PRACE Preparatory Access – 26rd cut-off evaluation in September 2016

Find below the results of the 26th cut-off evaluation of September 2016 for the PRACE Preparatory Access.

Projects from the following research areas:

 

Large scale configuration interaction studies of nuclear structure and level density

Project Name: Large scale configuration interaction studies of nuclear structure and level density
Project leader: Prof. Chong Qi
Research field: Fundamental Physics
Resource awarded: 50000 core hours on Marconi – Broadwell, 100000 core hours on Hazel Hen
Description

Large-scale shell-model configuration interaction calculations will be carried out to study the structure and decay properties of exotic atomic nuclei produced by leading radioactive beam facilities like FAIR in Germany, GANIL in France, FRIB at US and RIKEN and Tokyo. Predictions will be made on terra incognita nuclei that could be important for nuclear astrophysical processes and for the probe of the underlying nucleon-nucleon interaction.

top

Relativistic effects at non-standard temperature and pressure

Project Name: Relativistic effects at non-standard temperature and pressure
Project leader: Dr. Krista Steenbergen
Research field: Chemical Sciences and Materials
Resource awarded: 50000 core hours on SuperMUC, 100000 core hours on Piz Daint
Description

Mercury’s thermodynamic properties have long been of interest to experimental and theoretical researchers, as the only elemental metal that exists in the liquid state under standard conditions (room temperature, standard pressure). Recent simulations have shown that mercury is a liquid at room temperature due to relativistic effects. The interplay of relativistic effects and pressure have never been theoretically explored. We will complete a full set of simulations melting bulk mercury (alpha, beta phases) at a series of increasing pressures. Comparing the results from relativistic and non-relativistic models, we will explore how relativistic effects alter or dictate the material properties of this interesting metal under pressure.

top

Porting and scalability test of a multiscale online chemical-aerosol weather prediction system: the NMMB/BSC-CTM system

Project Name: Porting and scalability test of a multiscale online chemical-aerosol weather prediction system: the NMMB/BSC-CTM system
Project leader: Dr. Sara Basart
Research field: Earth System Sciences
Resource awarded: 50000 core hours on Marconi – Broadwell, 50000 core hours on MareNostrum, 50000 core hours on SuperMUC
Description

Some of the today’s most significant environmental concerns are related to the composition of the atmosphere. The increasing concentration of the greenhouse gases and the cooling effect of aerosols are prominent drivers of a changing climate, but the extent of their impact is still uncertain. At the Earth’s surface, aerosols, ozone and other reactive gases such as nitrogen dioxide determine the quality of the air around us, affecting human health and life expectancy, the health of ecosystems and the fabric of the built environment. Ozone distributions in the stratosphere influence the amount of ultraviolet radiation reaching the surface. Dust, sand, smoke and volcanic aerosols affect the safe operation of transport systems and the availability of power from solar generation, the formation of clouds and rainfall, and the remote sensing. To address these environmental concerns, there is the need to push further the boundaries of what numerical simulations can provide. The Earth Sciences Department of the Barcelona Supercomputing Center (ES-BSC) is currently developing a new fully on-line coupled chemical weather prediction system for research applications and experimental forecasts at sub-synoptic and mesoscale resolutions on global and regional domains. The NMMB/BSC-CTM system is based on the Non-hydrostatic Multiscale Model on the B-Grid (NMMB), developed at US National Centers for Environmental Prediction (NCEP). The main feature of NMMB/BSC-CTM is its online coupling of chemistry and meteorology. The new chemical system component solves the gas-phase tropospheric chemistry and the life cycle of the mineral dust, sea salt, black carbon, organic carbon, and sulphate. The direct effect of non-climatic aerosols on the radiative budget is already implemented and allows to study further mesoscale processes associated with air pollution and its interactions with meteorology, both at high resolution and on a global scale. Furthermore, the model also includes and ensemble-based data assimilation system for aerosols using data from satellites and ground-based observations. In data assimilation mode, the model is capable of ingesting observations to estimate improved model initial conditions, or analyses, as well as to generate retrospectively consistent analysis datasets over a long period (reanalyses). Reanalyses are key to monitor past state of the atmosphere. In the data assimilation mode multiple (independent) NMMB/BSC-CTM simulations are run to account for model uncertainty in the calculation of the data assimilation corrections. Such ensemble runs make the optimization of the model efficiency even more crucial. NMMB/BSC-CTM currently benefits of strong collaboration ties between ES-BSC, NCEP, the Technical University of Catalonia, the University of California Irvine and the NASA Goddard Institute for Space Studies. ES-BSC considers porting and evaluating NMMB/BSC-CTM on different high-performance computational platforms a top priority, before making the code generally available to the scientific community. Initial work on the CURIE system (PRACE Preparatory Access projects 2010PA0419 and 2010PA0627) with an early version of the model pointed out some issues to be addressed before performing scalability tests. Initial results show scalability up to 8000 cores. We aim to finalize the porting and investigate scalability and performance of the NMMB/BSC-CTM model on the MareNostrum, Marconi, and SuperMuc supercomputers. We target a high-resolution configuration reaching the foreseen capabilities of next year’s numerical weather prediction systems.

