NERSCPowering Scientific Discovery Since 1974

Publications, Presentations, Posters

Listed below are some of the publications, presentations, posters and other output of the Computational Science and Engineering Petascale Iniative's post-doctoral researchers.

Exploitation of Dynamic Communication Patterns through Static Analysis

Overlapping Communication with Computation using OpenMP tasks on the GTS magnetic fusion code

  Performance experiments on BOUT++

Download File: PraveenBOUT++.pdf | pdf | 3.6 MB

  BOUT++: Performance Characterization and Recent Advances in Design

Download File: icnsp2011-praveen.pdf | pdf | 1.7 MB

Poster presented at ICNSP 2011 describing BOUT++ design improvements and performance characterization.

  Petascale Pos-doc Program Reports and Slides

Download File: LBL-postdocsJune2010.pdf | pdf | 10 MB
Download File: ARRA-Koniges-030316.ppt | ppt | 3.1 MB

  Acceleration of Porous Media Applications on the Cray XE6 Platform

Download File: kfagnancug2011.pdf | pdf | 1.3 MB

Iterative Krylov solution methods for geophysical electromagnetic simulations on throughput-oriented processing units

A numerical scheme for including surface tension effects in hydrodynamic simulation: a full Korteweg type model without parasitic flows

  Modeling droplet breakup effects in warm dense matter experiments with diffuse interface methods in ALE-AMR code

Download File: wangyiliu.pdf | pdf | 302 KB

Conference paper

  Modeling Warm Dense Matter Experiments using the 3D ALE-AMR Code and the Move Toward Exascale Computing

Download File: IFSA-koniges2.pdf | pdf | 783 KB

Conference proceeding

  A Next Generation Light Source Facility at LBNL

Download File: PAC11NGLS.pdf | pdf | 406 KB

  GPU Computational Screening of Carbon Capture Materials

Download File: SciDacPosterJihan.pptx | pptx | 3.7 MB
Download File: kimjscidac.pdf | pdf | 3 MB

Application Acceleration on Current and Future Cray Platforms (Best Paper)

Calculation of RI-MP2 Gradient using Fermi GPUs

Compressive phase contrast tomography

When x-rays penetrate soft matter, their phase changes more rapidly than their amplitude. Interference effects visible with high brightness sources creates higher contrast, edge enhanced images. When the object is piecewise smooth (made of big blocks of a few components), such higher contrast datasets have a sparse solution. We apply basis pursuit solvers to improve SNR, remove ring artifacts, reduce the number of views and radiation dose from phase contrast datasets collected at the Hard X-Ray Micro Tomography Beamline at the Advanced Light Source. We report a GPU code for the most computationally intensive task, the gridding and inverse gridding algorithm (non uniform sampled Fourier transform). DOI: 10.1117/12.861946

Compressive auto-indexing in femtosecond nanocrystallography

  Performance Characterization and Implications for Magnetic Fusion Co-design applications

Download File: cug2011-praveen.pdf | pdf | 376 KB

  Influences of Architecture and Threading on the MPI Communication Strategies in an Accelerator Simulation Code

Download File: BAustinSciDacPoster.pptx | pptx | 810 KB


  Parallel Real-Time Ptychographic Image Reconstruction using GPUs

Download File: ptychoposter.pdf | pdf | 10 MB


  Modeling droplet breakup effects with applications in the warm dense matter NDCX experiment

Download File: poster.pdf | pdf | 613 KB


  Global Address Space Communication Techniques for Gyrokinetic Fusion Applications on Ultra-Scale Platforms

Download File: PreisslSciDacPoster.pdf | pdf | 646 KB


  Implicit, full coupling of the adaptive grid and physical problems in plasma simulation with additive Schwarz preconditioned inexact Newton

Download File: postericiam2011xuefeiyuan.pptx | pptx | 6.4 MB

Adaptive solution refinement through grid motion, with high solution variation regions serving as attractors for grid density or magnetic flux surfaces serving as constant-coordinate alignment targets, is a capability of increasing interest to continuum-based models of fusion devices. In this study, the parallel Newton-Krylov- Schwarz (NKS) iterative method is employed for a fully coupled, implicit simulation of magnetic reconnection using a moving adaptive grid. The four-field extended MagnetoHydroDynamics (MHD) equations are trans- formed into a time-dependent curvilinear coordinate system, which is a solution of the grid problem. However, the convergence of the Newton algorithm is not observed for high skin depth number di. The idea of nonlinear preconditioning has proven to be useful to remove the local high nonlinearity where the nonlinear function is replaced by a preconditioned function with more balanced nonlinearities. Convergence and accuracy results are presented demonstrating that an efficient fully coupled, implicit implementation is possible with increased accuracy.

  Iterative and direct linear solvers in fully implicit magnetic reconnection simulations with inexact Newton methods

Download File: postericnsp2011xuefeiyuan.ppt | ppt | 7.5 MB

Global Address Space Communication Techniques for Gyrokinetic Fusion Applications on Ultra-Scale Platforms

  Parallel Index and Query for Large Scale Data Analysis

Download File: FQSC11.pdf | pdf | 919 KB

  Modeling droplet breakup effects with diffuse interface methods in ALE-AMR code with application in modeling NDCX-II experiments

Download File: poster-1.pdf | pdf | 940 KB


  The Challenges Ahead

Download File: Challenges-Ahead.pdf | pdf | 4.8 MB

  Application of PDSLin to the magnetic reconnection problem

Download File: CSDRebeccaYuanetc.pdf | pdf | 371 KB