Current Talks

• Slides from a workshop presentation on the QCDOC machine at Brookhaven in Februrary 2003 are available here in ppt.

• Slides from a minisymposium presentation on fusion simulation given at the SIAM CS&E meeting in February 2003 are available here in ppt.

• Slides from a workshop presentation on parameter identification in wildland firespread modeling given at the IMA in January 2003 are available here in ppt.

• Slides from a 7-hour short course given at Peking University from July 29 to August 2, 2002 are available here in zipped pdf:

  "Parallel Implicit Methods for Nonlinear PDEs: Introduction, Motivation, and Fundamentals" (2 hours)
  "Practical Issues in the Terascale Simulation of PDEs: Scalability and Performance (2 hours)
  "Applications-specific Issues in the Terascale Simulation of PDEs" (1 hour)
  "Overview of the U.S. DOE Scientific Discovery through Advanced Computing (SciDAC) Project" (2 hours)

  Bill Gropp's portion of the same short course (another seven hours on the use of PETSc) is available here.

• Slides from three recent fusion energy simulation meetings in May, August, and September 2002 are available here in powerpoint.

  "Algorithms and Software for Scalable Fusion Energy Simulations" (Integrated Simulations of Fusion Systems Workshop at General Atomics)
  Solvers for Real People (Magnetoflow Workshop at General Atomics)
  "Algorithms and Interpretation" plenary and breakout session summaries (Community ISFOS Workshop at San Diego's Hyatt Regency Islandia)

• Slides from a 60-minute version of a talk entitled "Domain Decomposition in the Mainstream of Computational Science" given in many versions in early 2002 are here in ppt. (This is the April 10 version, Harvard Applied Mechanics Colloquium.)

• Slides from a 25-minute SIAM minisyposium at the 2002 Optimization Conference entitled "Lagrange-Newton-Krylov Methods in PDE-Constrained Optimization" are here in ppt.

• Poster images from "Lagrange-Newton-Krylov-Schur-Schwarz: Matrix-free Methods for PDE-constrained optimization" presented at The First SIAM Conference on Computational Science and Engineering (September 22, 2000, Washington, DC) are here (in pdf). (This poster is closely patterned on the following plenary; it is not a distinct presentation, but a subset.)

• Slides from a talk entitled "Parallel Algorithms for Large-scale Optimization" given as a plenary at The 8th AIAA/NASA/USAF/ISSMO Symposium on Multidisciplinary Analysis and Optimization (September 8, 2000, Long Beach) are here (in postscript). Related abstract (in html).

• Slides from a talk entitled ``Four Horizons for Enhancing the Performance of Parallel Simulations Based on Partial Differential Equations'' given as a plenary at Europar 2000 (August 30, 2000, Munich) and GaMM 2001 (February 16, 2001, Zurich) are here (in powerpoint). The abstract is here (in html).

• The slides files for a talk "Matrix-free Methods in Constrained Optimization" given at the International Symposium on Mathematical Programming (August 7-11, 2000, Atlanta) is available in pdf .

• The slides files for a talk "Performance Stresspoints for Parallel Implicit Nonlinear Solvers" given at the ITS Workshop on Scalable Nonlinear Solvers (July 26-28, 2000, Livermore) is available in pdf or postscript .

• The slides files for our recent two talks given at the SIAM Annual Meeting (July 10-14, 2000, San Juan) are here. "Operator-Split Preconditioning for Convection-Diffusion-Reaction Systems" is available in pdf or postscript . "Improving the Performance of Sparse Matrix-Vector Multiplication by Blocking" is available in pdf or powerpoint .

• The slides files for our recent Gordon Bell Award talk at Supercomputing'99, entitled "Achieving High Performance on an Unstructured Grid CFD Application" are here (in powerpoint) and here (in 2-to-a-page postscript - warning: very large file). The anstract is here (in html) and the updated proceedings paper is here (in pdf).

• Abstracts from our three short talks at the 12th International Conference on Domain Decomposition Methods are here, here, and here (in html).

• The abstract of a plenary talk entitled ``Krylov, Lagrange, Newton, and Schwarz: Combinations and Permutations'' at the overlapped Sixth SIAM Optimization Conference and the 1999 SIAM Annual Meeting (May 12, 1999, Atlanta) is here (in html). A similar version was retitled ``AD-based Matrix-free Methods in PDE-Constrained Optimization'' for the AD2000 meeting on Automatic Differentiation, which is the key technology that allows this new formulation to be successfully exploited. The slides are available in postscript or pdf.

• Discussion slides from a DARPA/DSRC Workshop on ``Numerical Simulation of Physical Systems: The State of the Art, and Opportunities for Further Advances'' given January 1999 at the Defense Sciences Resource Council is here (in postscript).

• The main part of a seminar entitled ``Death and Taxes, Nets and Caches: Facing Inevitabilities in Parallel PDE Simulations'' given July 1998 at Boeing is here (in postscript). The abstract is here (in html).

• The main part of a panel presentation entitled ``Modest Proposals for Directions for ASCI-era Algorithms Research'' given July 1998 at the National Science Foundation is here (in postscript).

• The main part of a plenary presentation entitled ``How Scalable is Domain Decomposition in Practice?'' given July 1998 at the 11th International Conference on Domain Decomposition Methods is here (in postscript). The abstract is here (in html).

• The abstract of an upcoming invited talk entitled ``Newton-Krylov Methods with Multilevel Preconditioning: Algorithm-Architecture Trade-offs in the Number of Levels'' at the Ninth Copper Mountain Conference on Multigrid Methods, April 11-16, 1999, Copper Mountain, CO) is here (in html). The slides are here (in postscript) or here (in pdf). An amplified version of the same talk will be given at the International Conference on Preconditioning Techniques for Large Sparse Matrix Problems in Industrial Applications, June 10-12, 1999, at the University of Minnesota Supercomputing Institute.

• As a SIAM Visiting Lecturer I have developed a collection of one-hour talks that are intended for introductory audiences, such as undergraduate mathematics and computer science majors. I am available, on average once per semester, to deliver these pre-professional presentations. Advance planning is essential, please.
  • What's Math got to do with it? (full talk in powerpoint)
  • How Many Computers Fit on the Head of a Pin? (abstract in html).
  • Algorithms for Petaflops Computing (abstract in html). The main part of the (always evolving) talk is here (in postscript), as given at Florida State in January 1999.
  • Think Globally, Act Locally: An Introduction to Domain-Based Parallelism (abstract in html).

• I also have a new 1.5-hour talk Introduction to Computational Science and Engineering (abstract in html).

back to David Keyes' home page