top

Activation mechanism of the ß2-adrenergic receptor

Project Name: Activation mechanism of the ß2-adrenergic receptor
Project leader: Prof. Ilpo Vattulainen
Research field: Biochemistry, Bioinformatics and Life sciences
Resource awarded: 100000 core hours on Hazel Hen, 100000 core hours on Piz Daint
Description

G protein-coupled receptors (GPCRs) are versatile signalling proteins that mediate diverse cellular responses. With over 800 members, GPCRs constitute the largest family of integral membrane proteins in human genome and represent roughly half of all drug targets in modern medicine. The human ß2-adrenergic receptor (ß2AR) is one of the best-characterized GPCRs. It is expressed in pulmonary and cardiac myocyte tissues and is a therapeutic target for asthma and heart failure. The functional diversity of ß2AR is associated with its structural dynamics. Recently found structures of ß2AR (published in Protein Data Bank) in the inactive and active states have provided valuable insights into the structure-function relationship of ß2AR. However, the dynamics and molecular details of the ß2AR activation process are still missing. Here we aim to fill this gap by using an extensive all-atom Molecular Dynamics (MD) simulations using GROMACS software. This will be further coupled with the Hamiltonian replica exchange MD to enhance the sampling of the simulations. System under investigation will be composed of the ß2AR embedded in the lipid bilayer of different lipid compositions. In order to provide a surface large enough to accommodate protein conformational changes and lipid types in a statistically significant quantity, large membranes must be used. A system of this size will require sufficient testing for ensure that the use of computational resources is done as efficiently as possible. In this project we will fine-tune and benchmark the performance of GROMACS for the solvated membrane-protein system and evaluate the scaling behaviour of enhanced sampling techniques.

top

Insights into the mechanism of oxygen reduction reaction on Pt over molybdenum carbide support

Project Name: Insights into the mechanism of oxygen reduction reaction on Pt over molybdenum carbide support
Project leader: Prof. Ilya Grinberg
Research field: Chemical Sciences and Materials
Resource awarded: 50000 core hours on MareNostrum, 100000 core hours on Hazel Hen, 100000 core hours on Judueen, 100000 core hours on SuperMUC
Description

Platinum nanorafts on molybdenum carbide support have been found to be an excellent electrocatalyst for the oxygen reduction reaction (ORR) and show enhanced performance in polymer electrolyte membrane fuel cells (PEMFCs). We use first-principles pseudopotential plane-wave calculations to investigate the formation of Pt nano rafts and their ORR catalytic activity on Mo2C. We will investigate the adsorption energies of Pt on Mo2C and of oxygen on Pt supported by Mo2C. Comparison of different O/Pt/Mo2C systems will reveal the chemical and structural origins of the high performance and stability of Pt nano rafts on Mo2C.

top

The Role of Black Hole Feedback in the Evolution of Massive Galaxies

Project Name: The Role of Black Hole Feedback in the Evolution of Massive Galaxies
Project leader: Prof. Robert Feldmann
Research field: Universe Sciences
Resource awarded: 50000 core hours on MareNostrum, 100000 core hours on Hazel Hen
Description

Galaxies in the nearby Universe typically belong to one of two distinct populations. Those that form stars actively out of interstellar gas and those that evolve passively after having shut down star formation a long time ago. Colors, gas fractions, masses, and structural properties also differ substantially between the two classes. However, although much progress has been made in analyzing these populations by observing increasingly larger and deeper samples, there is still no consensus on the physical processes that explain these observations. In fact, a large range of physically plausible, but not mutually consistent, scenarios have been suggested, many including feedback from central, supermassive black holes. We apply for a preparatory project to prepare a future PRACE project proposal targeting the origin of star forming and passive galaxies in the nearby Universe with ultra high-resolution, cosmological simulations. Our simulation code (GIZMO) and the physics modules (FIRE) are well tested, have been compiled and run on a number of HPC resources, and have been successfully applied to the study of galaxies in the nearby Universe (see http://fire.northwestern.edu for details). However, because of the associated high computational cost, we have not yet used the code to analyze nearby massive galaxies at high resolution. The preparatory project will allow us to test whether our code scales to the number of required cores and will enable us to prepare the necessary performance evaluation for a future PRACE project proposal.

top

Materials Genomic Approach to Characterising Water Behaviour in Metal Organic Frameworks

Project Name: Materials Genomic Approach to Characterising Water Behaviour in Metal Organic Frameworks
Project leader: Dr Peter Boyd
Research field: Chemical Sciences and Materials
Resource awarded: 50000 core hours on Marconi-Broadwell, 50000 core hours on MareNostrum, 5000 core hours on MareNostrum Hybrid Nodes, 100000 core hours on Hazel Hen, 50000 core hours on SuperMUC
Description

We will be simulating adsorption and diffusion of water in metal organic frameworks (MOFs). For this, we’ll be primarily using the Large-scale Atomic/Molecular Massively Parallel Simulator (LAMMPS). In this package, several algorithms will be utilized such as force field-based grand canonical Monte Carlo (GCMC) and molecular dynamics (MD) methods. This preparatory access will enable us to perform the necessary benchmarking required for a tier-0 “Project Access” application.

top

Scalability testing for large scale molecular dynamics simulations of respiratory complex I

