Advertisement

Parallel nearest neighbor calculations

  • Harold Trease
Session II: Mesh Management and Visualization
Part of the Lecture Notes in Physics book series (LNP, volume 395)

Abstract

We are just starting to parallelize the nearest neighbor portion of our free-Lagrange code. Our implementation of the nearest neighbor reconnection algorithm has not been parallelizable (i.e., we just flip one connection at a time). In this paper we consider what sort of nearest neighbor algorithms lend themselves to being parallelized. For example, the construction of the Voronoi mesh can be parallelized, but the construction of the Delaunay mesh (dual to the Voronoi mesh) cannot because of degenerate connections. We will show our most recent attempt to tessellate space with triangles or tetrahedrons with a new nearest neighbor construction algorithm called DAM (Dial-A-Mesh). This method has the characteristics of a parallel algorithm and produces a better tessellation of space than the Delaunay mesh. Parallel processing is becoming an everyday reality for us at Los Alamos. Our current production machines are Cray YMPs with 8 processors that can run independently or combined to work on one job. We are also exploring massive parallelism through the use of two 64K processor Connection Machines (CM2), where all the processors run in lock step mode. The effective application of 3-D computer models requires the use of parallel processing to achieve reasonable “turn around” times for our calculations.

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

References

  1. 1.
    H.F. Trease, “Three-Dimensional Free Lagrangian Hydrodynamics,” Proceedings of the first Free-Lagrange Conference, Lecture Notes in Physics, Springer-Verlag, Vol. 238, pp. 145–157, 1985.Google Scholar
  2. 2.
    M.S. Sahota, “Delaunay Tetrahedralization in a Three-Dimensional Free-Lagrangian Multimaterial Code,” Proceedings of the Next Free-Lagrange Conference, Jackson Lake Lodge, Wyoming, June 3–7, 1990, Springer-Verlag Press, this volume.Google Scholar
  3. 3.
    D.A. Mandell and H.E. Trease, “Parallel Processing a Three-Dimensional FreeLagrange Code: A Case History,” The International Journal of Supercomputer Applications, Vol. 3, No. 2, 1989, pp. 92–99.Google Scholar
  4. 4.
    D.M. Fraser, “Tetrahedral Meshing Considerations for a Three-Dimension FreeLagrangian Code,” Los Alamos National Laboratory report, LA-UR-88-3707, 1988.Google Scholar
  5. 5.
    J.C. Marshall and J.W. Painter, “Reconnection and Fluxing Algorithms in a Three-Dimensional Free-Lagrangian Hydrocode,” Proceedings of the Next Free-Lagrange Conference, Jackson Lake Lodge, Wyoming, June 3–7, 1990, Springer-Verlag Press, this volume.Google Scholar
  6. 6.
    J.H Cerutti and H.E. Trease, “The Free-Lagrange Method on the Connection Machine,” Proceedings of the Next Free-Lagrange Conference, Jackson Lake Lodge, Wyoming, June 3–7, 1990, Springer-Verlag Press, this volume.Google Scholar
  7. 7.
    M.S. Sahota, “Three-Dimensional Free-Lagrangian Hydrodynamics,” Los Alamos National Laboratory report, LA-UR-89-11-79, 1989.Google Scholar
  8. 8.
    M.S. Sahota, “An Explicit-Implicit Solution of the Hydrodynamic and Radiation Equations,” Proceedings of the Next Free-Lagrange Conference, Jackson Lake Lodge, Wyoming, June 3–7, 1990, Springer-Verlag Press, this volume.Google Scholar
  9. 9.
    M.S. Sahota and H.E. Trease, “A Three-Dimensional Free-Lagrange Code for Multimaterial Flow Simulations,” Proceedings of the ASME-JSME International Symposium on Liquid-Solid Flows, Portland, Oregon, June 24–26, 1991.Google Scholar

Copyright information

© Springer-Verlag 1991

Authors and Affiliations

  • Harold Trease
    • 1
  1. 1.Los Alamos National LaboratoryUSA

Personalised recommendations