15th Project Access Call – Awarded Projects

Results of the 15th Call for Proposals for Project Access.

Projects from the following research areas:

Biochemistry, Bioinformatics and Life sciences (4)

CryptoPocketSim. Understanding the mechanism of cryptic pocket formation at protein-protein interfaces: the case of TNFa

Project Title: CryptoPocketSim. Understanding the mechanism of cryptic pocket formation at protein-protein interfaces: the case of TNFa
Project Leader: Francesco Gervasio  
Resource Awarded: 16.8 million core hours on MareNostrum

Details

Team Members:
Federico Comitani, University College London, UNITED KINGDOM

Abstract:
The aim of this proposal is to understand the mechanism of formation of “cryptic” pockets and allosteric regulation at protein-protein interfaces in an actively investigated drug target (TNF) by means of enhanced-sampling simulations. “Cryptic” pockets, that is, sites on protein drug targets that only become apparent when a drug binds, offer an attractive opportunity for the development of allosteric drugs for difficult targets where most classic drug-design strategies fail. However, due to their “hidden” nature, they have been in most cases discovered by chance. What is more, the general molecular mechanism by which cryptic sites are formed is still not clear. My group has recently used enhanced sampling simulations to discover a previously unknown cryptic pocket in an anticancer target (FGFR)1. Building on this successful experience, we developed a Hamiltonian Replica Exchange-based approach (SWISH) and combined it with small molecular probes to investigate the nature of cryptic sites in four pharmacological targets, such as the beta-lactamase TEM1 (a target for antimicrobial resistance). Our results published in the JACS2 attracted a lot of interest from both academia and industry. Building on these successes here we want to study a more complex and pharmacologically-relevant system: the trimeric tumour necrosis factor (TNF). Rational drug discovery has so far failed to design viable inhibitors, however tool molecules that disrupt the formation of the trimer are known. Based on preliminary computational and experimental results, we will apply a combination of enhanced sampling methods (SWISH, Metadynamics and probes) to explore the mechanism of complex assembly as well as cryptic pocket formation and ligand binding at protein-protein interfaces of the TNF. The project will benefit from successful collaboration with a committed industrial partner that will provide substantial experimental data to validate the computational results.

 Understanding TET2 regulation and development of chemical probes that control its function

Project Title: Understanding TET2 regulation and development of chemical probes that control its function
Project Leader: Xavier Barril
Resource Awarded: 68.3 million core hours on Piz Daint

Details

Team Members:
Maciej Majewski, Universitat de Barcelona, SPAINSerena Piticchio, Universitat de Barcelona, SPAINSergio Ruiz Carmona, Universitat de Barcelona, SPAIN

Abstract:
Epigenetic mechanisms introduce a fundamental layer of regulation in the cell. The proteins that execute the basic epigenetic functions (writing, reading, erasing) are increasingly im-portant therapeutic targets, particularly in cancer. Ten-eleven translocation (TET) proteins are enzymes that demethylate DNA through an oxidation mechanism. TET loss of function, lead-ing to DNA hypermethylation is an early event in many forms of cancer, but TET regulation is also involved in disparate events, such as embryonic development or brain function. The discovery of chemical probes capable of regulating their function will be an important step towards the understanding of TET proteins involvement in health and disease, which is fundamentally lacking. In this project we aim to understand TET regulation at the molecular level and to develop novel small-molecule modulators of this protein class. These tool compounds will be used to dissect the biological function of this class of epigenetic enzymes and, perhaps, open new therapeutic venues.

 Genome and phenome-wide association analyses in 1Million individuals through imputation with sequenced-based reference panels

Project Title: Genome and phenome-wide association analyses in 1Million individuals through imputation with sequenced-based reference panels
Project Leader: David Torrents
Resource Awarded: 16.5 million core hours on MareNostrum

Details

Team Members:
Lorena Alonso, Barcelona Supercomputing Center, SPAINMarta Guindo, Barcelona Supercomputing Center, SPAINMercè Planas, Barcelona Supercomputing Center, SPAINMontserrat Puiggros, Barcelona Supercomputing Center, SPAINRomina Royo, Barcelona Supercomputing Center, SPAINJordi Valls, Barcelona Supercomputing Center, SPAINJorge Ejarque, Barcelona Supercomputing Center, SPAINJordà Polo, Barcelona Supercomputing Center, SPAINFriman Sánchez, Barcelona Supercomputing Center, SPAINJosep M Mercader, Broad Institute of MIT and Harvard, UNITED STATES

Abstract:
Complex diseases are a major healthcare problem worldwide and are caused by multiple factors, often combining a genetic predisposition with certain environmental factors and life styles. The genetic basis of these diseases are also complex, and often involve multiple loci, as it has been already described for several common disease, such as Type 2 Diabetes (T2D). The identification of these genetic factors is therefore key to understand the molecular basis of complex diseases, and to elaborate new treatment, diagnosis and prevention protocols. The current and standard approach to unravel the genetics behind complex diseases is called “genome-wide association studies” (GWAS), which relies on the comparison between thousands of case and control genotypes, to finally identify markers significantly associated with the disease. However, current GWAS workflows are complex, time-consuming and computational demanding, and, to date, no complete integrated framework has been developed for genome-wide genotype imputation and association testing for existing and upcoming large and heterogeneous GWAS datasets (comprising > 50,000 samples). For that reason, we developed GUIDANCE, an integrated framework that efficiently performs large-scale GWAS, including genotype imputation using multiple reference panels and cross-phenotype analysis, in parallel computing infrastructures and in a single execution. Using 10th and 13th PRACE resources, we have successfully applied this methodology to several genetic studies (Horikoshi, M. Nature, 2016, Galván-Femenía I., 2017, submitted manuscript), including a large re-analysis of 70,000K publicly available samples for a T2D study, where we identify seven novel T2D associated loci, two of them lead by a low-frequency and a rare variant (and Bonàs-Guarch, S., 2017, submitted manuscript). In this T2D study, we demonstrate that the reanalysis of publicly available data, using a combination of modern sequenced-based reference panels, allows the identification of new disease genetic markers in the form of risk variants. Here we propose to extend these approaches for the study of up to 44 complex diseases from the database of Genotypes and Phenotypes and the European Genotype Archive, and 1,717 phenotypes from UK Biobank for around 1M of individuals, not only to identify new risk variants, but also to understand their interaction within and across different phenotypes, defining their corresponding risk genetic network. The number of samples involved and the reference panels selected makes this study computationally demanding. Based on our previous studies, we envision the identification of a large number of new loci and novel causal variants, to study comorbidities and variant interaction, to ultimately start defining the genetic risk maps behind complex diseases and to define the corresponding prevention protocols.

GREaT – Genome wide identification of RNA editing sites in very large cohorts of human whole transcriptome data

Project Title: GREaT – Genome wide identification of RNA editing sites in very large cohorts of human whole transcriptome data
Project Leader: Ernesto Picardi
Resource Awarded: 30 million core hours on Marconi – KNL

Details

Team Members:
Tiziana Castrignanò, CINECA, ITALYTiziano Flati, CINECA, ITALYSilvia Gioiosa, CINECA, ITALYBruno Fosso, National Research Council, ITALY

Abstract:
RNA editing is a widespread post-transcriptional mechanism that alters primary RNA sequences through the insertion/deletion or modification of specific nucleotides. In humans, RNA editing affects nuclear and cytoplasmic transcripts mainly by the deamination of adenosine (A) to inosine (I) through members of ADAR enzymes. A-to-I modifications increase transcriptome and proteome diversity, and contribute in modulating gene expression at RNA level. RNA editing by A-to-I change is prominent in non-coding regions containing Alu repetitive elements, whereas the list of ADAR substrates in protein coding genes is relatively small. RNA editing modifies several human neurotransmitter receptors and plays important roles in modulating their physiology. Indeed, its deregulation has been linked to a variety of human diseases including neurological and neurodegenerative disorders, and cancer. Current technologies for massive transcriptome sequencing such as RNASeq are providing accurate maps of transcriptional dynamics occurring in complex eukaryotic genomes as in human and are facilitating the detection of post transcriptional RNA editing modifications with unprecedented resolution. The computational detection of RNA editing events in RNAseq experiments is quite intensive requiring the browsing of the human genome position by position. Therefore, the study of RNA editing in very large cohort of RNAseq data is precluded. Here, we propose to analyse RNA editing in thousand RNAseq data (>30000) from public projects such as GTEx and TCGA through the parallel version of our specialized software named REDItools. The aim is to unveil unknown biological aspect of RNA editing in health and disease, contributing to elucidate the real role of RNA editing in human and providing novel sources of biomarkers or new targets for innovative drugs. Results will provide the first comprehensive atlas of RNA editing in human and will be freely disseminated to scientific community.

top

Chemical Sciences and Materials (18)

Electronic and optical properties of high-performance monolayer and multilayer materials

Project Title: Electronic and optical properties of high-performance monolayer and multilayer materials
Project Leader: Nicola Marzari
Resource Awarded: 30 million core hours on Marconi – KNL

Details

Team Members:
Davide Campi, EPFL, SWITZERLANDMarco Gibertini, EPFL, SWITZERLANDAntimo Marrazzo, EPFL, SWITZERLANDNicolas Mounet, EPFL, SWITZERLANDThibault Sohier, EPFL, SWITZERLANDDeborah Prezzi, CNR-NANO, ITALYPaolo Umari, University of Padova, ITALY

Abstract:
There is major effort worldwide in exploring the capabilities of graphene and related materials for advanced technological applications; this is even more relevant for Europe, thanks to the 10-year effort spearheaded by the FET Graphene Flagship. While the effort has now focused mostly on a handful of materials – most notably graphene itself and transition-metal dichalcogenides – we believe that many more opportunities are available, and that computational screening and discovery can greatly accelerate the process of identifying the most promising materials for innovative applications. In fact, using high-throughput computational screening on more than 110,000 experimentally-known inorganic materials, we have recently identified close to 2000 layered candidates that can be exfoliated into mono- and multi-layers. These cover all the materials that give rise to the known 2D monolayers (from graphene to boron nitride, transition-metal chalcogenides, black phosphorus…), but obviously many more. Here, we want to urgently exploit and expand this portfolio, to identify as quickly as possible the systems with the most promising electronic and optical properties. This effort will involve a combination of rapid screening with approximate methods followed by a high-accuracy study of the most promising candidates. Deployment and dissemination will greatly benefit from our extensive expertise in high-throughput methods, powered by the AiiDA materials informatics platform (http://aiida.net), the use of state-of-the-art high-performance open-source codes for electronic-structure simulations, supported by the H2020 MaX Centre of Excellence (http://max-centre.eu), and our commitment towards full dissemination of the data and calculations’ workflows through the Materials Cloud (http://materialscloud.org).

 CATNIP – ChArge TraNsport In Perovskite solar cells

Project Title: CATNIP – ChArge TraNsport In Perovskite solar cells
Project Leader: Feliciano Giustino
Resource Awarded: 20 million core hours on MareNostrum

Details

Team Members:
Marina Filip, University of Oxford, UNITED KINGDOMSamuel Ponce, University of Oxford, UNITED KINGDOMMartin Schlipf, University of Oxford, UNITED KINGDOMGeorge Volonakis, University of Oxford, UNITED KINGDOMMarios Zacharias, University of Oxford, UNITED KINGDOMNourdine Zibouche, University of Oxford, UNITED KINGDOM

Abstract:
The energy production of society today largely depends on the conflict-laden extraction of fossil resources contributing to the accelerating climate change. To evolve the current unsustainable model into one where energy is produced renewable, converting sunlight into electric power via solar cells is a prominent solution. Aiming to obtain a affordable mass product, we require the development of efficient, reliable, and cheap materials. Recently, a novel material class – the so-called halide perovskites – was discovered. In the few years since their discovery, researchers managed to achieve efficiencies surpassing conventional solar cells. Our major research interest is to understand why these materials exhibit such an astonishing performance and what parameters one can tweak to develop even better solar cell materials. Fundamentally, a solar cell absorbs sunlight and converts it into electrical charges. These charges have to leave the solar cell so that they can used as electrical power. In CATNIP, we want to understand how the interaction between the electrical charges and the solar cell material limits the extracted electric power in particular in these novel halide perovskites. Building on the Nobel-prize winning work of Walter Kohn, we developed numerical algorithms allowing us to evaluate the electric transport inside the solar cell solving the fundamental physical equations. In this way, we can assess material properties in the computer without the need for empirical input. Within CATNIP, we will examine a set of materials to understand how changes to the chemical composition affects the electric properties. CATNIP is composed of three work-packages: We will examine the crystal structure of the materials (work-package 1), the behavior of the charge carriers (work-package 2), and the interaction of crystal and charge carriers determining the fundamental limit of charge transport (work-package 3). Based on the results obtained in CATNIP, we will aim to predict novel photovoltaic materials that our experimental collaborators in Oxford can synthesize. With their immense experience in growing perovskite solar cells, we are well-equipped for a fast track from computational design to experimental realisation. Our group’s philosophy is to provide optimal access to the results of our projects. Thence we will not only disseminate our results in high-profile scientific journals, but also provide open access to the input used to generate the data on our GitHub account (https://github.com/mmdg-oxford/papers) simultaneously to the publication.