Project Name: Scalability testing for large scale molecular dynamics simulations of respiratory complex I
Project leader: Dr Vivek Sharma
Research field: Biochemistry, Bioinformatics and Life sciences
Resource awarded: 50000 core hours on Marconi-Broadwell, 100000 core hours on Juqueen,
Description

Respiratory complex I is a key enzyme in the electron transport chains of mitochondria and bacteria. It couples the reduction of quinone to proton pumping across the biological membrane, and contributes ca. 40% of the total proton motive force required to drive ATP synthesis. Various neurodegenrative and mitochondrial disorders are known to be associated with complex I malfunction, which makes it one of the most critical enzymes in mitochondrial respiration. In this project, we aim to perform scalability tests for large scale molecular dynamics (MD) simulations of respiratory complex I from thermophilic bacterium (Thermus thermophilus). We will be using NAMD program for fully atomistic classical MD simulations on a state-of-the-art model system, and will perform scalability testing. This will allow us to generate scalability plots for our future large scale PRACE projects.

top

Scalability tests for FHI-aims electronic structure theory code

Project Name: Scalability tests for FHI-aims electronic structure theory code
Project leader: Dr Ville Havu
Research field: Chemical Sciences and Materials
Resource awarded: 50000 core hours on Curie,
Description

This project investigates the parallel scalability of the FHI-aims electronic structure theory code. FHI-aims is a full-potential all-electron electronic structure theory code based on numeric atom-centered orbitals. FHI-aims has a comprehensive set of features and can treat periodic and non-periodic systems on equal footing. Important ingredients for an effective parallel perfomance of FHI-aims are distributed handling of all grid-based operations (integration and update of the electron density), scalable eigensolver and distributed solver for the electrostatic potential. Further, the most expensive operations scale O(N) with respect to the size of the physical system (limiting factor being the O(N^3) cost for the eigensovler). FHI-aims has demonstrated good scalability on several platforms (Cray XC30/40, BG, IB-clusters, see, e.g., http://iopscience.iop.org/article/10.1088/0953-8984/26/21/213201) but there are no explicit scalability results on PRACE systems.

top

Characterization of extreme sea states using exact calculation of four wave nonlinear interactions

Project Name: Characterization of extreme sea states using exact calculation of four wave nonlinear interactions
Project leader: Dr. Sonia Ponce de Leon Alvarez
Research field: Earth System Sciences
Resource awarded: 50000 core hours on MareNostrum,
Description

This project aims at understanding the scalability properties of the WAVEWATCH III spectral wave model when using the Webb-Resio-Tracy (WRT) algorithm for the exact computation of the four-wave resonant non-linear interaction source term in realistic oceanographic applications. The use of approximate methods for the computation of the four-wave nonlinear interactions results in incorrect wave spectra that underestimate the spectra peak intensity and the instability bands around it. This region of the spectrum is very important when it comes to the prediction of oceanic rogue waves. This project intends to improve our current scientific knowledge into extreme sea states and their interaction with strong currents. The scalability study will focus in two different scenarios: The North Sea and the South Atlantic Agulhas current. The following approaches are to be adopted: a) Use of the exact WRT method (Resio and Perrie, 1991; Van Vledder, 2006) for the computation of the four-wave nonlinear source term b) Increase the frequency and direction resolution of the spectral grid for better representation of the spectral peak (Van Vledder and Hashimoto, 2013), The scalability data gathered in this preparatory step will allow a better understanding of the resources necessary to have accurate high resolution hindcasts of rogue sea states. The final objectives of this project are the better characterization of rogue waves and extreme sea states, advancing considerably the current knowledge in wave physics and an improvement of the capability of existing forecasting systems and the modeling of the interaction between waves and currents and its effects on the creation of rogue waves.

top

Molecular dynamics simulations of photoactive molecules in optical cavities: the effect of strong light-matter coupling on chemical reactivity

Project Name: Molecular dynamics simulations of photoactive molecules in optical cavities: the effect of strong light-matter coupling on chemical reactivity
Project leader: Dr. Gerrit Groenhof
Research field: Fundamental Physics
Resource awarded: 50000 core hours on Curie
Description

Strong coupling between confined light in optical cavities or plasmons, could be exploited to control chemical reactions and may thus provide an alternative to chemical catalyst or coherent control strategies. A theoretical model to describe and predict the dynamics of molecules coupled inside a cavity is currently missing, but such model will be essential to design cavities, or plasmonic materials with which photochemical reactions can be steered. To fill this gap, we have recently developed a new method for simulating both single molecule and ensembles of molecules in optical cavities. The new method is based on approximate QED in combination with excited state QM/MM simulations. The accuracy of the latter was recently demonstrated (Science 352 (2016) 725-729) and preliminary simulations on small numbers of molecules suggest that we can also predict the dynamics of photoactive molecules inside cavities. However, running the QED simulations of a realistically large number of molecules inside a optical cavity requires very large computational resources that are only available through PRACE. In this project we will run simulations of various molecules in optical cavities and compare their dynamics with recent experiments to validate our code and provide the hitherto unknown physical insights into the dynamics of these coupled molecules.

top

ATHERMAD: Ab-initio THERmal transport in layered MAterials with Defects.

