MS - Computation of large numbers of eigenpairs of large sparse matrices
Room TD 5
Peter Arbenz & Yousef Saad
› CORK: Compact Rational Krylov solvers for nonlinear eigenvalue problems - Karl Meerbergen, KU Leuven
10:20-10:50 (30min)
› Optimizing least-squares rational filters for solving interior eigenvalue problems - Jan Winkelmann, Aachen Institute for Advanced Study in Computational Engineering Science - RWTH Aachen
10:50-11:20 (30min)
› Inexact FEAST subspace iteration for solving large sparse eigenvalue problems - Eric Polizzi, University of Massachusetts Amherst
11:20-11:50 (30min)
› Latest Developments in PRIMME to Compute Many Interior Eigenpairs in Large Hermitian Matrices - Eloy Romero Alcalde, College of William and Mary
11:50-12:20 (30min)
10:20 - 12:20 (2h)
MS - Task-based scientific library on top of runtime systems
Room TD 6
H. Ltaief & F. Lopez
› The ParalleX Execution Model and HPX-5 Runtime System for Scalable Libraries - Thomas Sterling, Indiana University Center for Research in Extreme Scale Technologies
10:20-10:50 (30min)
› Scalable Task-Based Software at Petascale and Beyond - Martin Berzins, SCI Institute - University of Utah
10:50-11:20 (30min)
› Exploiting Kepler architecture in sparse direct solver with runtime systems. - Mathieu Faverge, HiePACS, Innovative Computing Laboratory
11:20-11:50 (30min)
› Concurrency is not complexity: A use case with PaRSEC - George Bosilca, Innovative Computing Laboratory, University of Tennessee, Knoxville
11:50-12:20 (30min)
MS - Computation of large numbers of eigenpairs of large sparse matrices
Room TD 5
Peter Arbenz & Yousef Saad
› A quadrature-based parallel eigensolver for large-scale simulations - Tetsuya Sakurai, University of Tsukuba
13:50-14:20 (30min)
› "TraceMIN: A Scalable Parallel Algorithm for Large Sparse Symmetric Eigenvalue Problems" - Ahmed SAMEH, Purdue University [West Lafayette]
14:20-14:50 (30min)
› Computing invariant subspaces in quantum chemistry - Joost VandeVondele, ETH Zurich
14:50-15:20 (30min)
› A Projected Preconditioned Conjugate Gradient Algorithm for Computing a Large number of Eigenpairs - Eugene Vecharynski, Lawrence Berkeley National Laboratory
15:20-15:50 (30min)
13:50 - 15:50 (2h)
MS - Task-based scientific library on top of runtime systems
Room TD 6
H. Ltaief & F. Lopez
› Task-based sparse Cholesky solver on top of runtime system - Florent Lopez, Numerical Analysis Group - STFC Rutherford Appleton Laboratory
13:50-14:20 (30min)
› Hierarchical Computations on Manycore Architectures - George Turkiyyah & Hatem Ltaief, KAUST
14:20-14:50 (30min)
› qr_mumps: a runtime-based Sequential Task Flow parallel solver - Alfredo Buttari, Institut de recherche en informatique de Toulouse
14:50-15:20 (30min)
› Towards Highly Parallel and Compute-bound Computation of Selected Eigenvectors given a Matrix in Schur Form - Lars Karlsson, Umeå University
15:20-15:50 (30min)
MS - Algorithms and applications of Krylov methods and preconditioning
Auditorium Gintrac
Karl Meerbergen
› Preconditioned Krylov solvers for non-linear Schrödinger equation with application in Superfluids and Superconductors - Wim Vanroose, University of Antwerp
16:20-16:50 (30min)
› Multilevel Variable-Block Schur-Complement Based Preconditioning for Large CFD Computation - Bruno Carpentieri, School of Science and Technology, Nottingham Trent University
16:50-17:20 (30min)
› MPI-GPU parallelism in preconditioned Krylov solvers for block-tridiagonal matrices - Alejandro Lamas Daviña, Universitat Politècnica de Valencia
17:20-17:50 (30min)
› Parallelization of the Rational Arnoldi Algorithm - Mario Berljafa, The University of Manchester
17:50-18:20 (30min)
16:20 - 17:20 (1h)
CP - High performance computing
Room TD 5
J. Roman
› Increasing Arithmetic Intensity using Stencil Compilers on Many-Core Architectures - Simplice Donfack, Inria
16:20-16:40 (20min)
› CANCELLED - A Parallel Algorithm for Scalable Bloom Filters Matrix - CANCELLED - Zhu Wang, Data Communication Technology Research Institute
16:40-17:00 (20min)
› Parallel solver for shifted systems - Nela Bosner, Department of Mathematics, Faculty of Science, University of Zagreb
17:00-17:20 (20min)
16:20 - 17:40 (1h20)
CP - Applications
Room TD 6
G. Houzeaux
› Optimizing linear operators within FETI DDM - Václav Hapla, IT4Innovations National Supercomputing Center, VSB - Technical University of Ostrava
16:20-16:40 (20min)
› Development of a Multi-physics Code with Adaptive Mesh Refinement - William Dai, Los Alamos National Laboratory
16:40-17:00 (20min)
› Hybrid programming in the legacy code AVBP using OMPSs - Cyril Fournier, Centre Européen de Recherche et de Formation Avancée en Calcul Scientifique
17:00-17:20 (20min)
› CANCELLED - High Intensity Matrix Polynomial Solvers for the Heat and Poisson Equations. - CANCELLED - Thierry Dumont, Institut Camille Jordan, CNRS UMR 5208.