BioTitan – Ab initio molecular dynamics of biomolecular adsorption on fully hydrated TiO2-water interfaces

Project Title: BioTitan – Ab initio molecular dynamics of biomolecular adsorption on fully hydrated TiO2-water interfaces
Project Leader: Alexander Lyubartsev
Resource Awarded:  50.2 million core hours on MareNostrum

Details

Team Members:
Lorenzo Agosta, Stockholm University, SWEDENErik Brandt, Stockholm University, SWEDEN

Abstract:
The nano-bio interface is the intersection between biology and solid-surface nanotechnology at the atomic scale. A wide range of nanotechnological applications rely on exploiting the circumstances where molecules from biology attach to, or detach from, solid surfaces. For example, diseases can be treated efficiently with targeted delivery of drug molecules that are carried by engineered nanoparticles. Biosensors selectively identify molecules with desired properties based on solid surface-recognition of adsorbing biomolecules. Medical implants, replacing malfunctioning body parts, rely on cell adhesion and interactions that are mediated by small biomolecules attached to the inorganic materials. However, concerns have been raised on the potential hazard of engineered nanomaterials on health and environment. Nanomaterials that enter living organisms become coated with biological molecules such as proteins, lipids, and sugars. The response from the immune system to the coated nanomaterial might lead to adverse outcomes in terms of nanotoxicity. Biological systems operate in water, so the nano-bio interface is fully hydrated at ambient condition. Adsorption at solid-liquid interfaces is more complex compared to their well-studied solid-gas counterparts, because when probed in experiments the signals from surface atoms is much weaker and hidden by that from the bulk. TiO2 is a highly biocompatible material, commonly used in the form of engineered nanoparticles. Since TiO2 is considered to be biologically inert, it is a commonly used material for medical implants, but also widely used in consumer products such as sun screens and tooth pastes. However, exposed atoms on the TiO2 surfaces can react with water, create harmful oxygen radicals, and significantly reshape the atomic surface structure, losing biocompability. At present, this atomic surface structure is unknown, the fact that TiO2 becomes increasingly reactive as the nanoparticle size decreases is well established. Here, we use accurate models base on quantum mechanics to simulate how protein fragments (amino acids) attach to TiO2 surfaces immersed in water. Large-scale ab initio molecular dynamics is used to simulate amino acid adsorption on fully hydrated TiO2 surfaces. Our calculations can be used to rank the binding strengths of each biofragment to TiO2, and identify the structure of each amino acid when bound to the surface. By comparing our computations to experimental measurements we will construct models with atomistic resolution for how larger biological molecules attach to TiO2 surfaces. A systematic methodology to study bio-nano interfaces and binding processes of large macromolecules in real time computer experiments will be achieved from our results. This approach will be pioneering the use of computers as practical tools in order to determine how living matter interacts with solid materials on the atomic level. Our calculations will lead to safe design principles for the manufacture of nanobiotechnology devices in the future by identifying possible health hazards of nanomaterials. Further, the calculations can be used in combination with experiments to reveal atomistic information for optimizing nano-devices, thus contributing to maintaining our current high standard of living.

 EXCIPHOCAT – Advancing our understanding of photocatalytic water splitting: effects of nanostructuring and hydroxylation on excited states in TiO2

Project Title: EXCIPHOCAT – Advancing our understanding of photocatalytic water splitting: effects of nanostructuring and hydroxylation on excited states in TiO2
Project Leader: Francesc Illas Riera
Resource Awarded:  40 million core hours on MareNostrum

Details

Team Members:
Stefan Thomas Bromley, Universitat de Barcelona, SPAINToni Macià, Universitat de Barcelona, SPAINAngel Morales, Universitat de Barcelona, SPAINRosendo Valero, Universitat de Barcelona, SPAINFrancesc Viñes, Universitat de Barcelona, SPAINVolker Blum, Duke University, UNITED STATES

Abstract:
Titanium dioxide (TiO2) is a fascinating material exhibiting unique photocatalytic properties which are exploited in many technological applications. Of particular importance is the capacity of TiO2 to photocatalyze the water splitting, resulting in H2 and O2. Hydrogen produced in this manner would constitutes a valuable, and inexhaustible, source of renewable energy. The rather large band gap of anatase and rutile TiO2 polymorphs, however, precludes its use for such applications under visible sunlight and suitable material modifications are being sought to improve its performance (e.g. nanostructuring, doping). Clearly, to design an effective modification requires a detailed knowledge of the electronic structure of TiO2 at the nanoscale, specifically with respect to its excited electronic states. In this project, we will make significant advances with respect to research work carried out in the framework of the previous COMPHOTOCAT PRACE projects 2014112608 and 2015133081. In this project first principles modelling of realistic anatase nanoparticles of up to 6 nm and containing more than 1000 atoms, was successfully achieved by means of all-electron density functional theory (DFT) based calculations. In the present project, we will use DFT based calculations to study the structures and thermodynamics of realistically sized anatase, rutile and mixed anatase-rutile TiO2 nanoparticles, both in vacuum and with hydroxylated surfaces, and, using advanced many body GW techniques, we will also investigate their excited states. Particular attention will be devoted to the anatase-rutile interface to clarify its role in electron-hole generation.

NANOCAT-ZN Modelling mechanism of reactions and active sites in Ziegler-Natta nano-systems, looking for reactivity indicators in heterogeneous catalysis

Project Title: NANOCAT-ZN Modelling mechanism of reactions and active sites in Ziegler-Natta nano-systems, looking for reactivity indicators in heterogeneous catalysis
Project Leader: Maddalena Damore 
Resource Awarded:  30 million core hours on SuperMUC

Details

Team Members:
Silvia Casassa, University of Torino, ITALYBartolomeo Civalleri, University of Torino, ITALYElena Clara Groppo, University of Torino, ITALYLorenzo Maschio, University of Torino, ITALY

Abstract:
Heterogeneous catalysts have often a complex nature that prevents the acquisition of the atomic-level knowledge on the active sites, especially in reaction conditions. The traditional experimental methods of use in surface science hardly provide an average picture of nano-catalysts, which are still optimized empirically in the industrial practice. On the other hand, HPC resources make now possible to run efficiently parallel codes with excellent scalability in terms of both speed-up and memory requirements. Due to the recent development in Density Functional Theory and to the availability of powerful computational facilities, theory is moving towards the computational design of efficient heterogeneous catalysts. In this project we address Ziegler-Natta (ZN) catalysts for olefin polymerization as prototypes of complex nano-sized catalysts. They are computationally highly demanding systems, with a long history of experimental and theoretical collection of data. Any description of ZN catalysts should take into account: 1) the presence of multiple sites on different MgCl2 surfaces and on defective positions; 2) the fact that the adducts of catalytic interest – involving TiCl4, aluminum alkyls and very bulky aromatic donors – have very low degree of coverage; and 3) the fact that after reduction of Ti species unrestricted DFT calculations are necessary. For these reasons, an accurate ab-initio prediction of thermodynamics and kinetics of our systems will require highly expensive calculations and huge computational resources. CRYSTAL in its Massive Parallel version (MPPCRYSTAL) will be the ab-initio code of election. Thanks to thousands cores, it will be possible to simulate nano-crystals of up to thousand atoms at high UB3LYP-D/TZVP level of computation. In addition, we will perform an analysis of plausible reactivity indicators, such as electron density transfers, in order to evaluate the occurrence of specific reaction mechanisms, on the basis of ELF and MPDs theories. The project stands on a synergetic cooperation between theoreticians, spectroscopists and industrial partners (DPI Project PO2.0-2016-005). In the short period, the results are expected to contribute to the rationalization of the relation between the structure of the active sites, and their reactivity and selectivity towards olefins. On a long-term scale our results might lead to improve both productivity and stereo-specificity of these catalysts, with a potential positive outcome from an industrial perspective, as well as to improve the performances of the code when running on thousands of processors.

 SuperquantumMC – Nuclear quantum effects in hydrogen-based high-temperature superconductors by quantum Monte Carlo methods

Project Title: SuperquantumMC – Nuclear quantum effects in hydrogen-based high-temperature superconductors by quantum Monte Carlo methods
Project Leader: Michele Casula
Resource Awarded: 40 million core hours on Marconi – KNL

Details

Team Members:
Tommaso Gorni, Université Pierre et Marie Curie, FRANCEClaudio Genovese, SISSA, ITALYSandro Sorella, SISSA, ITALY

Abstract:
The physics of hydrogen and hydrogen-dominant materials depends on a subtle interplay between nuclear quantum effects and electron correlation. This results in a very rich landscape of competing phases, which can be tuned by external thermodynamic parameters, such as pressure or temperature, leading to the emergence of new states of matter. A notable example is superconductivity in sulfur hydrides at Mbar pressures with record critical temperatures. Its quantitative description poses a formidable challenge to theory, as the quantum nature of hydrogen must be taken into account together with a high-accuracy resolution of the electronic structure. In the H3S sulfur hydride, it is believed that the highest ever measured superconducting critical temperature belongs to a phase where the hydrogen positions are symmetrized by quantum effects. However, the internal energies of the competing nearby structures are very close each other, and their predicted stability and phase boundaries strongly depend on the exchange-correlation functional used in density functional theory calculations. In this project, we are going to study the phase diagram of the H3S by very accurate quantum Monte Carlo simulations from first principles. Internal energies will be estimated at frozen ionic configurations, while nuclear quantum and thermal effects will be included by a recently developed fully quantum molecular dynamics. Phonon frequencies will be computed as byproduct. The outcome of this project is fundamental to validate the hypothesis of hydrogen symmetrization as driving mechanism to stabilize the highest-Tc superconducting phase, and more generally to shed light on the complex phase diagram of sulfur hydrides. This will be achieved thanks to the development and application of a quantum Monte Carlo framework that describes reliably the interplay between quantum and thermal effects, and electron correlations in this class of materials, where hydrogen is key to reach groundbreaking properties.

High-throughput simulations of transistors based on 2-D materials

Project Title: High-throughput simulations of transistors based on 2-D materials
Project Leader: Mathieu Luisier
Resource Awarded: 71 million core hours on Piz Daint

Details

Team Members:
Cedric Klinkert, ETH Zurich, SWITZERLANDChristian Stieger, ETH Zurich, SWITZERLANDAron Szabo, ETH Zurich, SWITZERLAND

Abstract:
Following the first experimental demonstration of graphene in 2004 2-D materials have rapidly received a wide attention from the scientific community and found their way in several applications. While the number of exfoliated 2-D compounds keeps increasing, most of them still remain to be discovered: there exist 5,619 3-D layered materials among which 1,844 could be possibly exfoliated. Some of them could play a major role in the future of nanoelectronics, whose driving force, Moore’s scaling law, is predicted to come to an end in the near future. The goal of this project is therefore to screen the available 2-D material design space and identify components that could build the channel of next-generation ultra-scaled transistors beyond the currently manufactured Si FinFETs and revive Moore’s law. To optimize the search process a high-throughput (HT) simulation environment will be used. It relies on the density functional theory code Quantum ESPRESSO (QE) to isolate stable 2-D crystals and compute their electronic structure, on the Wannier90 package to transform the plane-wave outputs of QE into a set of localized basis functions, and finally on the OMEN quantum transport solver to calculate the “current vs. voltage” characteristics of the investigated transistors and extract their most significant metrics.