Project Name: ATHERMAD: Ab-initio THERmal transport in layered MAterials with Defects.
Project leader: Dr. Ankita Katre
Research field: Chemical Sciences and Materials
Resource awarded: 50000 core hours on MareNostrum
Description

This preparatory access application aims at testing the scalability of the transport software developed in our group to study the effect of defects on thermal conductivities of the technologically relevant class of 2D layered materials from first principles. Our transport software, a part of H2020 project ALMA (www.almabte.eu), is based on the solutions of the Boltzmann transport equation for phonons where the frequencies of phonons and their scattering rates are determined using the inputs from density functional theory (DFT) calculations. Thermal conductivity is a crucial property for many technological devices, however with the increasing complexity of materials by doping or nanostructuring, the calculation of thermal conductivity in such materials is a non-trivial problem. In last few years, other ab-initio thermal transport software packages have been developed, however success has been achieved only for the cases of defect-free crystalline materials so far, whereas underlying materials in real devices are much more complex due to the presence of different defects. Considering the fact that the defects can have a huge impact on altering the thermal conductivity of a material and their effect cannot be captured with empirical models used commonly, a detailed approach based on Green’s functions in combination with DFT calculations has been implemented in our ALMA software. This approach has already been shown to capture correct experimental thermal conductivity of defective cubic compounds (Katre et al., J. Mat. Chem. A. 4, 15940, 2016 and Katcho et al., Phys. Rev. B., 90, 094117, 2014) which points towards the strong predictive power of this approach. Thus the next step is to test our method for more complex structures. This preparatory access, if granted, would help us to test the scalability of our software for 2D layered structures with defects. The DFT code (VASP: Vienna ab-initio Simulation Package) which is used in combination with our software for the thermal conductivity calculation, is already known for its excellent scalability. Thus this scalability test would provide us with statistics about our software, which will be helpful before applying for its deployment on massive parallel environments as a future goal.

top

Characterization of hypersonic flows in high non-equilibrium conditions during the entry of exploratory spacecrafts in planetary atmosphere using SPARK code.

Project Name: Characterization of hypersonic flows in high non-equilibrium conditions during the entry of exploratory spacecrafts in planetary atmosphere using SPARK code.
Project leader: Prof. Mario Lino da Silva
Research field: Engineering
Resource awarded: 50000 core hours on MareNostrum, 100000 core hours on Hazel Hen, 50000 core hours on superMuC
Description

The present project aims at assessing and improving the scalability of a recently developed code, for simulations of the aerothermodynamics in non-equilibrium hypersonic re-entry flows. An accurate prediction of the gas state conditions surrounding an inter-planetary exploratory spacecraft is still extremely difficult due to the high coupling between chemistry, radiation and the near-field flow generated during the entry of these exploratory spacecrafts in planetary atmosphere. This Software Package for Aerodynamics, Radiation and Kinetics (SPARK code) is a multidimensional finite volume method discretization solver for Navier-Stokes and reactive flow governing equations in structured meshes. SPARK is written in Fortran 03/08 and it explores the newly supported object-oriented feature which enables the encapsulation of different physical models, numerical methods, mesh-related operations and interface communications by means of derived-types and type-bound procedures. The code architecture has been thought to clearly separate the physical models from the numerical methods owing to an elegant and flexible way to dynamically assess/develop a given physical model and/or numerical method. This has the additional advantage of increasing code locality, which prevents unintentional data corruption and a efficient code parallelization. Two core physical models are implemented in SPARK for simulations of gas thermodynamics in non-equilibrium conditions: the multi-temperature model and the state-to-state model. In this preparatory access to PRACE infrastructure, a serial of short simulations will be performed in order to access the code scalability when considering each of these physical models. For this, a recently developed Fortran Coarray feature for parallel computation was implemented in the code and will be tested in different compilers during the project . Code scalability analysis will be performed by means of domain decomposition. The mesh is divided into multiple blocks which allows for multicore parallel computation. Preliminary results obtained with the SPARK code are very promising since they represent the first multi-dimensional computation of an re-entry capsule using a 5 species air model based on vibrational state-specific kinetics.

top

Structure and Phase Transitions on the Surface of Ice

Project Name: Structure and Phase Transitions on the Surface of Ice
Project leader: Prof. Luis G. MacDowell
Research field: Chemical Sciences and Materials
Resource awarded: 5000 core hours on MareNostrum
Description