17:20-17:40 (20min)
17:20 - 18:20 (1h)
CP - Sparse direct methods
Room TD 5
A. Buttari
› Scaling a Multifrontal Solver - Bob Lucas, Livermore Software Technology Corporation
17:20-17:40 (20min)
› Efficient parallelization of direct solvers for isogeometric analysis - Maciej Paszynski, AGH University of Science and Technology [Pologne]
17:40-18:00 (20min)
› Dissection : A direct solver with kernel detection for finite element matrices on multi-core supercomputers - Atsushi Suzuki, Cybermedia Center, Osaka University
18:00-18:20 (20min)
17:40 - 18:20 (40min)
CP - Eigensolver
Room TD 6
O. Coulaud
› Approximating Functionals of Hermitian Matrix Product Operators via a Block Lanczos Method - Moritz August, Technical University of Munich
17:40-18:00 (20min)
› ChASE: Chebyshev Accelerated Subspace iteration Eigensolver library on heterogeneous architectures - Edoardo Di Napoli, Juelich Spercomputing Centre -- Forschungszentrum Jülich GmbH, Juelich Aachen Research Alliance -- High-performance Computing
18:00-18:20 (20min)
MS - Parallel SVD/EVD solvers for large dense matrices
Room TD 5
Marian Vajtersic & Gabriel Oksa
› Computation of the CS and the indefinite CS decomposition - Sanja Singer, University of Zagreb, Faculty of Mechanical Engineering and Naval Architecture
10:20-10:50 (30min)
› Accelerating the SVD using GPUs - Mark Gates, Innovative Computing Laboratory, University of Tennessee
10:50-11:20 (30min)
› Asymptotic Quadratic Convergence of the Parallel Block-Jacobi EVD Algorithm for Hermitian Matrices - Gabriel Oksa, Institute of Mathematics, Slovak Academy of Sciences
11:20-11:50 (30min)
› High-Performance Parallelization Method of DSYRK for SVD and other Matrix Computations on Xeon Phi - Shuhei Kudo, The University of Electro-Communications
11:50-12:20 (30min)
10:20 - 12:20 (2h)
MS - Multilevel parallel linear solvers
Room TD 6
Eric Darve, Hadi Pouransari, Erik Boman & Siva Rajamanickam
› Fast Methods applied to BEM Solvers for industrial problems - Guillaume SYLVAND, Airbus Group Innovations / Inria
10:50-11:20 (30min)
› Performance and scalability of a parallel Block Low-Rank multifrontal solver - Theo Mary, Institut de recherche en informatique de Toulouse
11:20-11:50 (30min)
› Approximate factorization preconditioning with HSS matrices and random sampling - Pieter Ghysels, Lawrence Berkeley National Laboratory
11:50-12:20 (30min)
MS - Parallel SVD/EVD solvers for large dense matrices
Room TD 5
Marian Vajtersic & Gabriel Oksa
› Preprocessing Algorithm for the Generalized SVD on the Graphics Processing Units - Goran Flegar, University of Zagreb, Faculty of Science, Department of Mathematics
13:50-14:20 (30min)
› GPU-hosted SVD computations for large scale hierarchical matrices - George Turkiyyah, KAUST, American University of Beirut
14:20-14:50 (30min)
› ELPA: Algorithmic extensions and optimization - Michael Rippl, Fakultät für Informatik
14:50-15:20 (30min)
› CANCELLED - An accelerated tridiagonal divide-and-conquer algorithm on distributed memory architectures - CANCELLED - Shengguo Li, National University of Defense Technology (CHINA)
15:20-15:50 (30min)
13:50 - 15:50 (2h)
MS - Multilevel parallel linear solvers
Room TD 6
Eric Darve, Hadi Pouransari, Erik Boman & Siva Rajamanickam
› Feedback in the use of multilevel parallel linear solvers for thermomechanical studies - Olivier Boiteau, EDF Lab - Natacha Béreux, EDF Lab
13:50-14:20 (30min)
› Designing an efficient and scalable block low-rank direct solver for large scale clusters - Xavier Lacoste, Commissariat à l'Energie Atomique et aux Energies Alternatives
14:20-14:50 (30min)
› A massively parallel sparse linear solver ESPRESO - Alexandros Markopoulos, IT4Innovations National Supercomputing Centre
14:50-15:20 (30min)
› Parallel hierarchical solver - Eric Darve, Stanford
15:20-15:50 (30min)
MS - Advanced topics for EVP and SVD towards future scientific simulation