 DDFSDAM: Drug design for HPC systems: Fast Switching Double Annihilation Method (FS-DAM) at work

Project Title: DDFSDAM: Drug design for HPC systems: Fast Switching Double Annihilation Method (FS-DAM) at work
Project Leader: Piero Procacci
Resource Awarded: 17 million core hours on Marconi – Broadwell

Details

Team Members:
Marco Pagliai, University of Florence, ITALYGiorgio Signorini, University of Florence, ITALY

Abstract:
The determination of the binding affinity in ligand-receptor systems is placed right at the start of the drug discovery process, in a sequence of increasing capital-intensive steps, from safety tests, lead optimization, preclinical and clinical trials. The availability of a reliable computer-based tool for drug discovery, replacing or integrating the traditional end-states docking approach, is hence strategical for alleviating the cost of false positive, identified in the early stages of the multi-years drug-discovery pipeline. In silico screening using advanced simulation techniques, such as Free Energy Perturbation or Thermodynamic Integration alchemical methodologies, require expensive simulations that are typically plagued by convergence problems and reproducibility issues. Not surprisingly, commercial applications of equilibrium-based free-energy simulations have been limited due to the lack of large-scale validation coupled with the technical challenges associated with running these types of calculations. Exploiting some recent advances in non-equilibrium statistical thermodynamics, we have developed a new variant of the free energy alchemical method, that is specifically tailored for HPC system. The methodology, called Fast Switching Double Annihilation Method (FS-DAM) relies on the accurate Boltzmann sampling of the fully coupled bound and unbound states via Hamiltonian Replica exchange simulations with torsional tempering, followed by the production of many fast and independent non-equilibrium Molecular Dynamics trajectories with a continuous dynamical evolution of an externally driven alchemical coordinate, completing the decoupling of the ligand in a matter of few tens of picoseconds rather than nanoseconds. The drug-receptor binding free energies are recovered by using an unbiased unidirectional estimate derived from the Crooks’ theorem[GE Crooks. J. Stat. Phys., 90, 1481-1487, 1998] exploiting the inherent Gaussian nature of the ligand decoupling work. In order to assess its feasibility in an industrial context, FS-DAM will be tested on a HPC platform across a broad range of target classes and ligands, with retrospective results encompassing some hundreds of ligand-receptor systems with disparate binding affinities, including BACE, CDK2, JNK1, MCL1, p38, PTP1B, thrombin, and Tyk2 protein targets.

PEPBOND – PEPtide BOND formation at the air/water interface

Project Title: PEPBOND – PEPtide BOND formation at the air/water interface
Project Leader: MariePierre GAIGEOT
Resource Awarded: 29.3 million core hours on Curie

Details

Team Members:
Daria Galimberti, Universite d’Evry val d’Essonne, FRANCESimone Pezzotti, Universite d’Evry val d’Essonne, FRANCEFlavio Siro Brigiano, Universite d’Evry val d’Essonne, FRANCE

Abstract:
Determining the suitable conditions for the peptide condensation reaction is pivotal in rationalizing the origin of life on the planet Earth. The peptide bond formation is known to be disfavoured in bulk liquid water at ambient temperature and pressure, for both kinetic and thermodynamic reasons. The group of Vaida (USA) has opened a new route for the peptide bond formation with their 2012 experiment where the formation of polypeptides has been demonstrated through a condensation reaction between amino acid esters, catalysed by Cu2+, at the air/water interface. This work thus opened up a new promising playground for the peptide bond formation on prebiotic planet Earth: the air/water interface. Air/water interfaces have often been proposed to play key roles in altering the ionization state of species and in aligning (bio)molecules, providing a unique chemical environment. The 2012 Vaida’s experiment hence highlighted the importance of the air/water interface in the formation of peptide bonds, but the fundamental reasons for this chemical reaction to occur at this interface (and not in bulk water) has still to be unraveled. We propose to apply First Principles Molecular Dynamics (FP-MD) simulations in order to provide a detailed comprehension of why the air/water interface is a unique environment for the peptide bond formation. Our theoretical investigations will focus on structural properties at the air/water interface containing peptides and the Cu2+ catalyst, and on the relationships between structural properties and the chemical reaction mechanisms and energetics for the peptide bond formation. Our simulations will treat both the reactions at the air/water interface and in the bulk liquid water, in order to fully rationalize what the air/water environment allows and what the liquid water environment does not. Non-constrained and constrained FP-MD simulations will be performed in order to unravel all these inter-related properties. FP-MD simulations are the mandatory theoretical tools for our project, constrained FP-MD being essential for assessing the underlying energetics of the chemical reactions at play. The proposed FP-MD simulations are challenging and require Tier-0 computational resources that only the PRACE consortium can offer. With the PRACE allocation we will be the first group to offer a fundamental understanding of the peptide bond formation and to directly simulate the chemical reaction at the special environment given by the interface between air and water.

 Exploring new frontiers in Rayleigh-Bénard convection

Project Title: Exploring new frontiers in Rayleigh-Bénard convection
Project Leader: F. Xavier Trias
Resource Awarded: 33.1 million core hours on MareNostrum

Details

Team Members:
Firas Dabbagh, Technical University of Catalonia, SPAINCarlos David Pérez-Segarra, Technical University of Catalonia, SPAINAndrey Gorobets, Russian Academy of Sciences, RUSSIAN FEDERATION

Abstract:
The air flow surrounding us, the blood flow in the vessels of our bodies, the water flow in oceans and many other fluid flow are essentially classified under the discipline of fluid dynamics. Practically, fluid dynamics drives the energy transport process in many open/confined industrial applications. A famous example thereof is the buoyancy driven flow heated from below and cooled from above. It is classified under the name of Rayleigh-Bénard convection (RBC) and resembles a wide variety of circulations in nature, e.g. atmospheric, oceanic and mantle convection, and in industry, e.g. flows in nuclear reactors and solar thermal power plants. Most of these flows are at hard turbulent regime with a highly sensitive behavior to the physical circumstances. When we talk about turbulence, we first pass in our mind the disturbed streaming of smoke cigarette, the high rotational mixing in our coffee and the coherent vortices from an airplane wings. It is composed of a swirling fluid moving randomly around and about the overall direction of motion, being broken down into smaller and smaller eddies and transporting its kinetic energy to the level of viscous dissipation. Therefore, the key feature of turbulence physics is around disclosing the small scale motions. Studying the dynamics of the fine scales represents fundamental perspectives of the flow topology evolution, construction of the turbulent wind and the mechanism of kinetic energy cascade, to eventually broaden our knowledge of thermal turbulence and improve the design of its applications. With the rapid development of the computational technology, direct numerical simulation (DNS) forms, when possible, the most accurate and reliable tool to study the turbulence physics in RBC at hard turbulent regime. In this project, we propose to investigate the behavior of the fine-scale dynamics as the nonlinear production terms of enstrophy and dissipation, the flow topology change and the turbulent background growing events with increasing the Rayleigh number (Ra) in hard turbulent RBC. To do so, we propose to perform a DNS of an air-filled turbulent RBC at Ra=1e11, to be the first DNS performed in a rectangular configuration. On the other hand, other important research questions will be addressed in this project too. Namely, the performance of the novel S3PQR eddy-viscosity model for large-eddy simulation (LES) will be tested for the above described RB configuration. This model has been recently proposed by the applicants. Firstly, instantaneous results of the DNSs will be used to investigate a priori features in turbulence LES models. In this regard, we also expect to study the influence of the subgrid characteristic length in the performance of LES subgrid-scale (SGS) models. This is an on-going research conducted by the applicants. Altogether, we aim to extend these studies at higher Ra-numbers with the ultimate goal to improve the performance of current LES models in RBC. Altogether, these potential improvements in the modelization of SGS should be a key element to explore new frontiers (e.g. the ultimate regime) in Rayleigh-Bénard where DNSs are not possible yet.

MDMemRe – Combining Molecular Dynamics Simulations and Experiments to Understand Functional Consequences of Membrane Remodelling

Project Title: MDMemRe – Combining Molecular Dynamics Simulations and Experiments to Understand Functional Consequences of Membrane Remodelling
Project Leader: Mark Sansom
Resource Awarded:  16 million core hours on Curie

Details

Team Members:
Sarah-Beth Amos, University of Oxford, UNITED KINGDOMAnna Duncan, University of Oxford, UNITED KINGDOMElizabeth Jefferys, University of Oxford, UNITED KINGDOMWanling Song, University of Oxford, UNITED KINGDOM

Abstract:
Membrane remodelling helps to establish the complex architectures of the cells and to regulate many biological processes such as protein sorting, membrane protein tethering and membrane fusion, etc. However, the molecular mechanisms of such regulation are not clear. In this proposal, we will use coarse-grained molecular dynamics simulations to study three representative topics in membrane remodelling. Gaining access to PRACE machines will allow us to create large simulation systems which are directly comparable to experiments, thus enabling us to quantify the influence of membrane curvature via a series of membrane systems of differing undulation frequencies. The use of the simulation engine GROMACS ensures good scaling performance on all PRACE machines, especially on CURIE. Our computational results will be tested via different experimental methods with our collaborators in the UK and internationally. The results will advance our understanding of the relationship between biological function and mesoscale organization of cellular membranes, and will be of great interest to pharma companies.

 H2O-MaGic – Predicting the behavior of water in MOFs: a machine-learning approach to screening the Materials genome
Project Leader: Berend Smit

Project Title: H2O-MaGic – Predicting the behavior of water in MOFs: a machine-learning approach to screening the Materials genome
Project Leader: Berend Smit
Resource Awarded: 15 million core hours on SuperMUC

Details

Team Members:
Peter Boyd, Ecole Polytechnique Federale de Lausanne (EPFL), SWITZERLANDBerend Smit, Ecole Polytechnique Federale de Lausanne (EPFL), SWITZERLANDAmber Mace, Stockholm University, SWEDEN

Abstract:
Metal-organic frameworks (MOFs) have in the last decade received much attention for green energy related applications, for example, capturing carbon dioxide gas from coal-fired power plant exhaust. However, the behavior of water in a MOF is often conclusive for any given application it has been designed for. Consequently, understanding, characterizing and predicting how water is adsorbed on the diverse chemical and topological pore spaces typifying MOFs is the key challenge. This knowledge will be essential when identifying and/or designing a MOF for a certain application where humidity is a critical factor. We take on this challenge by proposing to develop a Materials Genomic approach to address the key question at study: Which chemical and topological properties in MOF pores diminish water uptake and which enhance it? How can we combine the metals and linkers to obstruct a material’s affinity to water or steer it to a certain level? However, with the nearly limitless potential for unique MOF materials, and the millions of hypothetical materials already at our disposal, an enormous amount of CPU time would be necessary to perform a brute force computational screening of simulated water in these structures. A new strategy must therefore be employed. Here, we propose a Machine-learning approach to screening the Materials genome. This novel approach will allow us to uncover essential characteristics that govern MOF-water interactions in the context of structure-activity relationships (SAR). State of the art in-house descriptors that uniquely classify porous materials, that, when trained with a deep neural network will enable the prediction of complex water behavior within these materials. It is anticipated that this will unlock a number of hitherto undiscovered similarities between structures, and will be a key parameter in our machine-learning program. The resulting model, trained from water data computed in tens of thousands of structures, will then be validated and applied to the millions of structures in the Materials Genome database. This will be at only a fraction of the cost associated with brute-force screening. Not only will this approach allow us to screen millions of MOFs for performance descriptors, it will also reveal important SARs greatly adding to our fundamental understanding of water adsorption in porous media. Such knowledge will allow us to postulate much needed guidelines on how to tailor a MOF with hydrophobic(/-philic) properties optimized for any given application.

PTILS

Project Title: PTILS
Project Leader: Enrico Bodo
Resource Awarded:  30 million core hours on Marconi – KNL

Details

Team Members:
Andrea Le Donne, University of Rome “Sapienza”, ITALY

Abstract:
Ionic liquids (ILs) are liquid salts made by sterically mismatched molecular ions and that present a low melting point because electrostatic interactions are weakened by charge delocalization and lattice formation is frustrated by entropic effects. In contrast to many traditional solvents, ILs have negligible volatility and, for this reason, represent a class of “green” solvents that should be inherently safer and environment-friendly. A special class of ILs is represented by Protic Ionic Liquids(PILs) where the liquid salt is simply obtained by an acid base reaction. When the difference of pKa between the acid and the base is large (>8 pKa units) the resulting liquid is completely ionized.The acidic proton has been transferred to the base, it is strongly bound to it and proton transfer is not possible. Conduction in these liquids is, as in other ILs, due to ion drift and inversely proportional to the liquid viscosity. One way to promote the formation of fast charge carriers is to have proton transfer from one molecule to another through what is commonly known as Grotthuss mechanism where charge is transferred through the H-bond chains. One possibility is that of using PILs made by molecular ions that carry an additional protic function. Compounds obtained from choline and aminoacid anions such as [Ch][Cys] and [Ch][Asp] have the right ingredients. The former has a weak -SH acid terminal and the latter has a second carboxyl group. The main focus of this project is the theoretical description of a series of liquids composed by choline cations combined with amino-acid anions with additional protic functions such as cysteine [Cys], Se-cysteine [Sec], omocysteine [HCys] and aspartic and glutamic acid [Asp], [Glu]. Given the complexity of these materials and the fact that chemical bond breaking and formation is taking place dynamically during their equilibrium dynamics, the use of ab-initio molecular dynamics is mandatory. The description of the bulk phase will be performed using Car-Parrinello molecular dynamics (CPMD) in order to describe their structure and the mechanisms of proton transfer processes that, in turn, could prompt their use a new solvents for electrochemistry.This study has the potential of opening an unexpected and advantageous route to the design of new, unconventional materials whose chemical/electrochemical properties can be tuned upon changes of their molecular structure or of the positive counterion.

