August 2012 - Allocations-admins - mailman3.cels.anl.gov

New LCRC report submitted for MRSA
by support＠lcrc.anl.gov 30 Aug '12

30 Aug '12

Full project report at https://accounts.lcrc.anl.gov/reporting/project/2012/?project=MRSA

1 0

New LCRC report submitted for DFT-defects-UO2
by support＠lcrc.anl.gov 30 Aug '12

30 Aug '12

Full project report at https://accounts.lcrc.anl.gov/reporting/project/2012/?project=DFT-defects-U…

1 0

New LCRC report submitted for MBM
by support＠lcrc.anl.gov 30 Aug '12

30 Aug '12

Full project report at https://accounts.lcrc.anl.gov/reporting/project/2012/?project=MBM

1 0

New LCRC report submitted for intermetallic
by support＠lcrc.anl.gov 29 Aug '12

29 Aug '12

Full project report at https://accounts.lcrc.anl.gov/reporting/project/2012/?project=intermetallic

1 0

New LCRC report submitted for UQCLM
by support＠lcrc.anl.gov 29 Aug '12

29 Aug '12

Full project report at https://accounts.lcrc.anl.gov/reporting/project/2012/?project=UQCLM

1 0

[LCRC Accounts] Yearly Allocation Request from perfopt
by accounts＠lcrc.anl.gov 29 Aug '12

29 Aug '12

Hello, A yearly allocation for the LCRC cluster has been requested with the following updated information: Submitter/PI: Prasanna Balaprakash Project Name: perfopt Division: MCS Project title: Optimization Algorithms for Empirical Performance Associated funding: DOE ASCR Other Systems: Surveyor, ANL: 100000 core hours NERSC: 100000 core hours Science: Our scientific objectives are: i) To advance the state-of-the-art in empirical performance tuning and adaptation of scientific applications on extreme scale architectures and thereby significantly increasing scientific productivity ii) To improve our understanding of large-scale empirical performance tuning iii) To develop novel mathematical optimization algorithms that exploits the characteristics of the search problem in performance tuning Project description: The rapid rate of innovations in computing architectures has widened the gap between the theoretical peak and the achievable performance of scientific codes. Often, scientific application programmers address this issue by manually rewriting the code for the target machine, but this approach is neither scalable nor portable. Empirical performance tuning or automatic performance tuning (in short, autotuning) is a promising approach to address the limitations of manual tuning. This approach consists of identifying relevant code optimization techniques (such as loop unrolling, register tiling, and loop vectorization), assigning a range of parameter values using hardware expertise and application-specific knowledge, and then either enumerating or searching this parameter space to find the best-performing parameter configuration for the given machine. Using this approach, several researchers have achieved considerable success in tuning scientific kernels for b oth serial and multicore processors Our goal is to advance the state-of-the-art in empirical performance tuning and adaptation of scientific applications on extreme scale architectures and thereby significantly increasing scientific productivity. To achieve this goal, we will systematically evaluate the strengths and limitations of optimization algorithms in empirical performance tuning. This study will improve our understanding of large-scale empirical performance tuning and help us to identify open issues that need to be addressed which will require developing novel, flexible optimization approaches. A main aspect of this line of research will be the systematic integration and experimental analysis on multiple problem classes on different architectures. Currently, we are developing novel numerical optimization algorithms that exploits the characteristics of the search problem in performance tuning. We will investigate the effectiveness of the optimization algorithms and will conduct an experimental study of some popular optimization algorithms. Our hypothesis is that appropriately modified local search algorithms can find high-performing code variants in short computation times. This is based on the rationale that the exploration component of global search algorithms is less beneficial in empirical performance-tuning problems, where finding high-performing configurations in short computation time is more important than finding the optimal configuration independent of the computation time required. On Fusion, we conduct an experimental study of some global and local search algorithms on a number of problems from the previously developed SPAPT test suite. We analyze the impact of initial configuration from which a search algorithm starts, input size, and search time constraints on the effectiveness of global and local search algorithms. Another benefit Fusion brings to this research is the ability to test the portability of empirical performance tuning techniques on Intel-based clusters. Eventually, we hope to deploy our techniques for improving the performance of scientific applications running on Fusion. We estimate 100,000 core hours for our experimental study: evaluation of the strengths and limitations of empirical performance tuning = 15,000 core hours; experimental analysis on multiple problem classes on different architectures = 30,000 core hours; empirical analysis for new approaches = 30,000 core hours; tests on parallel performance tuning approaches = 25,000 core hours; We will strongly involve in research dissemination activities such as writing scientific articles and giving seminars at Argonne and in conferences. Concerning the percentage of serial jobs, in the the first and second quarter, our jobs will be 80% serial. For the third and fourth quarter, we will have 30%. Project URL: http://www.mcs.anl.gov/research/project_detail.php?id=90 Current FY Hours Used: undetermined amount New FY Requested allocation: 95000 Q1: 15000 Q2: 30000 Q3: 25000 Q4: 25000 Justification: Thank You, The LCRC Accounts System