The interface of ice in air plays an extremely important role in a wide variety of extremely relevant phenomena. In the atmosphere, its properties are crucial for the assesment of the radiation budget in cirrus clouds . Also, it has an important role in the electrification of thunderclouds, or the scavenging of atmospheric trace gases. In geology the surface of ice has a crucial impact on the flow of glaciers, as well as the heaving of frozen ground, or the breakdown of rocks and concrete. In technology, it has impact in the damage of aircfats due to the freezing of moisture and rain drops. Whereas the important role of the ice surface is widely recognized our understanding of its structural properties remains very limitted. For example, it is well known since the pioneering studies of Nakaya that the habit of hexagonal ice crystal prisms grown from bulk vapor in the atmosphere change from plates, to columns, to plates and yet back to columns as temperature is cooled down below the triple point. It is clear that the habit changes observed experimentally must be related to sudden crossover in the growth rates of basal and prism facets, but we completely ignore why such crossovers take place and how they are related to the structure of the ice surface. In this project we plan to characterize the structure of the ice surface in a wide range of temperatures, using a surface fluctuation analysis that we have implemented succesfully in recent work (Benet et al. Phys.Rev.Lett. 117, 096101, 2916). By studying the behavior of surface fluctuations at low wave-vectors, we have been able to show the presence of a roughening transition on the prismatic face of ice very close to the triple point. Such transiton enhances the growth rate of prismatic faces, and is consistent with the appearence of plate like prisms close to the triple point as observed in nature. In this work we will pursue a comprehensive study of the surface properties for different temperatures and facets using our well tested methodology. Our aim is to resolve the long-standing problem of crystal-habit cross-over with temperature, with crucial impact on atmospheric sciences, but also, on our understanding of ice-surface physics. However, such study cannot be undertaken without the use of supercomputer facilities required to study systems of about two hundred thousand molecules over hundreds of nanoseconds at different ambient conditions typical of cirrus clouds.

top

Modeling of ribonucleotides elementary blocks under prebiotic conditions: Insights from ab initio molecular dynamics simulations

Project Name: Modeling of ribonucleotides elementary blocks under prebiotic conditions: Insights from ab initio molecular dynamics simulations
Project leader: Prof. Antonino Marco Saitta
Research field: Chemical Sciences and Materials
Resource awarded: 50000 core hours on Marenostrum, 100000 core hours on Juqueen, 50000 core hours on SuperMUC,
Description

Among the several hypothesis about origins of life, the “RNA world” is one of the most accepted by scientific community nowadays. This hypothesis states that ribonucleic acid molecules have been precursors to current living forms on Earth, and it is mainly supported by RNA capability of self-replication, catalysis, self-regulation, and so forth. One of the most relevant questions about the “RNA world”, is the spontaneous synthesis of RNA monomers, namely ribonucleotides, in the prebiotic earth. For this reason, our project is aimed at investigating the nucleotide breaking down mechanism and its reverse recombination (from smaller elementary entities) by means of ab initio molecular dynamics in the Born-Oppenheimer scheme (BOMD), in presence of explicit water molecules, and coupled with accelerated sampling methods such as metadynamics and umbrella sampling, in order to reconstruct the free energy profiles for the transitions of interest. During the last semester, we were able to tackle a first test case under hydrothermal prebiotic conditions, following the same framework described above. We explored the chemical reactivity of the uracil (UMP) and adenosine monophosphates (AMP) in presence of phosphorus species, observing different mechanisms that leads to the formation of a metabolite that plays a crucial role in the current de novo biosynthesis of nucleotides. Now our next step is focused to test both ambient and hydrothermal pressure/temperature conditions in a wider range of substrates of biological/prebiotic importance with the aim to model a realistic environment and obtain thermodynamic and kinetic insights about nucleotide synthesis, as well as provide clue features that bridge the prebiotic context with the current living systems. In particular, our simulations will include a methodology developed in our group, which consists in a new class of reaction coordinates that accurately track the changes of the chemical-bond network along a reaction and simultaneously take into account the active participation of the solvent. These coordinates, in combination with free energy methods, have unveiled non-trivial mechanisms (showing the relevant participation of the water molecules) for formamide (an important substrate for prebiotic chemistry), UMP and AMP molecules starting from a minimal knowledge about the system, rendering this approach a very powerful tool for this project. Additionally, a joint collaboration with experimental groups (lead by Marie-Christine Maurel at MNHN-Paris and François Guyot and Jean-François Lambert at UPMC-Paris) will test the nucleotide degradation pathways by spectroscopic methods including multinuclear solution NMR, vibrational spectroscopies, and ESI mass spectrometry, providing an important and complementary information to couple with the theoretical approaches.

top

Atomistic-level investigation of odor perception by olfactory receptor dynamics

Project Name: Atomistic-level investigation of odor perception by olfactory receptor dynamics
Project leader: Prof. Jerome Golebiowski
Research field: Biochemistry, Bioinformatics and Life sciences
Resource awarded: 50000 core hours on Marenostrum, 100000 core hours on Juqueen
Description