 SPINMOLSURF

Project Title: SPINMOLSURF
Project Leader: Eliseo Ruiz
Resource Awarded: 17.2 million core hours on MareNostrum

Details

Team Members:
Jordi Cirera, Universitat de Barcelona, SPAINJorge Echeverría, Universitat de Barcelona, SPAINAlejandro Martín, Universitat de Barcelona, SPAIN

Abstract:
The interaction between deposited molecules and the surfaces could result in new magnetic properties when compared with the isolated subsystems. Thus, the term Spinterface has been introduced to identify the changes in the spin distributions, and consequently in the magnetic properties, due to the interaction between molecules and surfaces. The electronic structure analysis of the Spinterface is crucial to study the magnetoresistance properties of molecular-based spintronic devices. In Spintronics, the key device component is the spin valve formed by a non-magnetic metal film coated between two magnetic metal layers. A step beyond the actual spintronic technologies implies the replacement of the metal or magnetic layer by molecules, whose spins would be used in a similar manner. This project focuses in the interaction of magnetic molecules and transition metal complexes either with metallic surfaces (gold) or magnetic oxides (magnetite Fe3O4). Thus, the first goal is to perform a theoretical study using periodic DFT methods applied to a family of metalloporphyrins with Co(II), Ni(II) and Cu(II) as cations anchored to gold electrodes through thiolpyridine ligands (in collaboration with experimental groups). The complexity of such systems can be enhanced by stacking two or three metalloporphyrins using some bridging ligands such as, for instance, pyrazine. The second aim of our research plan is to carry out a theoretical study based on DFT methods using periodic models that can provide with the exchange coupling constants for a clean Fe3O4 surface before and after the deposition of the molecule (intrasurface and molecule-surface exchange constants). Our study must provide with answers such as which role plays the inclusion of these molecules and its implications in the variation of the magnetic properties of the metal complexes and/or the magnetic surface.

NANOMETALS – Atomically precise, ligand-sabilized, silver-based nanoclusters in the metallic limit

Project Title: NANOMETALS – Atomically precise, ligand-sabilized, silver-based nanoclusters in the metallic limit
Project Leader: Hannu Hakkinen
Resource Awarded: 20.6 million core hours on MareNostrum

Details

Team Members:
Sami Kaappa, University of Jyvaskyla, FINLANDSami Malola, University of Jyvaskyla, FINLANDElli Selenius, University of Jyvaskyla, FINLAND

Abstract:
In this project, large-scale density functional theory calculations are applied to investigate the electronic structure of atomistically precise ligand-stabilized silver-based (silver and intermetallic Ag-Au, Ag-Pd and Ag-Pt) nanoclusters in the limit where the clusters become metallic. The studied clusters have been synthesized and structurally characterized to atomic precision. This offers, for the first time, an opportunity for direct correlation of computed STM topography, STS tunnelling properties, and plasmonic properties to known silver-based nanocluster structures. The results are relevant for understanding electronic conductance through the clusters, Coulomb blockade phenomena, development of plasmonic response and stabilizing factors from the organic ligand shell. These issues are important for future design of nano-structured materials and devices made out of silver-based nanocluster building blocks.

 PETMOI – Photo- and electrocatalysis at transition metal oxide interfaces

Project Title: PETMOI – Photo- and electrocatalysis at transition metal oxide interfaces
Project Leader: Daniel Opalka
Resource Awarded: 25.2 million core hours on Curie

Details

Abstract:
The production of fuels such as molecular hydrogen from renewable energy is widely considered one of the grand challenges to our societies. In this project two routes toward sustainable production of hydrogen are investigated: the photocatalytic and the electrochemical decomposition of liquid water. Atomistic models are developed to resolve the fundamental mechanisms in suitable catalyst materials and at water/catalyst interfaces to provide rational design principles for novel materials to achieve sustainable and commercially viable hydrogen production. The research focuses on transition metal oxide based materials which are among the most active catalysts, but exhibit a challenging electronic structure. Based on calculations from first principles thermodynamic stability, interfacial processes and doping of nanostructures are analysed.

SIDEX – SIze DEpendence of the EXcitonic properties of MoS2 nanoribbons from many-body perturbation theory

Project Title: SIDEX – SIze DEpendence of the EXcitonic properties of MoS2 nanoribbons from many-body perturbation theory
Project Leader: Pino D’Amico
Resource Awarded: 59 million core hours on Piz Daint

Details

Team Members:
Andrea Ferretti, Consiglio Nazionale delle Ricerche, ITALYElisa Molinari, Consiglio Nazionale delle Ricerche, ITALYDeborah Prezzi, Consiglio Nazionale delle Ricerche, ITALY

Abstract:
In the past few years the investigation of spontaneus polarization effects occurring in nanoscale systems has attracted much attention. The formation of 1D wires of carriers can be obtained and interesting physical phenomena related to the interplay between confinement and charge accumulation can be investigate also in view of possible application in solar-energy devices. In particular nanoribbons of MoS2 have been selected as a case study and investigated for different kind of edge terminations within the framework of Density Functional Theory, showing that the formation of 1D wires of carriers at the edges of the ribbons is a general trend in such systems. The spatial confinement and the polarization of the charges on the edges makes the MoS2 nanoribbons a good candidate to investigate electron-electron interaction in nanoscale systems. Results on the optical and the electronic properties of MoS2 bulk monolayers are present in literature but the study of the quasi-particle corrections and the excitonic effects due to the interaction is lacking. In this context, SIDEX will study the effect of the interaction in the electronic structure and in the optical spectrum of MoS2 nanoribbons as a function of the ribbon width. In particular the quasi-particle spectrum will be addressed by means of the GW approximation within the Many Body Perturbation Theory (MBPT) and the optical properties will be investigated through the Bethe Salpeter Equation (BSE) which takes into account electron-hole interaction. The effect of Spin-Orbit interaction will be also addressed for a limited number of cases, once the trends of the optoelectronic properties with respect of the ribbon size have been obtained.

Understanding of the molecular basis of the decoding process in prokaryotic and eukaryotic ribosomal A-sites

Project Title: Understanding of the molecular basis of the decoding process in prokaryotic and eukaryotic ribosomal A-sites.
Project Leader: Modesto Orozco
Resource Awarded: 31.9 million core hours on Piz Daint

Details

Team Members:
Pablo Ignacio Dans, Institute for Research in Biomedicine (IRB-Barcelona), SPAINHansel Gomez, Institute for Research in Biomedicine (IRB-Barcelona), SPAIN

Abstract:
The central dogma of molecular biology is an explanation of the flow of genetic information within a biological system. It states how information is transfered from DNA througth messenger RNA (mRNA) to final proteins. This mechanisms in two steps, involving a transcription and a translation, is essential for life, being the connection between an “inert“ code writed on the DNA to proteins, which are the main molecular effectors in any cell. Translation of an mRNA to a functional protein underlies complex interaction networks within a specialized cellular organell called ribosome, which offers high fidelity and eficiency during the process. Ribosomes, huge RNA-protein complexes, are formed by a small and a large subunit. The codon recognition occurs inside the A-site on the small subunit and provides for canonical matching of mRNA codon to tRNA anticodon. This process depends on the extra-helical flipping of two conserved adenines in the A-site, known as A1492 and A1493. While the secondary structure of the A-site is quite similar for eukaryotic and prokaryotic organisms, there are some sequential discrepancies which are the source of there markable differences between adenine flipping, especially regarding the effect of aminoglycoside antibiotics. Due to its major role in codon recognition, calculation of the free energy profiles along the extra-helical flipping of A1492/3 in different environments and the investigation of its dynamics and sequence dependency can give new insights on the mechanism of translational fidelity. This project focuses on the differences between the energetics of the extra-helical flipping of A1492/3 inside bacterial and yeast A-sites in different environment, using extensive Molecular Dynamics simulations, to obtain a better understanding of the decoding process in bacterial and yeast ribosomes.

top

Earth System Sciences (2)

ROMEO – Understanding the role of mesoscale eddies in the global ocean

Project Title: ROMEO – Understanding the role of mesoscale eddies in the global ocean
Project Leader: Doroteaciro Iovino
Resource Awarded: 67.7 million core hours on Marconi – KNL

Details

Team Members:
Andrea Cipollone, Fondazione Centro Euro-Mediterraneo sui Cambiamenti Climatici, ITALYPier Giuseppe Fogli, Fondazione Centro Euro-Mediterraneo sui Cambiamenti Climatici, ITALYSimona Masina, Fondazione Centro Euro-Mediterraneo sui Cambiamenti Climatici, ITALYEnrico Scoccimarro, Fondazione Centro Euro-Mediterraneo sui Cambiamenti Climatici, ITALYAndrea Storto, Fondazione Centro Euro-Mediterraneo sui Cambiamenti Climatici, ITALY

Abstract:
The complexity of mesoscale dynamics is a challenge for the ocean modelling community. While most global ocean/climate models have resolutions which are too coarse for eddies to form, the ROMEO project aims to more credibly simulate mesoscale dynamics and its role in climate evolution, by delivering a frontier-resolution representation of the global ocean circulation. Here, a global eddying ocean model, GLOB16, based on state-of-the-art NEMO framework (Nucleus for European Modelling of the Ocean), will be used to produce a long hindcast simulation in order to investigate how mesoscale features (eddies, fronts, sea-surface temperature gradients) govern the mean state and ocean variability. GLOB16 has 1/16 horizontal resolution at the Equator (increasing poleward) which allows to realistically represent the life cycle of baroclinic eddies almost over the entire domain. So far, GLOB16 represents the NEMO global configuration having the highest horizontal resolution. The strategy adopted to run the experiment follows a well-established protocol in the ocean modelling community, proposed by the CLIVAR Working Group on Ocean Model Development (WGOMD). It provides a framework to evaluate ocean model performance, to study mechanisms of ocean phenomena and their variability on different timescales. The protocol consists in simulating 5 repeating cycles of the atmospheric forcing dataset. In ROMEO, we will cover the period 1976 to 2015, the meteorological satellites era. This experiment design provides a simulated oceanic state in equilibrium at least in the upper ocean, but also credibly represents intermediate and deeper layers. The JRA-55 atmospheric forcing product used here has been bias corrected for ocean applications. Analysis of the GLOB16 fields during the 5th cycle will provide the basis for comparing to available observation-based estimates and assess the enhanced realism in representing mesoscale processes, with increased horizontal and vertical resolution and improved atmospheric surface forcing. GLOB16 constitutes the CMCC most-advanced global ocean component. It serves as the backbone for downscaling methodologies of ocean forecasting and reanalysis to the regional and coastal areas. The exploitation and protection of marine coastal regions require indeed predictive capability at space scales that so far have been forbidden due to the numerical and computational challenges involved.

