problem, we will first compare the data to ITM simulations then compare the data to phase-average models: two-fluid and mixture ASM for two-phase flow in the pipe, and boiling models for the convective flow problem.

The simulations will include resolution of all the important phenomena including how bubbles grow, depart and move along and away from the wall, and how and where they re-condense as they move into the colder liquid in the core of the channel. This entails calculating the temperature and velocity distributions as well as tracking the vapor/liquid interface of each bubble within the whole domain. These scale-resolving unsteady simulations require tremendous computational resources and storage.

The expected breakthrough outcomes are two folds: (1) to develop beyond state-of-the-art computational techniques (multiscale, multiphysics) which have never been undertaken hitherto, and (2) develop new models for heat transfer and interphase momentum exchange or revisit existing ones by means of detailed predictions techniques (LES, DNS and ITM) based on the new computational techniques developed within (1). Besides this, new benchmark exercises will be proposed with rich databases of flow to be used for model validation. The detailed simulations and benchmarks will provide data that are beyond reach of experiments, e.g. turbulent stresses very near the walls, correlations between turbulence and interface dynamics, etc.

Resource awarded: 11 000 000 core hours on Hermit @ GCS@HLRS, Germany

The Effect of Titanium-Derived Surfaces in modulating Protein Folding

Project leader: Reuter, Karsten; Technical University Munich, Germany

Collaborators: Julian Schneider, Technical University Munich, Germany;Emanuel Peter, University of California Santa Barbara, United States;Joan-Emma Shea, University of California Santa Barbara, United States

Abstract: The overarching aim of this proposal is to study the underlying physical principles that govern the folding of a small protein in the presence of modified titanium-based surfaces. It has been shown that e.g. a graphene surface denatures a Trp-Zip2 protein due to a combination of pi-pi interactions between the Trp residues and the surface, as well as due to a contribution related to the structuring of water near the graphene interface. Moreover, we have found that natively oxidized titanium surfaces also give rise to a very characteristic ordering of water, which strongly interacts with adsorbed peptide molecules. We anticipate that modified Titanium, in contrast to graphene, may enable the folding of this protein. We seek to determine how the solvent-environment at interfaces as well as the protein binding affinity is governed by the surface properties, with the long-term goal of designing artificial, synthetically modified chaperone surfaces. We plan to achieve this goals by using replica exchange molecular dynamics simulation to sample the conformational phase space of the protein at the surface. These technique requires a large number of replicas to be simulated in parallel, thus posing large demands towards the computational resources. By the including defects, doping, or covalent functionalization with of organic molecules we plan to modify the surface properties towards the desired behaviour. We expect that each investigated modification increases our understanding of the (un-)folding process at the surface, which allows for an increasingly rational design of new modifications.

Resource awarded: 6 600 000 core hours on Curie FN @ GENCI@CEA, France

Simulation of thermal environment of industrial combustors and comparison with experiments

Project leader: Roux, Anthony; TURBOMECA, France

Collaborators: Antoine Dauptain, CERFACS, France;Laurent Gicquel, CERFACS, France;Florent Duchaine, CERFACS, France;Sandrine Berger, TURBOMECA, France;Dorian Lahbib, TURBOMECA, France;Thomas Lederlin, TURBOMECA, France

Abstract: This project is concerned with the Large Eddy Simulation of an industrial combustor including the heat losses by radiation and conduction in solid parts. This is the natural extension from previous work which focused solely on the fluid dynamics on the turbine with adiabatic thermal conditions. The study will first focus on modelling the thermal properties of the main cooling component of a gas turbine: multi perforated plates. Then this model will be applied to a real gas turbine demonstrator using LES where an experimental thermal response is available. The LES will also include the assessment of radiation’s impact on the flame and the flow.