Our perception of an odor is the result of a complex cascade of events triggered by the interaction of odor molecules (odorants) with olfactory receptors (ORs). Although the odor of an odorant is fully encoded within its molecular structure, understanding structure-odor relationships requires decoding the combinatorial activation code of multiple ORs by the odorant. This represent a fundamental step and major challenge towards our comprehension of odor perception in the brain. ORs are the largest group of G protein-coupled transmembrane protein receptors (GPCRs)––the most heavily investigated drug targets in the pharmaceutical industry. The OR gene family is the largest in the mammalian genome, comprising over 450 genes in the human genome and more than 1,000 in mice [Yong et al. Hum Mol Genet 2002, 11: 535]. Human ORs can detect at least 1 trillion different odors [Bushdid et al. Science 2014, 343: 1370]. Although mainly known for their function in smell, a rising number of ORs reveal their roles in non-olfactory chemosensory pathways, emerging as potential drug targets for various disorders. Therefore, the large OR repertoires contain rich information on GPCR structure-function relations and may provide valuable data for pharmaceutical research. To this aim, two fundamental questions are to be answered: I) how do ORs differentiate agonists (e.g. odorants), antagonists and inverse agonists (e.g. odor blockers) and non-ligands; II) how do the ligands alter OR activation at molecular level. A major obstacle here is the paucity of structural information on ORs. Recent breakthroughs in GPCR crystallography have made homology modeling of OR structures possible, since the overall 3D structure, several key residues/motifs and the general activation mechanism are conserved across diverse GPCR families. In our recent work [de March et al. J Am Chem Soc 2015, 137: 8611], homology modeling followed by all-atom molecular dynamics (MD) simulations successfully reproduced mutagenesis data of OR activities and revealed molecular features previously inaccessible to experimental techniques. This computational approach is readily applicable to a wide range of ORs. The in silico models obtained will be assessed by in vitro / in vivo assays. We will gain atomistic-level insights into OR-ligand interactions, as well as agonist-dependent and agonist-independent activation of ORs. The study will also build a computational platform for screening and design of novel odorants, odor blockers and odor enhancers. The goal in a long term is to establish a ‘computational nose’ that can decode odor perception using merely the molecular structure of an odorant as input. The project outcome will have broad applications in pharmaceutical research and the flavor and fragrant industry.

top

Type B: Code development and optimization by the applicant (without PRACE support) (9)

High Performance Computing algorithms for the RefficientLib software.

Project Name: High Performance Computing algorithms for the RefficientLib software.
Project leader: Prof. Joan Baiges
Research field: Mathematics and Computer Sciences
Resource awarded: 100000 core hours on MareNostrum,
Description

This project aims to tune up the RefficientLib algorithm for adaptive mesh refinement in computational physics meshes in a distributed memory parallel setting. The proposed method has been developed for nodally based parallel domain partitions where the nodes of the mesh belong to a single processor, whereas the elements can belong to multiple processors. Some of the main features of the algorithm are the capability to handle multiple type of elements in two and three dimensions (triangular, quadrilateral, tetrahedral and hexahedral), the small amount of required memory per processor and the parallel scalability up to thousands of processors. The algorithm is also capable of dealing with on-balanced hierarchical refinement, where multi refinement level jumps are possible between neighbor elements. An algorithm for dealing with load-rebalancing has been also developed, which allows to move the hierarchical data structure between processors so that load unbalancing is kept below an acceptable level at all times during the simulation. A particular feature of the algorithm is that arbitrary renumbering algorithms can be used in the load rebalancing step, including both graph partitioning and space filling renumbering algorithms. The load-rebalanced adaptive mesh refinement algorithm is packed in the Fortran 2003 object oriented library RefficientLib, whose interface calls which allow it to be used from any computational physics code.

top

Scaling MESONH to next gen Intel KNL & CRAY/Nvidia-Telsa

Project Name: Scaling MESONH to next gen Intel KNL & CRAY/Nvidia-Telsa
Project leader: Dr ESCOBAR MUNOZ Juan
Research field: Earth System Sciences
Resource awarded: 10000 core hours on Piz Daint
Description

MesoNH is the non-hydrostatic mesoscale atmospheric model of the French research community. It has been jointly developed by the Laboratoire d’Arologie (UMR 5560 UPS/CNRS) and by CNRM-GAME (URA 1357 CNRS/Mto-France). MesoNH is now running in production mode in TIER1 computer upto 32K cores . The goal of this project is to prepare MesoNH to the next gen architecture, CRAY & NVIDIA/GPU & INTEL/KNL to gain 1 order of magnitude in the scalability of the code in production runs . ->Development will concentrate on the first porting of MESONH on the GPU with the CRAY+OPENACC tool-kit on multi-node clusters ( already in progress with PGI toolkit on other POWER+GPU prototype ) . ->The second goal of the project is to test the multi-node scalability of MesoNH on Multi rack Intel/KNL computer ( already tested in mono-switch intel/KNL prototype cluster) .

top

Domain Decomposition Method for Parallel Hybrid Linear System Solvers

Project Name: Domain Decomposition Method for Parallel Hybrid Linear System Solvers
Project leader: Prof. Cevdet Aykanat
Research field: Mathematics and Computer Sciences
Resource awarded: 100000 core hours on Marconi-Broadwell, 100000 core hours on MareNostrum, 250000 core hours on Hazel Hen, 250000 core hours on Juqueen, 100000 core hours on SuperMUC
Description

We propose a new domain decomposition method for parallel hybrid linear system solvers. The objective is to increase the performance and the scalability of the parallel sparse linear system solver. Our algorithm utilize the structure of the coefficient matrix and divides it into sub-matrices according to the number of processors in the parallel environment. Each processors solve the small linear system individually. An iterative solver is used to improve the partial solution iteratively. After a number of iteration the solution is acquired.

top

Test/improve the theoretical models of large scale structure of our universe

Project Name: Test/improve the theoretical models of large scale structure of our universe
Project leader: Dr. Chia-Hsun Chuang
Research field: Universe Sciences
Resource awarded: 10000 core hours on MareNostrum, 100000 core hours on SuperMUC
Description

