Publications

Research Keywords

  • scientific computing
  • Parallel and Distributed Computing
  • Parallel Algorithms
  • Computational Sciences

Presentations

  • Scalable multi-phase flows in complex domains using adaptive octree meshes, Center for Computational Mathematics, Flatiron Institute, New York, NY. Invited Talk/Keynote, Presented, 05/04/2022.
  • Scalable two-phase flows in complex domains, Imperial College, London. Invited Talk/Keynote, Presented, 04/25/2022.
  • Local timestepping and 4D tree-based adaptivity: Enabling spacetime adaptivity for scalable numerical simulations, Applied Physics Applied Mathematics Colloqium, Columbia University. Invited Talk/Keynote, Presented, 04/05/2022.
  • Solving PDEs in space-time: 4D tree-based adaptivity, mesh-free and matrix-free approaches, Numerical Analysis and Scientific Computing Seminar, Courant Institute of Mathematical Sciences, NYU. Invited Talk/Keynote, Presented, 02/18/2022.
  • Scalable adaptive PDE solvers in arbitrary domains, Department of Scientific Comput- ing, Florida State University. Invited Talk/Keynote, Presented, 10/20/2021.
  • Scalable Space-time adaptivity for Simulations of Binary Black Hole Intermediate-Mass-Ratio-Inspirals, Advances and Challenges in Computational Relativity, ICERM, Brown University. Invited Talk/Keynote, Presented, 09/16/2020.
    https://icerm.brown.edu/programs/sp-f20/w1/#worksh...
  • Scalable Space-Time Adaptivity for Simulations of Binary Black Hole Intermediate-Mass-Ratio-Inspirals SIAM PP'20. Invited Talk/Keynote, Presented, 02/13/2020.
    https://meetings.siam.org/sess/dsp_programsess.cfm...
  • Scalable Space-time adaptivity for Simulations of Binary Black Hole Intermediate-Mass-Ratio-Inspirals, CS Colloqium, University of Illinois at Urbana-Champagne. Invited Talk/Keynote, Presented, 10/09/2019.
    https://calendars.illinois.edu/detail/5598?eventId...
  • Scalable Space-time adaptivity for Simulations of Binary Black Hole Intermediate-Mass-Ratio-Inspirals, Center for Relativistic Astrophysics, Georgia Institute of Technology, Atlanta, GA. Invited Talk/Keynote, Presented, 06/05/2019.
    https://cra.gatech.edu/event/cra-seminar-hari-sund...
  • A Scalable Framework for Adaptive Computational General Relativity on Heterogeneous Clusters, Oden Institute, University of Texas at Austin, Austin, TX. Invited Talk/Keynote, Presented, 04/25/2019.
    https://www.oden.utexas.edu/about/events/1364/
  • Scalability & Adaptivity: Achieving Conflicting Goals in a Heterogeneous Computing Era, Mechanical Engineering Seminar, Iowa State University, Ames, IA. Invited Talk/Keynote, Presented, 04/09/2019.
  • dendro-GR: Enabling Adaptivity & Parallelism for Computational Relativity, Computational Challenges in Gravitational Wave Astronomy, IPAM, UCLA . Invited Talk/Keynote, Presented, 02/01/2019.
    https://www.ipam.ucla.edu/programs/workshops/compu...
  • Parallel Fast Gauss Transform, SIAM Parallel Processing for Scientific Computing, Mar 7 2018, Waseda University, Tokyo, Japan. Invited Talk/Keynote, Presented, 03/07/2018.
  • Efficient Parallel Streaming Algorithms for large-scale Inverse Problems - September 13, 2017 - IEEE High Performance Extreme Computing Conference, Waltham, MA. Conference Paper, Refereed, Presented, 09/13/2017.
  • Parallel Algorithms for the Computation of Cycles in Relative Neighborhood Graphs - August 16, 2017 - 46th International Conference on Parallel Processing, Bristol, UK. Conference Paper, Refereed, Presented, 08/16/2017.
  • A Scalable Hierarchical Semi-Separable Library for Heterogeneous Clusters, 46th International Conference on Parallel Processing (ICPP). Conference Paper, Refereed, Presented, 08/16/2017.
  • Machine and Application Aware Partitioning for Adaptive Mesh Refinement Applications, 26th International Symposium on High-Performance Parallel and Distributed Computing, 2017. Conference Paper, Refereed, Presented, 06/12/2017.
  • A Parallel Wavelet Approach for Binary Compact Object Mergers, Hyun Lim, Eric Hirschmann, David Neilsen, William Black, Matthew Anderson, Hari Sundar, Milinda Fernando. Other, Presented, 01/27/2017.
    http://meetings.aps.org/Meeting/APR17/Session/C5.3
  • H-to-P Efficiently: Solving HDG Systems via AMG Within The Nektar++ Framework, Hari Sundar and Robert Kirby, University of Utah, USA; Spencer Sherwin, Imperial College London, United Kingdom SIAM Annual Meeting, Boston, MA . Invited Talk/Keynote, Presented, 07/11/2016.
    http://meetings.siam.org/sess/dsp_programsess.cfm?...
  • A Nested Partitioning Algorithm for Adaptive Meshes on Heterogeneous Clusters, 29th ACM on International Conference on Supercomputing, Newport Beach, CA. Conference Paper, Refereed, Presented, 06/23/2015.
  • Parallel hp-Multigrid for HDG, SIAM Conference on Computational Science and Engineering, Salt Lake City, UT Feb 2015. Invited Talk/Keynote, Presented, 02/23/2015.
  • A Nested Partitioning Scheme for Adaptive Meshes on Parallel Heterogeneous Clusters SIAM Conference on Parallel Processing for Scientific Computing, Portland, OR. Other, Presented, 02/20/2014.
    http://www.siam.org/meetings/pp14/