Resource awarded: 15 000 000 core hours on Curie FN @ GENCI@CEA, France

LAIT – Light-harvesting in the time domain

Project leader: Rozzi, Carlo Andrea; CNR – Istituto Nanoscienze, Italy

Collaborators: Joseba Alberdi Rodriguez, University of the Basque Country UPV/EHU, Spain;Alain Delgado Gran, CNR – Istituto Nanoscienze, Italy;Stefano Pittalis, CNR – Istituto Nanoscienze, Italy

Abstract: Organic and hybrid solar cells are very complex systems, setting a formidable challenge for first-principles modeling. They raise both fundamental scientific questions, and technological problems. The main difficulty in the calculations consists in the fact that the chain of phenomena involved in the photovoltaic conversion of energy occurs on a wide spatial and temporal scale. Although the isolated constituent units of a cell might be appropriately described by accurate quantum-chemical methods, this is not the case when such units are linked, and eventually included in a complex liquid interacting environment. The computational cost of such methods simply makes them impossible to be applied to large systems.The main goal of this project is to obtain a first principle time-resolved description of the light-harvesting process in nano-sized units for photo-energy conversion in dye-sensitized solar cells. The theoretical investigation is carried out in close collaboration with experimental and industrial partners, responsible of validating and testing the results. This way the theoretical and computational work will be carried out on materials for which it is granted the the possibility of performing time-resolved state of the art spectroscopic measurements, and actively interacting with t
he industrial department in charge for producing, characterizing and commercializing the samples.To pursue the goal of the project a suitable compromise between accuracy and computational cost of the simulations must be reached. The theoretical tool of choice for the project is Time-Dependent Density-Functional Theory (TDDFT). TDDFT has been successfully employed in the past to describe excitations and optical properties of a wide variety of systems. It is computationally much more efficient than standard quantum chemical methods, which makes it particularly suitable for large-scale computing. In particular, as implemented in the real-space, real-time open source code “octopus” TDDFT offers a unique opportunity to investigate the excited states of the systems on a rigorous ground in a supercomputing environment.

Resource awarded: 20 000 000 core hours on FERMI @ CINECA, Italy

LHC-ABS – The optical absorption spectra of a real Light Harvesting Complex from first-principles: the spinach case

Project leader: Rubio, Angel; European Theoretical Spectroscopy Facility, Spain

Collaborators: Joseba Alberdi-Rodriguez, Universidad del Pais Vasco/Euskal Herriko Unibertsitatea UPV/EHU, Spain;Miguel Marques, CNRS, France;Fernando Nogueira, University of Coimbra, Portugal ;Micael Oliveira, University of Coimbra, Portugal ;Bruce Milne, University of Coimbra, Portugal; Xavier Andrade, Harvard University, United States

Abstract: More than 50% of the light captured by green plants for use in photosynthesis is absorbed by the light harvesting complex (LHCII) . LHCII is a trimeric protein assembly with three-fold symmetry. Each monomer unit contains, in addition to the protein, 14 chlorophyll and 4 xanthophyll chromophore molecules which are the key functional units of the LHC system and perform the actual light-harvesting. In all, LHCII contains approximately 16,900 atoms, 7000 of which belong to the chromophores. Although usually bound within cell membranes in vivo, LHCII is a robust structure and survives intact the processes of purification and crystallisation meaning that the crystal structure is known accurately and also that it has become possible to perform experimental studies of the isolated LHCII in a fairly routine manner.

LHCII has for decades been the focus of experimental research aimed at understanding the biological aspects of what is truly the power-house of life on Earth and arguably the most important proteinaceous structure in Nature. More recently, the extremely high light-capturing quantum efficiency of LHCII has attracted a great deal of interest from a completely different group of scientists, namely those working toward developing solar energy capture devices.

After absorption of light the excited-state energy is transferred from the LHCII antenna to the photosynthetic reaction center. At low light levels the quantum efficiency of this process approaches unity. Models based on excitonic coupling between the chlorophyll units and/or modulation of the chromophores’ energy levels by the protein environment have been invoked in an attempt to explain the high efficiency of this process. Whereas in bacterial light-harvesting antennae the former, excitonic, mechanism appears to be dominant, there is evidence pointing towards a more micro-environment driven mechanism being mostly responsible for the super-radiance observed in plant LHCII. Although a balance between both effects will contribute to the functioning of LHCII, it is extremely difficult to experimentally discern the relative importance of the two.

In this project we propose to study the various contributions made by each of the building blocks of the light-harvesting antenna to its characteristic electronic structure and optical properties by performing simulations from first-principles of the optical response of the full LHCII complex and of its components using time-dependent density functional theory. Computationally we can easily separate the protein, chlorophyll and xanthophyll assemblies and study them in isolation and at various levels of combination whilst maintaining them at the original geometries found in the complete complex. Thus, it will be possible to clearly separate the spectral features corresponding to the separate components and assess how the interplay between them leads to the building up of the optical response of the whole system.