We are developing a theoretical model for the large scale structure of our universe. We would like to verify it by comparing with N-body simulations, i.e. MultiDark simulations and Dark Sky simulations. Since we need to scan a wide parameter space when testing the model, we would need to access supercomputers for this task. In addition, we are going to do the same test for all the models used in recent data analysis. This project is expected to be very useful for understanding our universe from the on-going and future large scale structure surveys

top

Atomic-scale modelling of Elastic and Inelastic transport in semiconductor devices

Project Name: Atomic-scale modelling of Elastic and Inelastic transport in semiconductor devices
Project leader: Prof Kurt Stokbro
Research field: Engineering
Resource awarded: 100000 core hours on MareNostrum, 20000 core hours on MareNostrum Hybrid Nodes, 20000 coure hours on Curie, 100000 core hours on SuperMUC
Description

Atomic-scale modelling of the I-V characteristics of semiconductor devices is becomming important for calibrating the quantum corrections in traditional TCAD simulators used in the semiconductor industry. This requires atomic-scale simulations of systems with upto a million atoms. The objective in this project is to test and improve the parallel scalability of the Atomistix ToolKit (ATK) for approaching million atom systems. ATK is used for nano device simulations by major semiconductor companies and leading academic group. The objective is to perform a full simulation of a nanoscale transistor including electron-phonon interactions. In our previous prepatory access the elastic part of the code was tested and improved. Since then we have improved both the elastic part and added new inelastic simulation methods. We would now like to test the scalability of the new algorithms on different large scale parallel architechtures, to tune the methods and determine their limitations.

top

Tiling Library for Heterogeneous Systems

Project Name: Tiling Library for Heterogeneous Systems
Project leader: Asst. Prof. Didem Unat
Research field: Mathematics and Computer Sciences
Resource awarded: 100000 core hours on Piz Daint
Description

In this project we are extending a synchronous tiling library to run asynchronously on heterogeneous systems. We focus on the development of an asynchronous runtime system and its scheduler. Previously we have developed the TiDA library which runs only on the homogeneous multicore systems. We are extending it to exploit GPUs and KNL systems. With the help of the PRACE systems, we will be developing and testing its performance on the emerging parallel architectures on our combustion simulations.

top

Pressure drop simulation for a compressed gas closed system

Project Name: Pressure drop simulation for a compressed gas closed system
Project leader: …
Research field: Engineering
Resource awarded: 100000 core hours on MareNostrum
Description

The objective of the project is to calculate by simulation the pressure drop of a firefighting pressure regulated discharge valve for inert gas agent, to determine the functioning of the outlet pressure regulation system of the valve and to assure the quantity of gas that can be discharge across the valve. For doing that, it is needed to simulate the discharge of a pressurized closed canister through the valve under realistic conditions. The development of simulation techniques for discharging compressed gas across a valve installed over a closed system when the inlet condition changes with time will be a big step forward in the design of these valves as nowadays there is no commercial simulation tool that can manage it.

top

Numerical simulation of accidental fires with a spillage of oil in large buildings

Project Name: Numerical simulation of accidental fires with a spillage of oil in large buildings
Project leader: …
Research field: Engineering
Resource awarded: 100000 core hours on MareNostrum
Description

The objective of the project is to develop a fire engineering analysis (Performance-Based design) of the steel structure buildings that belong to the ITER (International Thermonuclear Experimental Reactor) industrial complex in Cadarache (France), which is devoted .

top

ecPic3D

Project Name: ecPic3D
Project leader: Dr Giovanni Lapenta
Research field: Universe Sciences
Resource awarded: 100000 core hours on Marconi Broadwell, 100000 core hours on MareNostrum, 100000 core hours on SuperMUC
Description

The particle in cell (PIC) technique has been employed for more than sixty years to model the microscopic, complex and highly non-linear kinetic behavior of collisionless plasmas [1, 2]. In the PIC algorithm, macroparticles representative of several real plasma particles are moved under the influence of the self-consistent electromagnetic fields. The latters are calculated on a grid via Maxwell’s equations (ME), where the source terms are obtained by weighting the discrete particles into the grid. The scheme is quite straightforward and has provided very valuable results over the years. However, as it models physics at a microscopic level, it is computationally demanding, especially when the aim is to study realistic astrophysical phenomena, which span over a multitude of time and spatial scales. The semi-implicit version of the PIC algorithm relaxes the stability constraints on the fine spatial and temporal discretization, because the equations of motion for the particles and the ME are solved according to an implicit scheme [3]. However, due to the approximations introduced computing the source term of ME to avoid the solution of a non-linear system, it does not conserve the energy. Ensuring the conservation of energy is fundamental to investigate phenomena that involve energy conversion from particles to fields or vice-versa and where, therefore, the energy balance is particularly important, as for instabilities or magnetic reconnection. That is why the field solver and the particle mover of iPIC3D [4] have been replaced with the scheme described in [5]. This new variation of the code has been called ecPIC3D and it conserves energy down to round-off precision, representing a tremendous breakthrough. As ecPIC3D is based on the implicit moment method (a variant of the semi-implicit scheme), like iPIC3D, the fields are numerically computed from ME through iterative methods. At the current status, the GMRES method has been implemented. However, recent profiling tests have shown how communication among processes in this portion of the code is particularly time demanding, decreasing considerably the performance of ecPIC3D. Therefore, in order to optimize the code, we propose to use the PETSc library [6]. PETSc routines are massively parallel and highly optimized for solving PDEs. They will require minor arrangements in the code at the level of the domain decomposition and they are expected to enhance the performances. Moreover, PETSc will introduce a great flexibility in the code, since the users will be able to choose among several solvers. Furthermore, the profiling has also shown that a considerable portion of time is spent during the computation of the mass matrix, which is needed to solve Ampere equation. In this case, almost no communications among processes are involved, but we believe that vectorizing the calculations could enhance the performances. [1] Birdsall and Langdon, Plasma Physics via Computer Simulation, McGrawHill, New York, 1985. [2] Hockney and Eastwood, Computer Simulation Using Particles, Taylor and Francis, 1988. [3] Brackbill and Forslund, J. Comput. Phys. 46, 271 (1982). [4] Markidis et al., Math. Comput. Simulat 80, 1509 (2010). [5] Lapenta, arXiv preprint arXiv:1602.06326 (2016). [6] http://www.mcs.anl.gov/petsc