1 0

[LCRC Accounts] Yearly Allocation Request from ApsRenewalLattice
by accounts＠lcrc.anl.gov 29 Aug '12

29 Aug '12

Hello, A yearly allocation for the LCRC cluster has been requested with the following updated information: Submitter/PI: Michael Borland Project Name: ApsRenewalLattice Division: ASD Project title: Lattice Development and Beam Dynamics for APS Upgrade Associated funding: Department of Energy,Office of Science, Office of Basic Energy Sciences Other Systems: ASD "weed" cluster, operated by ASD/AOP group. weed has 56 Nehelem processors with a lustre filesystem and infiniband network. I typically have access to about 50% of the cluster full time. Science: As part of the APS Renewal, the accelerator lattice (the location and powering of the magnets) will be modified to incorporate a number of long straight sections for insertion devices and other purposes, such as Short-Pulse X-rays (SPX). This involves breaking the symmetry of the ring, which has serious implications for particle stability. Using simulation-based techniques developed at APS, we can tune the nonlinear elements to improve the dynamic and momentum aperture. Rapid turn-around is essential in order to respond to requests from the APS user community for evaluation of new options. In addition, simulations related to SPX beam dynamics will be performed, since this complex, unique system has the potential to affect the beam in undesirable ways. Project description: APS has developed a method of using multi-objective genetic algorithms (MOGA) for optimization of accelerator lattices. In past years, this method has been applied successfully using fusion, intrepid, and a small APS cluster to development and evaluation of lattices for the APS Upgrade. In the coming year, we will perform detailed evaluation of configurations leading up to the CD2 review. This will primarily consist of detailed multiparticle tracking simulations for the short-pulse x-ray (SPX) system. These simulations will confirm expected performance and tolerances. Beyond that, we will refine the configurations and respond to requests for changes. For example, we will look at the possibility of reoptimizing SPX to use a lower rf frequency. Another configuration that may be explored is the possibility of beta-function manipulation to enhance the performance of SPX. The codes being used have all been used in past years on fusion. These include Pelegant, a parallel accelerator simulation code used primarily for particle tracking and acceptance determination, as well as a Tcl script used for genetic optimization. The simulations are close to embarrassingly parallel and thus scale very well. Project URL: Current FY Hours Used: undetermined amount New FY Requested allocation: 450000 Q1: 150000 Q2: 100000 Q3: 100000 Q4: 100000 Justification: These codes have been used in past years on fusion, intrepid, and the APS weed cluster and show excellent scaling. The basic simulations involve single particle beam dynamics, which is close to embarrassingly parallel. In addition, the genetic algorithm script does little more than manage jobs in the batch queue. Thank You, The LCRC Accounts System

1 0

New LCRC report submitted for dynamics
by support＠lcrc.anl.gov 29 Aug '12

29 Aug '12

Full project report at https://accounts.lcrc.anl.gov/reporting/project/2012/?project=dynamics

1 0

[LCRC Accounts] Yearly Allocation Request from Drude
by accounts＠lcrc.anl.gov 29 Aug '12