Resource awarded: 21 560 000 core hours on MareNostrum @ BSC, Spain

MULTIPORE – Dynamics of multi-component Fluid dynamics in porous structures

Project leader: Sbragaglia, Mauro; Univ. Tor Vergata, Italy

Collaborators: GIorgio Amati, Univ. Tor Vergata, Italy;Luca Biferale, Univ. Tor Vergata, Italy;Eric Foard, Univ. Tor Vergata, Italy;Patrizio Ripesi, Univ. Tor Vergata, Italy;Riccardo Scatamacchia, Univ. Tor Vergata, Italy;Federico Toschi, Tecn. Univ. Eindhoven, The Netherlands

Abstract: The dynamics of multi-component and/or multi-phase flows in porous media is crucial for many disciplines and many applications as, e.g., for oil recovery techniques enhancement, the dispersion of pollutants in aquifers, filtration and centrifuging in many chemical engineering applications and wetting of feathers, the air dynamics in lungs and the design of bone tissueperfusion microreactors.In this project, we aim to perform a systematic investigations of multi-component and/or multi-phase flows dynamics in porous matrices adopting a bottom-up, multi-scale approach, based on a Lattice Boltzmann Method. The final goal is to go beyond the actual state-of-theart for simulating porous media dynamics, exploring crucial effects for the local/globalresponse of the system as: (i) Heterogeneous wettability distributions (ii) Non zero slip velocity properties at pore boundaries (iii) A strong heterogeneous pore matrix geometrical structure.

Resource awarded: 9 150 160 core hours on Hermit @ GCS@HLRS, Germany

ContQCD — The continuum limit of QCD with up, down and strange quarks

Project leader: Schaefer, Stefan; CERN, Switzerland

Collaborators: Martin Luescher, CERN, e;Fabio Bernardoni, DESY, Germany;Piotr Korcyl, DESY, Germany;Stefano Lottini, DESY, Germany;Alberto Ramos, DESY, Germany;Hubert Simma, DESY, Germany;Rainer Sommer, DESY, Germany;Patrick Fritzsch, Humboldt Universitaet zu Berlin, Germany;Tomasz Korzec, Humboldt Universitaet zu Berlin, Germany;Ulrich Wolff, Humboldt Universitaet zu Berlin, Germany;Gunnar Bali, Universitaet Regensburg, Germany;Vladimir Braun, Universitaet Regensburg, Germany;Sara Collins, Universitaet Regensburg, Germany;Benjamin Glaessle, Universitaet Regensburg, Germany;Meinulf Goeckeler, Universitaet Regensburg, Germany;Paula Perez-Rubio, Universitaet Regensburg, Germany;Enno Scholz, Universitaet Regensburg, Germany;Wolfgang Soeldner, Universitaet Regensburg, Germany;Andre Sternbeck, Universitaet Regensburg, Germany;Francesco Knechtli, University of Wuppertal, Germany;Jochen Heitger, Westfaelische Wilhelms-Universitaet Muenster, Germany;Michele Della Morte, IFIC (CSIC) Valencia, Spain;John Bulava, Trinity College Dublin, Ireland;Mauro Papinutto, Universita` di Roma “La Sapienza” & INFN sezione di Roma, Italy

Abstract: The parameters of the standard model of particle physics need to be determined from a matching of quantities computed within this theory to quantities determined in experiments.This project aims at computing some of these parameters at a new level of rigor: the strong coupling constant, the masses of strange and charm quark and also the decay constants of charmed mesons needed to extract matrix elements of the CKM matrix.This will be achieved by calculations in lattice QCD, which for these observables is the method that delivers the highest level of accuracy.

Determinations of fundamental parameters with minimal experimental input are an important ingredient into precision tests of the theory, since they probe very particular aspects, complementary to the analysis of experimental high energy physics data: while the strong coupling constant can, e.g., be extracted from fits of perturbation theory to scattering data, once this quantity is known from a lattice QCD calculations, the same data can serve as a very stringent test of the theory, augmenting its predictive power. The accuracy and rigor which we will be able to achieve with these simulations will be at thelevel required for such tests.

This project is part of the CLS effort involving lattice field theorists from all over Europe to perform computations in lattice QCD in a coordinated way, aiming at high quality results. The data obtained in these simulations will therefore lay ground for further results beyond the observables of this proposal.