Room TD 5
Toshiyuki Imamura & Tetsuya Sakura
› Parallel dense eigenvalue solver and SVD solver for post-petascale computing systems - Toshiyuki Imamura, RIKEN Advanced Institute for Computational Science
16:20-16:50 (30min)
› A real-valued method for solving complex symmetric linear systems arising in contour integral eigensolver - Yasunori Futamura, University of Tsukuba
16:50-17:20 (30min)
› A new implementation of the orthogonal QD algorithm and its application - Kinji Kimura, Graduate School of Informatics, Kyoto University
17:20-17:50 (30min)
› Performance analysis of the quadruple precision eigensolver library QPEigenK on the K computer - Yusuke Hirota, RIKEN Advanced Institute for Computational Science
17:50-18:20 (30min)
16:20 - 18:20 (2h)
MS - Fast sparse selected inversion techniques
Room TD 6
Olaf Schenk & Matthias Bollhöfer
› Applications of trace estimation techniques - Yousef Saad, University of Minnesota
16:20-16:50 (30min)
› Parallel Selected Inversion of Asymmetric Matrices - Mathias Jacquelin, Lawrence Berkeley National Laboratory
16:50-17:20 (30min)
› Takahashi formulas and the generalized Takahashi structure - Klaus Gaertner, Olaf Schenk, Università della Svizzera italiana
17:20-17:50 (30min)
› Highly parallel stochastic estimation of the diagonal of inverse matrices - Fabio Verbosio, Università della Svizzera italiana, Institute of Computational Science
17:50-18:20 (30min)
MS - Efficient computation of inner eigenvalues of large sparse matrices
Room TD 5
Achim Basermann & Gerhard Wellein
› Efficient subspace iteration with Chebyshev-type filtering - Bruno Lang, University of Wuppertal
10:20-10:50 (30min)
› Parallel methods for computing interior eigenvalues in linear and nonlinear eigenvalue problems in SLEPc - Jose E. Roman, Universitat Politècnica de Valencia
10:50-11:20 (30min)
› Filtered thick restart Lanczos algorithm and the EVSL package - Yousef Saad, University of Minnesota
11:20-11:50 (30min)
› Block Krylov-type complex moment-based nonlinear eigensolver with hierarchical parallelism - Akira Imakura, University of Tsukuba
11:50-12:20 (30min)
10:20 - 12:20 (2h)
MS - Sparse matrix and tensor computations
Room TD 6
Bora Uçar & Albert-Jan Yzelman
› High Performance Parallel Sparse Tucker Decompositions - Oguz Kaya, École normale supérieure de Lyon
10:20-10:50 (30min)
› An Exploration of Optimization Algorithms for High Performance Tensor Completion - George Karypis, Computer Science & Engineering Department, University of Minnesota
10:50-11:20 (30min)
› An Empirical Study of Sparse BLAS on Emerging Heterogeneous Processors - Weifeng Liu, University of Copenhagen
11:20-11:50 (30min)
› Making effective sparse matrix--vector multiplication both portable and usable - Albert-Jan Yzelman, Huawei Technologies France
11:50-12:20 (30min)
› Exploring Vectorization Possibilities on the Intel Xeon Phi for Solving Tridiagonal Systems - Ioannis Venetis, University of Patras [Patras]
13:20-13:40 (20min)
› Streamwise numbering for Gauss-Seidel and bidiagonal preconditioners in convection dominated flows - Paula Córdoba, Barcelona Supercomputing Center
13:40-14:00 (20min)
› SpMV Algorithm for High Bandwidth Memory Based Application Specific Hardware Accelerator - Fazle Sadi, Department of Electrical and Computer Engineering - Carnegie Mellon University
14:00-14:20 (20min)
13:20 - 14:20 (1h)
CP - Preconditioning
Room TD 6
W. Vanroose
› Parallel preconditioners for a higher order time integration methods - Radim Blaheta, Institute of Geonics AS CR
13:20-13:40 (20min)
› Scaling preconditioners for parallel CG-accelerated projection techniques - Dan Gordon, Dept. of Computer Science
13:40-14:00 (20min)
› Large-Scale Sparse Inverse Covariance Estimation - Matthias Bollhoefer, TU Braunschweig, Institute for Computational Mathematics
14:00-14:20 (20min)
Auditorium Gintrac