 Glob15km – Global 15km coupled climate simulations

Project Title: Glob15km – Global 15km coupled climate simulations
Project Leader: Virginie Guemas
Resource Awarded:  24 million core hours on MareNostrum

Details

Team Members:
Mario Acosta, Barcelona Supercomputing Center, SPAINJuan Acosta, Barcelona Supercomputing Center, SPAINRoberto Bilbao, Barcelona Supercomputing Center, SPAINPierre-Antoine Bretonniere, Barcelona Supercomputing Center, SPAINLouis-Philippe Caron, Barcelona Supercomputing Center, SPAINMiguel Castrillo, Barcelona Supercomputing Center, SPAINEleftheria Exarchou, Barcelona Supercomputing Center, SPAINNeven Fuckar, Barcelona Supercomputing Center, SPAINDomingo Manubens, Barcelona Supercomputing Center, SPAINMartin Menegoz, Barcelona Supercomputing Center, SPAINChloe Prodhomme, Barcelona Supercomputing Center, SPAINOriol Tinto, Barcelona Supercomputing Center, SPAINEtienne Tourigny, Barcelona Supercomputing Center, SPAIN

Abstract:
Recent studies have established that the typical atmospheric and oceanic resolutions used for the CMIP5 coordinated exercise (Coupled Model Intercomparison Project, phase 5), i.e around 40km-150km globally, are a limiting factor to correctly reproduce climate mean state and variability. BSC has developed a coupled version of the EC-Earth 3.2 climate model at a groundbreaking resolution of about 15km in all the climate system components (ocean, sea ice, land and atmosphere). To date, BSC has been able to run two years of simulation on marenostrum 3. The HighResMIP coordinated exercise, as part the Sixth Phase of the Coupled Model Intercomparison Project (CMIP6), offers a framework for building a large multi-model ensemble of high resolution simulations with a low resolution counterpart following a common experimental protocol, i.e. a common integration period, forcing and boundary conditions. This coordinated exercise will allow for identifying the robust benefits of increased model resolution based on multi-model ensemble simulations. The Glob15km project proposes to follow the entire HighResMIP protocol for coupled climate simulations with this ground-breaking resolution configuration of EC-Earth 3.2. Its experimental protocol consists in running a 50-year spinup under perpetual 1950 conditions followed by: 1) a historical simulations covering the 1950-2050 period, 2) a control simulation under perpetual 1950 conditions run for 100 years. This experimental protocol has been chosen as a compromise between limiting the computational cost to ensure that a maximum number of participating institutes can afford it and allowing for a minimization (thanks to the spinup) and subtraction (thanks to the control) of the model drift. This makes a total of 250 years of simulation. BSC will therefore contribute to HighResMIP with 3 different resolutions : a standard resolution version (approximately 100km) and a high resolution version (approximately 25km) resolution with its internal resources as well as an extremely high resolution version (approximately 15km) thanks to the Glob15km project. BSC will hence not only contribute with the highest resolution used for HighResMIP but also with the set of simulations allowing for the most robust assessment of the impact of increasing resolution worldwide. These simulations will represent an extensive source of information for the writing of the next Assessment Report of the Intergovernmental Panel on Climate Change which will take place in 2018. Our main scientific objective will be to pin down physical and dynamical reasons behind differences in model representation induced by resolution change. Process-based assessment will focus on the representation of mean state, variability and teleconnections on a wide range of timescales. The applying team will focus on a large range of processes that can be affected by resolution: sea ice dynamics and thermodynamics, El Nino Southern Oscillation and tropical-extratropical teleconnections, monsoons and heat waves and droughts, Tropical Instability Waves, the Gulf Stream and its influence on the atmosphere, the jet streams and Euro-Atlantic blockings, tropical cyclones.

top

Engineering (8)

TREC – Time resolved evolution of the energy-containing scales in turbulent channel flow at Ret=5000

Project Title: TREC – Time resolved evolution of the energy-containing scales in turbulent channel flow at Ret=5000
Project Leader: Javier  Jimenez
Resource Awarded: 122.4  million core hours on Piz Daint

Details

Team Members:
Adrian G. Gutierrez, Universidad Politecnica Madrid, SPAINMiguel P. Encinar, Universidad Politecnica Madrid, SPAINAlberto Vela-Martin, Universidad Politecnica Madrid, SPAIN

Abstract:
Wall-bounded flows play an important part in numerous common applications, and have been intensively studied for over a century. About 5% of the total energy used by advanced economies, and a disproportionate amount of the resulting CO2 emissions, are due to turbulent wall friction. One of the most powerful tools for this study has proved to be direct numerical simulation, especially if time-resolved series of high-Reynolds number flows can be stored for postprocessing. In particular, it is known that most of the velocity difference and energy dissipation at high Reynolds numbers resides in the “logarithmic” layer that forms at intermediate distances from the wall, but its structure remains controversial. Even if DNS has allowed a lot of progress in the last decade, it is not even completely clear if the dynamics of the structures in the logarithmic layer are local, or if they are caused by events nearer or farther from the wall. The formulation of possible control strategies that manipulate these structures cannot be tackled until questions such as these can be clarified. This we propose to do by means of a new simulation of a turbulent channel that stores the temporal evolution of the flow for a long enough time to allow the “slow” evolution of these intermediate structures to be made explicitly available for postprocessing. Existing data sets either span too short evolution times, have computational boxes which are too small, or a Reynolds number which is too low, to answer these causality questions completely. This next step in turbulence research has been made possible by the new heterogeneous CPU/GPU systems. A new high resolution hybrid CUDA-MPI code exploits the advantages of GPU co-processors on distributed memory systems, making simulations up to Reynolds numbers Re_tau = 5,000, in large boxes and for sufficiently long evolution times, not only possible but computationally affordable. The main problem remains the cost of storing the time-resolved data, but we will drastically reduce it by retaining only spatially filtered data about the large and intermediate scales that dominate the logarithmic layer, taking care to keep all the variables needed to fully reconstruct the filtered flow at the level of second-order statistics, including the energy and the momentum fluxes. The results will be made publicly available, both on-line and by means of on-site workshops in our department.

 HiLESProp – High Fidelity LES on High Skewed Propellers with Upstream Disturbances

Project Title: HiLESProp – High Fidelity LES on High Skewed Propellers with Upstream Disturbances
Project Leader: Riccardo Broglia
Resource Awarded: 30.6  million core hours on Marconi – KNL

Details

Team Members:
Antonio Posa, CNR-INSEAN, ITALYElias Balaras, The George Washington University, UNITED STATES

Abstract:
Propellers may operate in the wake of upstream rudders, adopted for manoeuvring. The influence of such rudders on the operation of propellers and their wake signature is not well documented in the literature. Such an analysis requires indeed huge computational resources, since an accurate simulation of the wake flow relies upon the use of eddy-resolving techniques, which are still at the forefront of research in naval hydrodynamics and in fluid mechanics more in general. Here we propose the use of high fidelity eddy resolving computations (i.e., Large Eddy Simulations), coupled with a non-conforming grid methodology (Immersed-Boundary), to characterize the wake of a high skewed seven-bladed propeller, under experimental investigation at CNR-INSEAN. Our specific interest here is towards interaction and coupling between tip vortices in the wake of propellers with large blade numbers and with a strong swirl along their radius, together with the impact of such physics on wake turbulence. Especially, we focus on the computation of turbulence statistics and the generation of a database, to use for future hydro-acoustic analyses. Such a work will be guided by the availability of PIV results, already produced by another research group in the same institution. At the same time our three-dimensional fields will allow to extract information that experiments are not able to provide, complementing the findings of the measurements campaign. We also plan to study for three different configurations of an upstream hydrofoil, how a manoeuvring rudder affects the wake of the propeller, in terms of stability of its coherent structures (tip and hub vortices) and in terms of its turbulence signature. The computational methodology and the same solver we are going to utilize in the proposed study were already largely validated on both canonical problems and practical configurations, including rotating machinery and naval hydrodynamic flows and specifically propeller flows, simulated in HPC environment on computational grids composed of billions of cells and supercomputers with huge core counts. Our earlier works on propellers proved that our numerical approach and solver are well suited to simulate such class of flows, showing good comparisons with experiments and ability to capture the complex physics of the instability of tip vortices, typical of the wake of multi-bladed high skewed propellers. Significant computational resources are obviously required to utilize the full potential of our numerical tools and provide innovative results.

DiNoSAUR – DNs of SquAre dUct at high Reynolds number

Project Title: DiNoSAUR – DNs of SquAre dUct at high Reynolds number
Project Leader: Sergio Pirozzoli
Resource Awarded: 40  million core hours on Marconi – KNL

Details

Team Members:
Francesco Grasso, CNAM, FRANCEDavide Modesti, CNAM, FRANCEPaolo Orlandi, Sapienza, University of Rome, ITALY

Abstract:
We study turbulent flows in square ducts at high enough Reynolds number to reach flow conditions which are representative of fully developed turbulence, and much higher than previous studies. The main scientific goal is to establish the nature and the dynamical relevance of the secondary motions which arise because of the duct geometry. First, we aim at establishing which is the correct velocity scale for those motions, and how they impact the global flow properties, namely friction and heat transfer. For that purpose, we carry out direct numerical simulations (DNS) up to bulk Reynolds number Reb=82000 (where Reb = 2 h ub / ν, 2 h being the duct side length, ub the duct bulk velocity, and ν the fluid kinematic viscosity), about a factor of four higher than reached in previous DNS. Numerical simulations will be carried out using a compressible Navier-Stokes solver made to operate at low Mach number, in such a way that the flow behavior is effectively incompressible. Despite stronger time step restrictions, this approach is deemed to be more effficient for use on massively parallel computers as it allows three-dimensional domain decomposition, and it limits the use of collective data transfer operations. Runtime statistics will be collected for the streamwise vorticity and momentum equations, which will allow to shed light on the mechanisms responsible for mean cross-flow sustainment, and on the relative importance of the budget terms, with the ultimate goal of developing predictive models, and to provide useful data for the design of effective turbulence models based on the RANS approach and of wall models for LES. Last, we expect that the understanding of flows in ducts with complex shape can also shed light on more canonical flows, as it allows to probe in greater detail inner-outer layer interactions, and to distinguish scalings associated with the local wall friction and with the imposed pressure gradient.

COSI – COntrolled Source sImulation

Project Title: COSI – COntrolled Source sImulation
Project Leader: Florian Bleibinhaus
Resource Awarded: 15  million core hours on Marconi – Broadwell

Details

Team Members:
Kormann Jean, Montanuniversitaet Leoben, AUSTRIAKatrin Peters, Montanuniversitaet Leoben, AUSTRIACornelia Tauchner, Montanuniversitaet Leoben, AUSTRIAJens Zeiß, Montanuniversitaet Leoben, AUSTRIAJosep de la Puente, Barcelona Supercomputing Center (BSC), SPAINMiguel Ferrer, Barcelona Supercomputing Center (BSC), SPAINNatalia Gutierrez, Barcelona Supercomputing Center (BSC), SPAINMauricio Hanzich, Barcelona Supercomputing Center (BSC), SPAINJuan Esteban Rodriguez, Barcelona Supercomputing Center (BSC), SPAIN

Abstract:
Quarries, and occasionally open pit mines, below called blasting site, are increasingly being established near to communities, and existing blasting sites are being encroached by expanding communities. Thus the unavoidable side effects of rock blasting, such as ground vibrations, air blast, fly-rock, are an issue of increasing importance. This is enhanced by a decreasing acceptance level of the public to such disturbances. In Europe, ground vibrations from blasting are subject to legal restrictions and norms. Houses and other structures are classified as to the vibration level they are supposed to tolerate without creating structural damage. The operation permit of a blasting site demands that those limits are not exceeded. Additional limits may be imposed based on public disturbance considerations. Those limits are often expressed in terms of peak particle velocity (PPV) as a function of frequency. Mechanical structures in general, including buildings, have eigenfrequencies around which an excitation, e.g. by ground vibrations, will cause significantly more damage compared to other frequencies. These eigenfrequencies lie typically in the range 2-100 Hz for residential houses and should be avoided. The blasting site operator has several tools by which he can influence the PPV levels that radiate out from the blasting site. He could, e.g., decrease the size of the charges that detonate using smaller drill-holes, or using decked charges, or using fewer blast-holes in a round. Such actions are most probably associated with higher costs and poorer productivity. He could also change the initiation direction and delay times in the round while avoiding the other cost driving actions. Most work of the latter type is based either on trial-and-error, or on very simplified models that don’t take the actual site specific factors into account, such as the viscoelastic subsurface properties, and the topography. Having a systematic way of adapting the geometry and initiation pattern of a blasting round so that the vibration amplitudes at critical frequencies were minimized would allow the blasting site to reach a more economic compromise with the legitimate claims of the public

High-fidelity LES/DNS simulation of full-span turbine passage

Project Title: High-fidelity LES/DNS simulation of full-span turbine passage
Project Leader: Koen Hillewaert
Resource Awarded: 68.5  million core hours on MareNostrum

Details

Team Members:
Jean-Sebastien Cagnone, Cenaero, BELGIUMmichel rasquin, Cenaero, BELGIUMAntonio Garcia-Uceda Juarez, NUMECA, BELGIUMFrancesco Bassi, Università degli Studi di Bergamo, ITALYAlessandro Colombo, Università degli Studi di Bergamo, ITALYDavid Gutzwiller, NUMECA USA, UNITED STATES

Abstract:
The proposed research aims at high-fidelity LES/DNS simulations of a full-span low-pressure turbine, including the diverging end-walls of the tunnel section, and the resulting corner-vortex flow. This computational campaign is organized under the umbrella of the TILDA H2020 European project, focused on innovative methods for accurate predictions of complex aerodynamic turbulent flows. This study will concentrate on this previously unreleased three­-dimensional turbine configuration, whose experimental measurements will be made publicly available in parallel with the simulation data. With this new test case, we plan to analyse the horse-shoe vortex phenomenon, initiated by the interaction of the incoming wall boundary-layer with the blade profile. A primary outcome is the unprecedented understanding of this separation mechanism, and the development of a high-fidelity database dedicated to the improvement of reduced-fidelity models such as LES wall-function, transition models and RANS calibration. A second important outcome is the confrontation of three different computational methods with distinct numerical properties, and an assessment of their turbulence-resolving capabilities for complex three-dimensional compressible flows.