Resource awarded: 39 910 000 core hours on SuperMUC @ GCS@LRZ, Germany

MODYQA – Molecular dynamics simulation of liquid water by quantum Monte Carlo

Project leader: Sorella, Sandro; SISSA, Italy

Collaborators: Ye Luo, SISSA, Italy;Guglielmo Mazzola, SISSA, Italy;Andrea Zen, University of Rome, Italy;Matteo Barborini, Università degli Studi de L”Aquila, Italy;Emanuele Coccia, Università degli Studi de L”Aquila, Italy;Leonardo Guidoni, Università degli Studi de L”Aquila, Italy

Abstract: The correct description of the structural and dynamical properties of liquid water in molecular simulations represents a challenge for computational physicists and chemists since the early days of molecular dynamics.So far, first principles molecular dynamics simulations based on density functional theory have significant discrepancies with respect to experimental data. For instance the calculated oxygen-oxygen radial distribution function is over structured, the calculated diffusion coefficient is too small and the average number of hydrogen bonds per molecule is too high.In the present work we propose to simulate the liquid water at room temperature and experimental density using a very accurate ab-initio method where the electronic calculations are based on a quantum Monte Carlo (QMC) method. Among all methods, capable to describe the electron correlation in a very accurate manner, the QMC method is one of the most promising one, because can be extended to large number of atoms ( 100) due to a computational cost scaling rather well with the number of electrons and the extremely good performances in modern supercomputers. In the water system considered, it has been previously shown that a simple and very efficient QMC approach -namely the variational Monte Carlo based on an a very accurate and compact many-body wave function- allows us to calculate with high accuracy both the total energy and the ionic forces, and to use these forces to perform molecular dynamics simulations. Indeed our variational Monte Carlo method correctly describes the potential energy surface of a water molecule around its structural minimum, and both the dipole moment and the binding energy of water dimer, the latter being crucial for a proper description of the hydrogen bond and liquid water properties.

Resource awarded: 46 000 000 core hours on FERMI @ CINECA, Italy

DNS4RISC – Deterministic Numerical ground motion Simulations for RIsk hazard in Santiago de Chile

Project leader: Stupazzini, Marco; MunichRE, Germany

Collaborators: Paola Antonietti, Politecnico di Milano, Italy;Luca Formaggia, Politecnico di Milano, Italy;Ilario Mazzieri, Politecnico di Milano, Italy;Alfio Quarteroni, Politecnico di Milano, Italy;Roberto Guidotti, Politecnico di Milano, Italy;Roberto Paolucci, Politecnico di Milano, Italy;Chiara Smerzini, Politecnico di Milano, Italy

Abstract: Physics-based numerical modeling of the seismic response of arbitrarily complex earth media has gained major relevance in recent years, owing, on one side, to the ever-increasing progress in computational algorithms and resources, and, on the other side, to the growing interest towards the development of deterministic scenarios as input within seismic hazard and risk assessment studies.

The increasing need for certified numerical models apt to include the coupled effects of the seismic source, the propagation path through complex geological structures and localized superficial irregularities, such as alluvial basins or/and man-made infrastructure poses challenging demands on computer methods and resources due to the coexistence of very different spatial scales, from a few tens of kilometers, with reference to the seismic fault, up to a few meters, or even less, when considering some structural elements. Up to our knowledge, no one has ever tried to systematically include the following features in physics-based numerical scenarios: (i) non linear behavior, and ii) full 3D description of the central business district (CBD) for a large set of scenarios within the same numerical technique. The only attempt in this direction has been performed by Isbiliroglu et al. (2013) and Taborda et al. (2012) on a quite limited number of scenarios or coupling different numerical techniques (Krishnan et al. 2006) neglecting the Dynamic Soil-Structure Interaction (DSSI) or Site-City Interaction (SCI) problem.

The main objectives of the project can be summarized as follows: (a) generation of massive 3D deterministic scenarios to fully describe the seismic hazard in Santiago de Chile (including crustal, subduction inter- and intra-plate events); (b) inclusion, for a selected set of earthquake rupture scenarios, of the 3D description of middle rise and high rise buildings in the city of Santiago, in order to assess the importance of the Dynamic Soil-Structure and Site-City Interaction (DSSI and SCI, respectively) effects.