top

Type C: Code development with support from experts from PRACE (2)

Optimization of REDItools package for investigating RNA editing in thousands of human deep sequencing experiments.

Project Name: Optimization of REDItools package for investigating RNA editing in thousands of human deep sequencing experiments.
Project leader: Dr Ernesto Picardi
Research field: Biochemistry, Bioinformatics and Life sciences
Resource awarded: 100000 core hours on Markconi-Broadwell,
Description

RNA editing by A-to-I deamination is the prominent co-/post-transcriptional modification in humans. It is carried out by ADAR enzymes and contributes to both transcriptomic and proteomic expansion. RNA editing has pivotal cellular effects and its deregulation has been linked to a variety of human disorders including neurological and neurodegenerative diseases and cancer. Despite its biological relevance, many physiological and functional aspects of RNA editing are yet elusive. Indeed, RNA editing can alter codon identity, create or destroy splice sites and affect base-pairing interactions in secondary and tertiary RNA structures. The advent of high-throughput sequencing technologies has largely improved the computational detection of RNA editing events at genomic scale, revealing its pervasive nature in the human transcriptome. Recently, we have developed the first computational tool, called REDItools, to identify A-to-I events in large transcriptome sequencing experiments (RNAseq). REDItools have been designed to traverse individual genomic positions of multiple read alignments in order to detect genome to transcriptome changes due to RNA editing. Thanks to REDItools, we have profiled, for the first time, RNA editing in 55 human body sites, identifying more than 3 millions of A-to-I events differently distributed across tissues. Despite these findings, many functional aspects of RNA editing are yet unknown and further investigations are needed to elucidate the dynamic regulation of editing sites. To shed light on potential functional roles of RNA editing, it should be studied in very large collections of RNAseq experiments such as those stored in dedicated databases such as dbGAP (including the TCGA repository or the GTEx project). Handling huge amount of RNAseq data, however, is quite computationally intensive. Each RNAseq experiment includes at least 50 million reads and REDItools take on average from 10 to 24 hours of cpu time to provide a complete RNA editing profile. It means that a complete overview of RNA editing in physiological as well as pathological conditions is precluded on standard servers. High performance computing (HPC), employing parallelization on a large number of computing nodes, can drastically reduce computational times. However, REDItools are not optimized to run on HPC systems and, thus, software modifications are strictly needed. Here we propose to optimize and redesign the main REDItools algorithm in order to efficiently scale on large HPC systems. Optimization steps will include the development of a novel algorithm supporting parallelization and the inclusion of dedicated functions in high performance languages (like C and C++) in order to strongly decrease computational times. Our optimization will make REDItools ready for HPC systems and able to handle really huge amount of RNAseq experiments without any precedent.

top

Simulation of the Ocean Cleanup Array

Project Name: Simulation of the Ocean Cleanup Array
Project leader: …
Research field: Engineering
Resource awarded: 100000 core hours on MareNostrum
Description

Every year we produce about 300 million tons of plastic, a portion of which enters and accumulates in the oceans. Due to large offshore current systems called gyres, plastic concentrates in certain offshore areas, of which the Great Pacific Garbage Patch between Hawaii and California is the best-known example. The Ocean Cleanup (www.theoceancleanup.com) is a foundation that develops technologies to extract plastic pollution from the oceans and prevent more plastic debris from entering ocean waters. The main technology is the Ocean Cleanup Array which utilizes long floating barriers to capture and concentrate the plastic such that the system is a passive barrier. Computational Fluid Dynamics (CFD) is being used to study the catch efficiency debris of different sizes and densities, the transport of plastic along the containment boom, and the forces acting on it in order to determine the appropriate shape for their passive barrier concept. A study for the wave and boom influence on particle trajectories has to be done with CFD to investigate the effects of wind-and- wave-induced turbulence on the boom capture efficiently as well as to include the interaction between particles and the dynamic structure in the CFD analyses. This simulation will be carried out by Alya code, which has demonstrated its high potential to face complex problems. The project contains many difficulties that can only be solved by a code like Alya.

top

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