Large-Eddy-Simulations of the unsteady aerodynamics of oscillating airfoils at moderately high Reynolds numbers.

Project Title: Large-Eddy-Simulations of the unsteady aerodynamics of oscillating airfoils at moderately high Reynolds numbers
Project Leader: Dan Henningson
Resource Awarded: 40  million core hours on Hazel Hen

Details

Team Members:

Abstract:
Prediction of transition and flow separation is crucial for adequate design and control of aircraft systems. This necessarily implies the in-depth understanding of boundary layer characteristics in unsteady applications, where both transition and separation may dramatically alter flow characteristics. Such scenarios are especially relevant for Natural Laminar Flow (NLF) airfoils which can exhibit a sensitive dependence on the angle of attack, leading to large changes in aerodynamic forces for even small changes in the angle of attack. Such small changes are inevitable in normal flight conditions and can arise from small structural deformations of the wing structure or incoming gusts. In such cases, current simple models of unsteady flow prediction fail to be reliable and a more detailed understanding is required for better control strategies and aircraft wing designs. A recent report by NASA has identified prediction of transition and separation in general unsteady flows as one of the fundamental challenges of CFD for the coming decade. Additionally, such studies necessarily need to be undertaken in the relevant Reynolds number range of the intended application since transition and separation characteristics on wings can change substantially with Reynolds numbers. The current work investigates the unsteady flow characteristics for a specific case of small-amplitude pitch oscillations in an NLF airfoil. Such small-amplitude pitch oscillations mimic the changes in operating conditions due to structural deformations. The NLF airfoil exhibits the sensitive dependence on angle of attack, with large changes in transition and separation characteristics within a small angle of attack change. Wall-resolved large-eddy-simulations are performed to investigate the flow-field evolution in the unsteady case with a focus on flow transition, separation and the mutual interaction of these two critical boundary layer characteristics.

PlaJet – Mixing enhancement for plasma-controlled turbulent jets

Project Title: PlaJet – Mixing enhancement for plasma-controlled turbulent jets
Project Leader: Sylvain  Laizet
Resource Awarded: 32  million core hours on Marconi – KNL

Details

Team Members:
Eric Lamballais, P, FRANCEVasilis Ioannou, Imperial College London, UNITED KINGDOMJ. Christos Vassilicos, Imperial College London, UNITED KINGDOMTatsuya Yasuda, Imperial College London, UNITED KINGDOMYi Zhou, Imperial College London, UNITED KINGDOMCharles Moulinec, STFC Daresbury Lab, UNITED KINGDOM

Abstract:
In recent years, the development of devices known as plasma actuators has advanced the promise of controlling flows to achieve and improve beneficial effects and/or to reduce negative effects of turbulence. Dielectric barrier discharge (DBD) plasma actuators consist of two electrodes, one exposed to the ambient fluid and the other covered by a dielectric material. When an Alternating Current (A.C.) voltage is applied between the two electrodes the ambient fluid over the covered electrode ionizes. This ionized fluid is called the plasma and results in a body force vector which exchanges momentum with the ambient, neutrally charged, fluid. Studies based on DBD plasma actuators have covered a wide range of engineering applications such as viscous drag reduction, boundary layer separation control in low-speed flows, shock wave modification, wave drag reduction in supersonic and transonic flows and supersonic boundary layer transition control. However, very little has been done for the manipulation of turbulent jets. The control of turbulent jets has significant consequences for many applications. For instance, the reduction of aircraft noise and mixing enhancement in various industrial problems such as the dispersion of pollutants from combustion. In combustors it is important to enhance turbulent mixing of the chemical species in order to achieve high combustion efficiency and to reduce the emission of pollutants. In jet engines the aerodynamic noise can be reduced by controlling flow unsteadiness to minimise the environmental impact of air transport. For this project, turbulence-resolving simulations will be carried out in order to demonstrate the potential of DBD plasma actuators located near the nozzle-exit to act on the spatial evolution of turbulent jets. This idea is to propose efficient active control solutions (which can be turned off if not needed in order to save energy) for turbulent jet mixing enhancement. This research project is a first step in the development of new technologies based on plasma actuators not only for mixing enhancement but also for noise reduction purposes and for a better efficiency of jet engines. The large number of parameters (location of the actuator, orientation, size, relative placement of the embedded and exposed electrodes, applied voltage, frequency) affecting the performance of plasma actuators makes their development, testing and optimisation a very complicated task. As a result, experimental approaches require numerous high-cost and time consuming trial-and-error iterations. Computational Fluid Dynamics (CFD) has the potential to investigate rigorously how DBD plasma actuators can be designed to manipulate a turbulent jet in order to achieve mixing enhancement.

FIRELES: FInite-Rate chEmistry modeling for Large-Eddy Simulation of turbulent reactive flows in realistic geometries

Project Title: FIRELES: FInite-Rate chEmistry modeling for Large-Eddy Simulation of turbulent reactive flows in realistic geometries
Project Leader: Vincent MOUREAU
Resource Awarded: 23.5  million core hours on Curie

Details

Team Members:
Pierre BENARD, CORIA – CNRS UMR6614, FRANCEGhislain LARTIGUE, CORIA – CNRS UMR6614, FRANCE

Abstract:
The FIRELES project is dedicated to the prediction of combustion performances in small to medium-size methane/air burners using Large-Eddy Simulation, finite-rate chemistry and high- performance computing. The prediction of conversion efficiency and pollutant emissions is particularly challenging as they result from very complex interactions of turbulence, chemistry, and heat exchanges at very different space and time scales. Large-Eddy Simulation (LES) has proven to bring significant improvements in the prediction of reacting turbulent flows and has shown its potential to accurately predict pollutant formation in complex real combustors. Over the last five years, a dedicated finite-rate chemistry module has been developed in the massively parallel YALES2 solver, which relies on operator splitting, robust stiff integration of the chemical source terms with analytical Jacobian and dynamic task sharing, and implicit diffusion. This module can be coupled to the dynamic thickened flame model to take into account the sub-grid flame/turbulence interactions and to a dynamic mesh adaptation module, which enables to refine and coarsen the mesh locally to follow the premixed front. In order to fully demonstrate the capabilities of LES coupled to finite-rate chemistry, a semi-industrial aeronautical swirl burner has been chosen, which operates with a premixed stream of air and methane and which has been extensively studied experimentally and used for LES validation. Despite the large number of studies of this so-called PRECCINSTA burner, there still remain a number of open questions concerning the combustion performances and pollutant emissions. None of the numerical studies achieve a correct reproduction of both the temperature field and the CO mass fraction due to the limited accuracy of the chemistry description and the assumption of adiabatic walls, which is detrimental in corner recirculation zones. The objective of the project is to reach very fine mesh resolutions in order to capture most of the flame/turbulence interactions and to study the influence of the chemical mechanism on the flame dynamics.

top

Fundamental Constituents of Matter (8)

Radiation imprint of ultra-intense laser heating of solids

Project Title: Radiation imprint of ultra-intense laser heating of solids
Project Leader: Thomas Kluge
Resource Awarded: 109 million core hours on Piz Daint

Details

Team Members:
Michael Bussmann, Helmholtz-Zentrum Dresden-Rossendorf, GERMANYAxel Hübl, Helmholtz-Zentrum Dresden-Rossendorf, GERMANYRene Widera, Helmholtz-Zentrum Dresden-Rossendorf, GERMANY

Abstract:
We propose to perform – to our knowledge for the first time – 3D particle-in-cell simulations of the interaction of an ultra-intense short pulse laser with solids including the picosecond time span prior to the main laser pulse, in order to study the influence of material and pre-pulse laser conditions on the plasma dynamics and its imprint on Bremsstrahlung and synchrotron radiation, generated by relativistic laser accelerated electrons. This time period is thought to be decisive for the following main pulse interaction, yet it is poorly explored – partly due to the immense numerical needs to cover the plasma kinetically with full precision. Here, we will bridge scales hitherto unaccessible, from attosecond plasma oscillations over few femtosecond laser oscillations and transient, non-equilibrium plasma dynamics on the tens of femtosecond laser duration to picosecond preplasma development. The aim is to infer radiative signatures of the plasma dynamics and link it to isochoric heating and instability development and other complex dynamics. Beyond gaining a fundamental understanding of the governing fundamental principle plasma dynamics, the results will be used to develop a novel diagnostics analyzing the Bremsstrahlung and synchrotron radiation in order to experimentally probe the sub ps interaction.

JOREK_P; Non-linear MHD modelling of pellet injection for the control of MHD instabilities in fusion plasmas

Project Title: JOREK_P; Non-linear MHD modelling of pellet injection for the control of MHD instabilities in fusion plasmas
Project Leader: Shimpei Futatani
Resource Awarded: 15 million core hours on MareNostrum

Details

Team Members:
Shimpei Futatani, Barcelona Supercomputing Center, SPAINMervi Mantsinen, Barcelona Supercomputing Center, SPAINXavier Saez, Barcelona Supercomputing Center, SPAINGuido Huijsmans, CEA, FRANCEDi Hu, ITER, FRANCEAlberto Loarte, ITER, FRANCEStanislas Pamela, CCFE, UNITED KINGDOM

Abstract:
The project is dedicated to the nuclear fusion physics research in close collaboration with existing experimental fusion devices and the ITER organization (www.iter.org) which is an huge international nuclear fusion R&D project. The goal of the ITER project is to demonstrate an energy production by nuclear fusion which is the reaction that powers the sun. One of the ideas of the nuclear fusion on the earth is that the very high temperature ionized particles, forming a plasma can be controlled by a magnetic field, called magnetically confined plasma. This is essential, because no material can be sustained against such high temperature reached in a fusion reactor. The magnetic field structure is torus, donut-shape structure in order to close the magnetic field. This is called ‘Tokamak’ which is the currently furthest developed magnetic confinement devices. However, it is a demanding task to achieve a sufficiently good confinement for a ‘burning plasma’ due to various kinds of plasma instabilities. One of the critical undissolved problems is MHD (MagnetoHydroDynamics) instabilities at the plasma boundary, called Edge Localized Modes (ELMs). Fusion reactors can be damaged by ELMs as they release large energy from plasmas to the reactor wall. One of the methods of ELM mitigation is injection of small pellet (small deuterium ice cube) to induce small ELMs before large ELM occurs. The technique is experimentally proved [Baylor 2013, Lang 2014, Loarte 2014], but the theoretical understanding is not yet established as well as the numerical simulation points of view. This proposal aims to improve the understanding of the physics processes involved in ELM control by pellet injection, using the three-dimensional non-linear MHD simulation code JOREK which solves pellet ablation physics self-consistently with the MHD activity. This work is necessary to be carried out to establish the requirements for the pellet size and speed for the ITER pellet injector for ELM control. In addition, the consequences for the power fluxes at the ITER divertor during pellet triggered ELMs and the triggered ELM density losses will be evaluated and consequences for ITER operation will be estimated.