At the moment, physics-based numerical techniques are meaningful only for selected areas worldwide where the availability of well-constrained information on the risk under consideration justifies the higher computational costs. The multi-scale ground shaking maps, produced within the project with deterministic simulations, will constitute an improved database that will be used for risk assessment (hazard and vulnerability) in the Santiago de Chile region.

It is worth recalling here that the choice of Santiago de Chile was motivated mainly by two criteria: i) the high population level (more than one third of the total Chilean population live in the great Santiago region) together with the high concentration of value at ri
sk during a major seismic event; ii) the large amount of detailed information regarding the geophysical model of the Santiago basin and the building stock as well, as collected during reconnaissance surveys after the recent 27th of February 2010, Maule earthquake.

Resource awarded: 40 000 000 core hours on FERMI @ CINECA, Italy

Controlling Hydrogen Binding to Corrugated Graphene

Project leader: Tozzini, Valentina; Isituto Nanoscienze del Cnr, Italy

Collaborators: Riccardo Farchioni, Istituto Nanoscienze del CNR, Italy;Antonio Rossi, Istituto Nanoscienze del CNR, Italy;Riccardo Nifos, Istituto Nanoscienze del Cnr, Italy

Abstract: Graphene, a recently discovered new material, is a 2D honeycomb lattice of sp2 hybridized carbon, that can be obtained by separating the graphite layers. Its relatively low molecular weight and its uncommon mechanical properties (flexibility and resistance) make it a suitable material to be considered as hydrogen storage medium. In addition, the fact that H saturated graphene is insulator, and partially hydrogenated graphenes are tunable semiconductors, allows the possibility of building graphene-based nano-electronic devices based on controlled hydrogenation.Hydrogen can bind to graphene either by means of physisorption or chemisorption. The latter is more advantageous because it provides a stable binding, but problems arise due to the difficulty in overcoming the adsorption/desorption barriers.Density Functional Theory – Car-Parrinello dynamics studies performed by the proponents on a small model system of 200 C atoms, indicate that the hydrogen binding/release can be achieved by controlling the curvature of graphene. This is due to the strong dependence of reactivity and H binding stability on the local curvature at a given C site.Although this preliminary study shows the possibility of using this property in a device, the minimal model that can include realistic macroscopic effects has the size of at least 1000-1500 atoms. In fact, in practice, the curvature can be obtained by growing graphene on SiC substrates, whose minimal super-cell contains 338 C atoms plus a part of substrate; and the curvature control could be obtained manipulating a multi-layer system by means of external electric fields. In both cases curvatures are obtained and controllable on larger scales than those of the test system, that implies the simulation of supercells of the above mentioned size.The calculations on the smaller system were performed in part on the Fermi BG/Q system, since it has been put in production in sep 2012, with the CPMD code. This combination of system/software shows a good scalability up to the necessary size (2000-4000 cores). On the other hand, smaller computing systems than Tier0 do not have the capability of addressing a molecular system of the size of the larger one, which is the object of this proposal.The results of the study here proposed might have a large impact. We expect to verify that the preferential binding on convex areas and release by curvature inversion is observed also in the model system with realistic size. These results could be directly compared with the epitaxial graphene hydrogenation experiments (performed by the experimental group in the proponent’s research lab (NEST)), since the system simulated reproduced the supercell of epitaxial graphene on SiC. In addition we expect to show how the external fields could be used to control the curvature and produce release in a single or multilayer system.The results are likely to give indications for building a device based on graphene exploiting the corrugation control, with an impact in the field of energy storage and nanoelectronics.

Resource awarded: 38 000 000 core hours on FERMI @ CINECA, Italy

TEST SEA – Turbulence Events and Sediment Transport at SEAbed

Project leader: Vittori, Giovanna; University of Genoa, Italy

Collaborators: Markus Uhlmann, Karlsruhe Institute of Technology, Germany ;Marco Mazzuoli, University of Genoa, Italy

Abstract: A direct numerical simulation of the oscillatory boundary layer close to a flat rough wall is performed with the aim to detect the inception of turbulence and the formation and development of the turbulence structures. In particular the experiments by Carstensen et al (2012), who observed experimentally the formation of turbulent spots (TS) in an oscillatory boundary layer close to a rough wall, will be reproduced numerically and the characteristics of the TS will be quantified as well as their role in turbulence dynamics. The wall roughness will be obtained by placing spheres in a regular hexagonal array over a plane wall. In the first part of the investigation the spheres are not allowed to move. The numerical algorithm is based on a factional step method and makes use of the immersed boundary technique (Uhlmann, 2005) to force the boundary conditions. The second part of the project is aimed at extending the existing code to consider a layer of spherical particles moving in close contact to the rough wall. The numerical algorithm will be modified to allow particle movement. The obtained results will give new insight into the dynamics of sediment close to the sea bottom.