Research Groups

  • David van Komen, Graduate Student. 01/18/2021 - present.
  • LeAnn Leslie, Graduate Student. 08/17/2020 - present.
  • Eric Taylor Heisler, Graduate Student. 08/26/2019 - present.
  • Songzhe Xu, Postdoc. 06/03/2019 - 10/01/2021.
  • Bobby King, Undergraduate Student. 01/07/2019 - 12/23/2019.
  • Liam Moynihan, Graduate Student. School of Computing. 01/07/2019 - present.
  • Han Duc Tran, Graduate Student. 08/20/2018 - present.
  • Masado Ishii, Graduate Student. 08/20/2018 - present.
  • Maxx Carlson, Graduate Student. School of Computing. 08/14/2017 - present.
  • Weerahannadige Milinda Shyamala Fernando, Graduate Student. 08/17/2015 - 07/30/2021.
  • Seyed Majid Rasouli-Pichahi, Graduate Student. 08/17/2015 - 07/30/2021.
  • Christopher Mertin, Graduate Student. 08/17/2015 - 07/10/2017.

Languages

  • Hindi, fluent.
  • Tamil, fluent.
  • Urdu, fluent.

Software Titles

  • Dendro-GR GPU . A portable, highly-scalable, extensible, and easy-to-use public infrastructure for general relativity simulations that is able to run efficiently on modern GPU clusters. The goal of this work is to perform advanced, massively parallel numerical simulations of IMRIs with mass ratios on the order of 1/100 to generate waveforms that can be used in LIGO data analysis and to calibrate approximate methods. Release Date: 11/16/2021. Inventors: Hari Sundar, Milinda Fernando, David Nielsen.
  • Dendro-LTS. Package for locally adaptive time-stepping on octree meshes. This has been extended from previous versions to support GPUs. Release Date: 09/15/2021. Inventors: Milinda Fernando, Hari Sundar.
  • Finch. Finch is a Julia-based domain specific language for large-scale PDE simulations. It allows a high level mathematical description of the problem and automatically generates scalble C++ code that can be run efficiently on large HPC clusters. finchdsl.org . Release Date: 09/15/2021. Inventors: Eric Heisler, Hari Sundar .
  • UQ-SKetch. Code to compress simulation data on the fly for UQ problems, thereby making the gradient and Hessian computations more efficient. Release Date: 08/18/2020. Inventors: Liam Moynihan, Milinda Fernando, Hari Sundar.
  • EigenMM: Scalable Generalized Eigensolvers. We present highly scalable generalized eigensolvers for computing the full spectrum of operators, mainly for the purpose of evaluating fractional operators. Release Date: 08/03/2020. Inventors: Max Carlson, Hari Sundar.
  • Saena: Scalable Algebraic Multigrid. Saena is an extremely scalable, large-scale AMG solver and preconditioner for solving large elliptic problems. Release Date: 10/15/2018. Inventors: Majid Rasouli, Hari Sundar.
  • DendroGR. A portable, highly-scalable, extensible, and easy-to-use public infrastructure for general relativity simulations that will be forward-compatible with next-generation heterogeneous clusters. The goal of this work is to perform advanced, massively parallel numerical simulations of IMRIs with mass ratios on the order of 1/100 to generate waveforms that can be used in LIGO data analysis and to calibrate approximate methods. Release Date: 07/18/2018. Inventors: Hari Sundar, Milinda Fernando, David Nielsen.
  • esort: energy and power efficient sorting on distributed GPU clusters. The package implements communication avoiding distributed sorting algorithms along with highly optimized GPU sorting routines for node-local sorting. The code is highly tuned and provides parallelism using MPI, OpenMP, CUDA and SIMD vectorization. Release Date: 11/16/2015.
  • homg. High-order finite-element package using hexahedral elements. The code is a testbed for geometric multigrid approaches for high order discretizations. The current implementation supports setting up a combination of h and p heirarchy. The following smoothers are supported, * Jacobi * Chebyshev-accelerated Jacobi * block Jacobi * Symmetric SOR . Release Date: 02/24/2014. Inventors: Hari Sundar.
  • hyksort. Highly scalable distributed sorting and selection library. The package im- plements BitonicSort, MergeSort, SampleSort and HykSort. The code is highly tuned and provides parallelism using MPI, OpenMP and SIMD vectorization. Release Date: 02/02/2014. Inventors: Hari Sundar, Dhairya Malhotra.
  • pfgt. A distributed memory implementation of the fast Gauss Transform. Fast adaptive parallel algorithms to compute the sum of N Gaussians at M points using the fast Gauss Transform. We use parallel octrees and a new scheme for translating the plane-waves to efficiently handle non-uniform distributions. Release Date: 02/15/2011.
  • Dendro. A C++ library for constructing and balancing octrees in parallel. It also generates hexahedral meshes from the octrees and extends PETSc’s distributed ar- ray framework to support octree-based meshing. Basic routines for solving PDEs on such meshes using the finite element method are also provided. Release Date: 03/16/2009. Inventors: Hari Sundar, Rahul Sampath.