JOREK_MHD: Validation of simulations of MHD instabilities in tokamak plasmas for extrapolation to ITER
Project Leader: Stanislas Pamela

Project Title: JOREK_MHD: Validation of simulations of MHD instabilities in tokamak plasmas for extrapolation to ITER
Project Leader: Stanislas Pamela
Resource Awarded: 30 million core hours on MareNostrum

Details

Team Members:
Matthias Hoelzl, IPP Garching, GERMANYGuido Huijsmans, CEA, FRANCEEric Nardon, CEA, FRANCE

Abstract:
The research towards nuclear fusion energy production has entered a new phase with the ongoing construction of the ITER machine in Cadarache France. In the ITER tokamak, plasmas of Deuterium and Tritium will be confined by strong magnetic fields and heated to the required temperature of ~3×108 C. One of the key issues in nuclear fusion research is the handling of the output power onto the plasma-facing components. In ITER, plasma facing components are designed to withstand the expected steady state heat loads. However, uncontrolled MHD instabilities may cause fast, transient energy exhausts from the plasma, that could potentially erode/melt those plasma facing components. Although experimental extrapolation from current experiments is available, and although control/mitigation of these MHD events are being developed on current machines in view of ITER, an improved physics understanding of these MHD instabilities (and their control) through numerical simulations could provide more accurate predictions for future devices like ITER. The current state of the JOREK code development is such that validation has now become one of the main priorities. This requires high resolution simulations approaching as much as possible realistic experimental conditions and plasma parameters. These large scale non-linear MHD simulations can only be executed on tier-0 resources. There is currently strong pressure from the European and international fusion communities for JOREK to be validated against experiments and produce predictions for ITER before its operation. This quantitative validation requires not only advanced HPC facilities, but also a large availability of resources on these cluster due to the large amount of simulations required to provide modelling of various physics experiments on multiple machines. Although some support is provided by EUROfusion with the MARCONI cluster, additional resources are necessary to obtain the validation of JOREK on this short time scale. We hope this project proposal can convince the PRACE team to allow some HPC support for the JOREK team.

Light front wave functions in coordinate space

Project Title: Light front wave functions in coordinate space
Project Leader: Gunnar Bali
Resource Awarded: 50 million core hours on Marconi – KNL

Details

Team Members:
Vladimir Braun, Universitaet Regensburg, GERMANYPeter Georg, Universitaet Regensburg, GERMANYMeinulf Goeckeler, Universitaet Regensburg, GERMANYFabian Hutzler, Universitaet Regensburg, GERMANYPiotr Korcyl, Universitaet Regensburg, GERMANYDaniel Richtmann, Universitaet Regensburg, GERMANYAndreas Schaefer, Universitaet Regensburg, GERMANYWolfgang Soeldner, Universitaet Regensburg, GERMANYPhilipp Wein, Universitaet Regensburg, GERMANYJianhui Zhang, Universitaet Regensburg, GERMANY

Abstract
Protons, neutrons and pions are complicated objects with a very complex internal quark and gluon structure. The most important aspects of the structure of these hadrons if scattered or produced at high momentum transfers, e.g., at the large hadron collider (LHC) at CERN or the planned electron ion collider (EIC) in the US, can be described in terms of parton distribution functions (PDFs), quantum field theoretical analogues of quantum mechanical probability densities, and distribution amplitudes (DAs), which correspond to light front wave functions. The latter objects will be computed for the pion and the kaon in lattice simulations of the underlying theory, QCD. This is important to theoretically describe exclusive processes in upcoming new experiments. Previously, only moments of these light front wave functions could be accessed in such simulations. However, with new algorithmic inventions and methods it is now possible to simulate these directly in coordinate space. We will not only provide accurate results on the pion and kaon DAs but also test the computationally very demanding new approach by confronting the results obtained against the lowest moments computed in the “traditional” way. Coordinate space methods require perturbatively small distances and large momenta to be resolved and, therefore, unprecedentedly small lattice spacings a. Here we will realize a=0.039 fm. Realizing large momenta has only become possible with the momentum smearing technique that was very recently invented by some of us.

MagGen: Magnetic field generation of kinetic plasmas, from onset to turbulence

Project Title: MagGen: Magnetic field generation of kinetic plasmas, from onset to turbulence
Project Leader: Kevin Schoeffler
Resource Awarded: 15 million core hours on MareNostrum

Details

Team Members:
Thomas Grismayer, Instituto Superior Tecnico, PORTUGALGiannandrea Inchingolo, Instituto Superior Tecnico, PORTUGAL

Abstract:
Magnetogenesis, the question as to how magnetic fields throughout the universe were generated, remains a topic of interest relevant to both large galactic size fields, and smaller scaled systems like in accretion discs around black holes. As many astrophysical regimes are in the collisionless regime where kinetic effects of both electrons and ions play an important role, this project aims to take advantage of particle-in-cell simulations (OSIRIS) to help answer some major related questions. Turbulent dynamos, which are thought to be the major source of these fields need to be initiated with small, but significant seed fields. These seed fields may be caused by the Biermann battery from a sudden motion that results in misaligned density and temperature gradients, and fast instabilities such as the MagnetoRotational instability due to free energy in differentially rotating systems, and the kinetic small-scale Weibel instability due to temperature anisotropies. Can small-scale Weibel magnetic fields cascade to the large scale fields measured today? Do kinetic electron scales affect the growth of the MagnetoRotational instability? Do turbulent plasmas with small mass ions differ from the infinitely massive ions typically studied? We intend to carefully address these important questions and to make a contribution to the understanding of the workings behind Magnetogenesis.

Characterization of turbulence and flow generation in Tore Supra tokamak

Project Title: Characterization of turbulence and flow generation in Tore Supra tokamak
Project Leader: Timo Kiviniemi
Resource Awarded: 30 million core hours on MareNostrum

Details

Team Members:
Laurent Chôné, Aalto University, FINLANDSusan Leerink, Aalto University, FINLANDPaavo Niskala, Aalto University, FINLANDRonan Rochford, Aalto University, FINLAND

Abstract:
Nuclear fusion as a basis for clean commercial energy production has been intensively investigated worldwide for decades and so far the tokamak magnetic confinement device has been most successful in creating nuclear fusion conditions on earth. A major setback however is caused by turbulence induced transport which causes chaotic property changes in the plasma and requires first principle computer simulations due to its highly non-linear nature. During the last decade the gyrokinetic code ELMFIRE has been developed to study turbulent transport from first principle basis with the Particle-​In-​Cell method while solving for the full gyrokinetic ion and drift kinetic electron distribution which allows for investigation of complex interplay between turbulence and self-consistently evolving background plasma and e.g. our long term efforts on studying the geodesic acoustic modes (GAMs). In the present project, we are aiming to simulate GAMs and other meso-scale oscillations in context of Linear to the Saturated Ohmic Confinement regime (so called LOC/SOC transition) in the Tore Supra tokamak. In SOC also the staircasing of density fluctuations has been observed experimentally which can be investigated with ELMFIRE. Two major issues in developing a commercial fusion power plant are the degradation of plasma confinement due to turbulent transport and the plasma-wall interactions in the region where the plasma particles hit the material wall. Scrape-off-layer (SOL) turbulence is a research area which connects these two. This interplay between the plasma fluxes from the core to SOL, the turbulent transport and flows in open field lines, and the losses to the vessel walls, determine the peak heat loads at the vessel which is a critical issue for the ITER and all future tokamaks. Many aspects of core turbulence are well understood but transport properties are dramatically different in the SOL where e.g. the assumption that the turbulence amplitude is small compared to the mean value is not valid, and the presence of open field lines complicates the analysis. In the SOL region, the solid wall interacts with the plasma controlling impurity dynamics and recycling level. This interaction has an important role in overall plasma confinement and is also assumed to affect the L-H transition power threshold, which is of critical importance for ITER and future devices. The present project aims to investigate SOL flows for Tore Supra tokamak. The results are compared to SOL measurements, including poloidal and radial distribution of density, temperature and electrostatic potential. Parametric dependence of SOL width and width of power load distribution on targets is studied.

UNPIC3D

Project Title: UNPIC3D
Project Leader: François Pechereau
Resource Awarded: 18 million core hours on Curie

Details

Team Members:
Bénédicte Cuenot, CERFACS, FRANCE | Valentin Jonquières, CERFACS, FRANCE | Olivier Vermorel, CERFACS, FRANCE | François Pechereau, CERFACS, FRANCE | Gabriel Staffelbach, CERFACS, FRANCE | Anne Bourdon, LPP, FRANCE | Trevor Lafleur, LPP, FRANCE

Abstract:

Electric propulsion (EP) is a type of space technology used to provide propulsion for satellites by using electric and/or magnetic fields. One of the most successful EP systems to date is the Hall-effect thruster which was first studied in the USSR in the 1960s. The main advantage of EP systems compared with chemical systems is the larger propellant exhaust velocities that can be obtained, which consequently result in lower propellant mass requirements, and hence significant cost savings. This was demonstrated, for example, by the

SMART-1 mission to the moon where a chemical propulsion system would have required 53 tons of propellant, whereas the PPS-1350 Hall-effect thruster (developed by SNECMA) made the journey with only 80 kg of Xenon. With current launch prices of the order of $10 000 per kg, cost savings due to EP systems are of great interest even though manoeuvre times can be longer due to the lower thrust of these devices. In 2012, more than 300 satellites made use of EP thrusters as the main propulsion system. The same year, both ESA and Boeing introduced “fully electric” satellite architectures without any chemical propulsion. These systems have been the main drive in recent worldwide studies on EP. However, despite this renewed interest, many EP technologies, and in particular Hall-effect thrusters, are not yet fully understood. Issues related to anomalous particle transport, current oscillations, and operational lifetime still constitute a scientific challenge.

As the simulation of a full 3D real thruster is still out of reach, it is proposed to compute a 3D sector of a simplified Hall-effect thruster for which experimental data exists. The first objective is to investigate the anomalous particle transport in Hall-effect thrusters conditions by using for the first time fully self-consistent 3D computations, which are regarded as the only way to accurately describe this phenomenon. The second objective is to study the dynamics of current oscillations in Hall-effect thrusters and to numerically reproduce experimental results where current oscillations can be suppressed thanks to non-homogenous azimuthal injection of neutral gas into the thruster.

A unified unified computational protocol for QCD nuclei

Project Title: A unified unified computational protocol for QCD nuclei
Project Leader: Alessandro Lovato
Resource Awarded: 37.5 million core hours on Marconi – KNL

Details

Team Members:
Assumpta Parreño, University of Barcelona, SPAINLorenzo Contessi, University of Trento, ITALYFrancesco Pederiva, University of Trento, ITALYAlessandro Roggero, Institute for Nuclear Theory, UNITED STATESMartin Savage, Institute for Nuclear Theory, UNITED STATESZohreh Davoudi, MIT, UNITED STATESWilliam Detmold, MIT, UNITED STATESPhiala Shanahan, MIT, UNITED STATES

Abstract:

We propose to bridge the gap between the fundamental theory of the strong interactions, Quantum Chromo Dynamics (QCD), and low-energy nuclear observables, such as nuclear masses and radii. We aim to study the evolution of the nuclear chart as a function of the quark-masses, which are input parameters of the Standard Model. A first exploration of the stability of biological membranes, a necessary structure for the existence of carbon-based life, with regard to variations of Standard Model parameters will thus become possible. We will achieve this goal using a nuclear potential derived within a pion-less effective field theory. The low-energy constants will be fitted to Lattice-QCD (LQCD) calculations for few-baryon systems, which constitute an essential part of the present proposal. This potential will employ state-of-the art Quantum Monte Carlo (QMC) techniques to precisely compute properties of 4He, 12C, 16O, 40Ca. These calculations will significantly extend the domain of Lattice-QCD predictions in a model-independent fashion. The Green’s function Monte Carlo (GFMC) and the Auxiliary Field Diffusion Monte Carlo (AFDMC) methods are ideal to solve the Schrödinger equation of the pion-less Hamiltonian. They have no limitations in using spin-dependent two- and three-body forces characterised by stiff short-range behaviour. Nucleon-nucleon scattering amplitudes and the bindings of the deuteron, di-neutron, 3He and 4He will be calculated with high precision at a pion mass of approximately 800 MeV using existing large ensembles of isotropic clover gauge-field configurations at two values of the lattice spacing. Close coordination and feedback between the QMC and LQCD productions, to fine-tune both sets of calculations at intermediate stages of production, is expected to be essential, and our production schedule is designed to accommodate this. All of our codes are highly optimised for standard Intel CPU, KNL and GPU architectures, and have been proven to efficiently scale on Leadership-Class computing facilities, such as Mira and Theta, hosted at Argonne National Laboratory or Titan at Oak Ridge National Laboratory.

top

Mathematics and Computer Sciences (0)

top

Universe Sciences (6)

ProbPhysGrav — Probing fundamental physics with gravity

Project Title: ProbPhysGrav — Probing fundamental physics with gravity
Project Leader: Helvi Witek
Resource Awarded: 15 million core hours on MareNostrum