Resource awarded: 35 000 000 core hours on FERMI @ CINECA, Italy

LATTQCDNf3 – Chiral properties of Lattice QCD with three dynamical quark flavours

Project leader: Vladikas, Anastassios; INFN – Istituto Nazionale Fisica Nucleare, Italy

Collaborators: Martin Luescher, CERN, Switzerland;Piotr Korcyl, DESY, Germany;Stefano Lottini, DESY, Germany;Alberto Ramos, DESY, Germany;Hubert Simma, DESY, Germany;Rainer Sommer, DESY, Germany;Tomasz Korzec, Humboldt Universitaet zu Berlin, Germany;Gunnar Bali, Universitaet Regensburg, Germany;Vladimir Braun, Universitaet Regensburg, Germany;Sara Collins, Universitaet Regensburg, Germany;Bemjamin Glaessle, Universitaet Regensburg, Germany;Meinulf Goeckeler, Universitaet Regensburg, Germany;Paula Perez-Rubio, Universitaet Regensburg, Germany;Enno Scholz, Universitaet Regensburg, Germany;Wolfgang Soeldner, Universitaet Regensburg, Germany;Andre Sternbeck, Universitaet Regensburg, Germany;Gregorio Herdoiza, University of Mainz, Germany;Harvey Byron Meyer, University of Mainz, Germany;Georg von Hippel, University of Mainz, Germany;Hartmut Wittig, University of Mainz, Germany;Jochen Heitger, Westfaelische Wilhelms-Universitaet Muenster, Germany;Germany Della Morte, IFIC (CSIC), Spain;Carlos Pena Ruano, Universidad Autonoma de Madrid, Spain;Georg Engel, University of Milan Bicocca, Italy;Leonardo Giusti, University of Milan Bicocca, Italy;Federico Rapuano, University of Milan Bicocca, Italy;Mauro Papinutto, University of Rome La Sapienza, Italy

Abstract: The strong interaction of fundamental particles is successfully described by Quantum ChromoDynamics (QCD). At high energies, the theory displays the outstanding property of weakly coupled quarks and gluons, known as asymptotic freedom. This justifies the use of perturbation theory in
these energies and the resulting theoretical predictions are known to agree well with experiment. On the contrary, the most natural framework for low energy studies of hadronic interactions consists of QCD with discretized space-time in a large finite volume. Extensive computer simulations of this theory, known as lattice QCD, are performed for several quark masses and lattice spacings, and the physical high precision results are extrapolated to the continuum and chiral limits.

The present proposal aims at a high-precision determination of the free parameters in the quark sector of the Standard Model, a detailed quantitative understanding on how chiral symmetry is realized in QCD, and a precise determination of some of the low-energy constants of the QCD chiral dynamics. More specifically we will compute the QCD chiral condensate, light and strange quark masses and meson decay constants and the axial charge of the nucleon. These results will provide stringent tests of the low energy, strongly interacting sector of the Standard Model. An important element of the project consists in the novel treatment of an important systematic effect, namely the entrapment of simulations in topological charge sectors of field space. This important effect is habitually overlooked in lattice QCD simulations close to the continuum limit. We expect that, with most systematic effects carefully under control, the impact of our work will be that of providing benchmark estimates of important Physical quantities which crucially depend on the correct approach to the chiral limit.

We will perform high precision lattice QCD simulations for a lattice spacing of 0.09 fm and volumes of about 3.0-6.0 fm, aimed at a thorough investigation of the chiral limit. The three lighter quark flavours (up, down, strange) are dynamical degrees of freedom. The simulated up and down quarks are such that pions are as light as 140 MeV. Reliable computation of physical observables at such low masses is achieved by using state of the art algorithms, suitably optimized for the computer architectures of PRACE, with carefully selected boundary conditions of the finite lattices.

Resource awarded: 70 000 000 core hours on FERMI @ CINECA, Italy