29 Aug '12

Hello, A yearly allocation for the LCRC cluster has been requested with the following updated information: Submitter/PI: Benoit Roux Project Name: Drude Division: BIO Project title: Development and validation of new polarizable force field for molecular dynamics Associated funding: LDRD: 2009-047-R1, "Novel Computing Methodologies for the Simulation of Complex Molecular Systems" NIH grant from NIGMS R01-GM072558: Polarizable Force Field For Proteins and Lipids Other Systems: BG/P, XSEDE, Anton, kbt, Beagle Science: The functional form and parameters of a new force field accounting for induced polarization in molecular dynamics simulations must be developed, optimized, and validated for various biomolecular systems. Project description: At the present time, MD simulations are not only limited by the computational resources but also by the physical correctness and accuracy of the underlying atomic models. Without a systematic approach to generate, test and validate the accuracy of the force fields, the value of MD simulations will always remain in question. The development of potential functions for computer simulations has a long history and is still a very active field. Essentially all of the MD simulations of complex molecular systems (biological or not) are currently carried out using fixed-charge non-polarizable force fields, even though there is a broad consensus that electronic polarization effect ought to be included. The important role of polarization is particularly obvious for self-assembling amphiphilic molecules, containing polar or charged groups as well as nonpolar moieties (see below). A number of research groups are working at developing new algorithm to treat polarizati on, either with inducible point dipoles, fluctuating charges, or classical Drude oscillators. The latter approach, which we have adopted, is the most advantageous due to its simplicity and computational efficiency. The goal is to develop, optimize, and validate the functional form and parameters of a new force field accounting for induced polarization in molecular dynamics simulations. We are particularly interested in ion solvation properties. # computational methods The computational method involves ab initio quantum mechanics, as well as classical molecular dynamics simulations and free energy perturbation. The induced polarization is represented by introducing a small auxiliary charged particle attached to the nucleus by a spring. The pair forms a Drude oscillator. It can be shown that this construction accounts properly for the physics of induced polarization in molecular interactions. # scalability issues The new force field is developed to be implementable in a wide range of existing molecular dynamics codes. Currently, the model is supported by CHARMM, and in NAMD. Scalability is typically better with NAMD. But that is for simple brute force simulations. For alchemical free energy perturbation, we can run replica-exchange simulations FEP/REMD using CHARMM/REPD, which is extremely scalable. # computations planned We are now refining ion solvation models in liquid N-methylacetamide (NMA) for divalent cations. The work will comprise - high level ab initio on small clusters - FEP divalent ions solvation in NMA # software requirements We use Gaussian, Qchem, CPMD, CHARMM, and NAMD. # sizes of calculations run on and/or planned for Fusion or other systems The work always starts with some ab initio in small clusters. The most demanding part of the project corresponds to the large number of FEP computations needed to fully optimize the force fields. The FEP simulations are extremely scalable via CHARMM/REPD when we carry them as replica-exchange MD. The osmotic pressure require several MD at increasingly high concentrations with large systems. Lastly, we also use CPMD for some systems when required (like we simulated Ca2+ and Ba2+ in NMA to clarify issues of coordination numbers). # expected number of project members: The number of members is expected to be ~3, including Benoit Roux, Chris Rowley and Hui Li (divalent ions), from the University of Chicago. In the coming year we need to parametrize divalent cations solvated in N-methylacetamide. This will require both ab initio calculations as well as solvation free energy computations. We carry out the latter using our highly scalable replica-exchange methodology. Request: The dominant part of the project is the FEP calculations. To estimate the number of SUs, it is sufficient to focus on what is needed for this. For one case, we typically require about 100 trials to scan the space of parameters and search for the optimal model. We have to do 3 cases (Ca+2, Mg2+, Ba+2 in liquid NMA). A single solvation FEP/lambda-REMD calculation: 75 windows/replica One replica use one node (8 cores). Sampling of 0.2 ns takes 4 hours. Total needed: 75*4*8 = 2,400 core-hours Total needed for one complete parametrization : 240,000 core-hours The total needed for 3 complete parametrization: 720,000 core-hours. We also need about 80,000 core-hours for various ab initio and CPMD simulations. Grand total requested: 800,000 core-hours. RECENT REFERENCES Yu, H; Mazzanti, CL; Whitfield, TW, et al. A Combined Experimental and Theoretical Study of Ion Solvation in Liquid N-Methylacetamide. JOURNAL OF THE AMERICAN CHEMICAL SOCIETY Volume: 132 Issue: 31 Pages: 10847-10856 Published: 2010 Baker, CM; Lopes, PEM; Zhu, X, et al. Accurate Calculation of Hydration Free Energies using Pair-Specific Lennard-Jones Parameters in the CHARMM Drude Polarizable Force Field. JOURNAL OF CHEMICAL THEORY AND COMPUTATION Volume: 6 Issue: 4 Pages: 1181-1198 Published: 2010 Yu, HB; Whitfield, TW; Harder, E, et al. Simulating Monovalent and Divalent Ions in Aqueous Solution Using a Drude Polarizable Force Field. JOURNAL OF CHEMICAL THEORY AND COMPUTATION Volume: 6 Issue: 3 Pages: 774-786 Published: 2010 Project URL: http://thallium.bsd.uchicago.edu/RouxLab/ Current FY Hours Used: undetermined amount New FY Requested allocation: 800000 Q1: 200000 Q2: 200000 Q3: 200000 Q4: 200000 Justification: All the codes and methodologies that we use are high performance and highly scalable. Thank You, The LCRC Accounts System

1 0

New LCRC report submitted for LiBattModeling
by support＠lcrc.anl.gov 29 Aug '12

29 Aug '12

Full project report at https://accounts.lcrc.anl.gov/reporting/project/2012/?project=LiBattModeling

1 0