Details

Team Members:
Diego Blas, CERN, European Organization for Nuclear Research, SWITZERLANDKaty Clough, Georg-August-University, GERMANYMiguel Zilhao, University of Barcelona, SPAINJames Cook, King’s College London, UNITED KINGDOMThomas Helfer, King’s College London, UNITED KINGDOMEugene Lim, King’s College London, UNITED KINGDOMHans Bantilan, Queen Mary University of London, UNITED KINGDOMPau Figueras, Queen Mary University of London, UNITED KINGDOMWilliam Cook, University of Cambridge, UNITED KINGDOMMarkus Kunesch, University of Cambridge, UNITED KINGDOMRoxana Rosca, University of Cambridge, UNITED KINGDOMUlrich Sperhake, University of Cambridge, UNITED KINGDOM

Abstract:
Humankind has always been on a quest to unravel the mysteries of our universe. Despite the recent breakthrough discoveries of gravitational waves, black holes or the Higgs boson, big questions concerning the very nature of gravity are still open. In fact, we still lack a profound understanding of the evolution of our universe and why it is expanding the way it does, or what dark matter – non-visible matter that drives the formation and evolution of galaxies – is made of. Instead of traditional searches using electromagnetic observations or insight from high-energy particle collisions, we will employ black holes and gravitational waves to find new answers to these outstanding questions. For example, the interplay between black holes and dark matter candidates can lead to characteristic features in a gravitational wave signal that is potentially observable with LIGO – the gravitational wave observatory that directly detected gravitational waves for the first time little more than a year ago. The future space-based detector LISA will also be able to pick up these signatures if they come from supermassive black holes that live at the center of most galaxies where we also expect dark matter rich regions. Additionally, LISA will be able to listen to the whispers from inflation, i.e., the early phase of the universe’s evolution that is still poorly understood. We intend to predict the expected gravitational wave signals – called waveforms – so that they can be employed in the template banks of gravitational wave detectors to search for these characteristic signatures. Because the detectors are still on the edge of their observational capabilities, gravitational wave signals typically do not stick out and we need to know what to look for in order to find them. This makes it an even more pressing task to predict them since otherwise we may just miss crucial smoking gun effects that could tell us more about nature’s secrets.

Dynamics of a solar active region

Project Title: Dynamics of a solar active region
Project Leader: Boris Gudiksen
Resource Awarded: 72 million core hours on Marconi – KNL

Details

Team Members:
Mats Carlsson, University of Oslo, NORWAYViggo Hansteen, University of Oslo, NORWAY

Abstract:
The solar corona is extremely hot. The temperature is more than 1 million degrees and no obvious heating mechanism is at play. The magnetic field of the sun is playing a big part in the heating of the corona, but so far we have not had a full understanding of how the sun is able to produce these enourmous temperatures. In this project we will simulate the hottest parts of the solar corona to understand the heating mechanism and the to reveal if the same heating mechanism is at play everywhere or just in places where the solar magnetic field is the strongest.

FIREbox: Simulated Galaxies with Resolved Structure and Multiphase Interstellar Medium in Cosmological Volumes

Project Title: FIREbox: Simulated Galaxies with Resolved Structure and Multiphase Interstellar Medium in Cosmological Volumes
Project Leader: Robert Feldmann
Resource Awarded: 29.9 million core hours on MareNostrum

Details

Team Members:
Onur Catmacabak, University of Zurich, SWITZERLANDAlexander Hobbs, University of Zurich, SWITZERLANDLichen Liang, University of Zurich, SWITZERLANDPhil Hopkins, Caltech, UNITED STATESDaniel Angles-Alcazar, Northwestern University, UNITED STATESClaude-Andre Faucher-Giguere, Northwestern University, UNITED STATES

Abstract:
Observational surveys that cover large fractions of the sky and sample millions of galaxies have been instrumental in quantifying the statistics of galaxy properties and in measuring their evolution across cosmic time. Hydrodynamic simulations of large cosmological volumes have been paramount in connecting these results to galaxy formation theory. However, these cosmological simulations model core physical processes (such as star formation, stellar feedback, and supermassive black hole growth) using finely tuned sub-grid prescriptions, reducing their predictive power and compromising their ability to adequately capture the internal structure of galaxies. Zoom-in simulations have begun to implement much more detailed models of the interstellar medium (ISM), but those are available only for small numbers of galaxies limiting their statistical power. We propose FIREbox, a new set of simulations designed to bridge the gap between current large-volume and zoom-in simulations. Their high numerical resolution (comparable to today’s best zoom-ins) and the relatively large volume they cover (similar to some cosmological box simulations) will make them ideally suited to address a large variety of science questions. These range from the evolution of galaxies in Local Group like environments, over the role of inflows and outflows in regulating the gas content of galaxies, to predictions of the faint end slope of the luminosity function at high redshift.

UNITSIMS: Universe N-body Simulations for the Investigation of Theoretical Models of Large Scale Structure.

Project Title: UNITSIMS: Universe N-body Simulations for the Investigation of Theoretical Models of Large Scale Structure
Project Leader: Gustavo Yepes
Resource Awarded: 25 million core hours on MareNostrum

Details

Team Members:
Yu Liang, Tsinghua University, CHINACheng Zhao, Tsinghua University, CHINAChia-Hsun Chuang, Leibniz-Institute for Astrophysics Potsdam (AIP), GERMANYFrancisco Shu Kitaura, Instituto de Astrofísica de Canarias, SPAINMarco Pellejero-Ibañez, Instituto de Astrofísica de Canarias, SPAINAlexander Knebe, Universidad Autonoma de Madrid, SPAINSergio Rodriguez Torres, Universidad Autonoma de Madrid, SPAINYu Feng, University of California at Berkeley, UNITED STATES

Abstract:
The statistical analysis of the large scale structures of the galaxy distribution is one of the most important tools in present day cosmology to understand the nature of dark matter and dark energy in the Universe. To this end, a considerable observational effort is put forward to map the 3D galaxy distribution in the Universe at unprecedented scales by means of large collaborations that will measure the distance to millions of galaxies using both photometric and spectroscopic surveys (e.g. BOSS, eBOSS, DESI, J-PAS, Euclid, LSST, WFIRST-AFTA, 4MOST, etc). To robustly extract the cosmological constraints from these surveys, we need to be sure that the potential systematic error from the theoretical models is well below the statistical uncertainties caused by cosmic variance. To reach such a goal, we would need to run simulations with much larger effective volume than those probed by the surveys and with enough mass resolution to be able to resolve the dark matter halos hosting the typical galaxies detected in those surveys. However, the computational resources needed to account for this task are at the edge of the current (petaflop) computational power. In this proposal we intend to explore an alternative way of reaching the same level of required accuracy but with far less computational resources. We propose to run a series of pair simulations with fixed amplitude and opposite phases to effectively remove the cosmic variance. This technique has proved to be very effective in doing so. For instance, a single pair of such simulations would be equivalent to having an ensemble of 50 different random realizations of the same computational box. We plan to run a maximum of 15 pair simulations of a large volume of 1/h Gpc size with enough resolution to resolve the halos hosting the Emission Line Galaxies and the H-alpha emitting galaxies, which will be the type of galaxies surveyed by the upcoming DESI and Euclid spectroscopic surveys. In particular, we will focus on the Euclid survey which will probe a larger volume. With 15 pairs of 1/h Gpc boxes and almost 4000^3 particles each, we will get errors in the statistical measurements of dark matter equivalent to having more than 10 times the volume of the Universe sampled by Euclid. We will generate halo catalogues and merger trees, using the publicly available ROCKSTAR halo finder, together with density and velocity fields on a mesh for later construction of light cone distribution of galaxies and weak lensing maps. We will use state of the art techniques to produce thousands of catalogues, extracting the statistics of the UNITSIMS. All these data will be made publicly available through databases and web portals for the general use of the astrophysical community.

Kinetic modelling of cometary-induced magnetospheres

Project Title: Kinetic modelling of cometary-induced magnetospheres
Project Leader: Pierre Henri
Resource Awarded: 15 million core hours on Curie

Details

Team Members:
Vyacheslav Olshevsky, KU Leuven, BELGIUMNicolas Gilet, Centre national de la recherche scientifique (CNRS), FRANCERajkumar Hajra, Centre national de la recherche scientifique (CNRS), FRANCEJerome More, Centre national de la recherche scientifique (CNRS), FRANCEAndrey Divin, St. Petersburg State University, RUSSIAN FEDERATIONStefano Markidis, KTH Royal Institute of Technology, SWEDENAnders Eriksson, Swedish Institute of Space Physics, Uppsala, SWEDENJan Deca, University of Colorado Boulder, UNITED STATES

Abstract:
After more than 2 years of operations in the vicinity of comet 67P/Churyumov-Gerasimenko, ESA’s highly successful exploratory space mission Rosetta came to its conclusion at the end of 2016. Rosetta has provided the space community the largest amount of data ever collected around a comet. While some observations were expected, many others were a complete surprise and have risen more questions than they have brought answers. Many of these unexpected findings are directly related to the structure and dynamics of the comet’s ionized environment and to its interaction with the surrounding supersonic plasma flow in which it is embedded: the solar wind. They have redefined unknowns about the cometary plasma environment and interaction that were at the very core of the Rosetta mission science objectives: How does the cometary environment evolve as it approaches the Sun? What are the physical processes steering the interaction? What are the key processes that define the shape and plasma flows in the near-cometary environment? Our project “Kinetic modelling of cometary-induced magnetospheres” aims at answering these questions using fully kinetic simulations of the interaction between the solar wind and a comet. With our code, iPIC3D, we plan to explore both the collisionless and the weakly collisional regimes (a strongly collisional regime would not require a fully kinetic approach). Next to the global dynamics of the solar wind and cometary plasma species, we will focus in particular on the physics that controls the electron dynamics. Our fundamental model can also provide robust feedback to hybrid or fluid simulations that model the electron dynamics through a Generalized Ohm’s law. Our project is timely and will strongly enhance the science return of the Rosetta mission by adding the indispensable physical insight to the observations gathered during more than 2 years. Finally, as the solar wind-comet plasma interaction is, in essence, that of a mass-loading plasma, our work will be beneficial as well for the broader space physics community, e.g., to help decipher gas release processes in space. It will pave the way to prepare future European space missions dedicated to the exploration of small bodies in the solar system.

GAFFER – Galaxy Formation For the Epoch of Reionization

Project Title: GAFFER – Galaxy Formation For the Epoch of Reionization
Project Leader: Andrei Mesinger
Resource Awarded: 30 million core hours on Curie

Details

Team Members:
Gillet Nicolas, Scuola Normale Superiore, ITALY

Abstract:
The birth of the first luminous objects and the eventual reionization of our Universe remain at the very forefront of modern cosmology research. This poorly-understood epoch of reionization (EoR) is set to experience a Big Data revolution, driven mainly by upcoming observations of the cosmic 21-cm signal. Next-generation interferometers such as HERA and the SKA will deliver a 3D map of the first billion years of our Universe! The patterns of this map encode the properties of the undiscovered first galaxies and intergalactic medium (IGM) structures. But how can we interpret this wealth of data? The astrophysics of collapsed objects is intractable from first principles. The EoR involves a huge range of scales, with small-scale physics (involving stars, galaxies, gas clumps) influencing large-scale, highly non-Gaussian cosmic signals. As a result, all large-scale EoR simulations must include missing physics via sub-grid prescriptions. Unfortunately, it is difficult to provide reliable sub-grid models for the emission and absorption of ionizing photons. Typically, one uses a handful of small-scale simulations for calibration. However, these have many available ‘tuning knobs’, with different parameter combinations fitting the scant available data equally well. Thus, sub-grid models calibrated only on one or a few small-scale simulations are effectively educated guesses and do not quantify the range of astrophysical uncertainties. In this project, we propose to do a numerical parameter study by generating a very large suite (several hundred) of medium scale and resolution hydrodynamic simulations. We will efficiently sample the conditional probabilities of emission and absorption of ionizing photons, quantifying their dependence on general properties of host dark matter halos and the local environment. These will then be adapted as sub-grid models into our efficient, large-scale (>Gpc) EoR simulations, exploring a large range of physically-motivated astrophysical uncertainties. As part of the official SKA data analysis, these EoR simulations will allow Europe to profit from its large investment in this next-generation interferometer by quantifying what we can learn from the cosmic 21-cm signal.

top

Share: Share on LinkedInTweet about this on TwitterShare on FacebookShare on Google+Email this to someone