Verification of Computer Codes in
Computational Science and Engineering

Dr. Patrick Knupp
Sandia National Laboratories

To navigate, use pull-down menus above or go to: 7th USNCCM Site Map

Post-Congress Short Course

Thursday, July 31, 2003
8:30 am - 12:30 pm (4 hours)
Albuquerque Hyatt Hotel
(Conference Hotel)


Course Description

This short course will summarize methods for computer code verification, emphasizing primarily a particular method known as the Method of Manufactured Solutions. We begin with an informal discussion of code verification to establish context and historical background. We then provide a brief review of terminology for partial differential equations and basic numerical methods to facilitate subsequent discussions. A step-by-step procedure for verifying code order-of-accuracy is described for use by code developers, testers, and analysts. Issues related to the design of a suite of coverage tests suitable for order-verification are considered in depth. Manufactured solutions are carefully described as a method for generating complex yet exact solutions to the governing differential equations. We show by examples the benefits of verifying code order-of-accuracy, namely, reliable detection of many coding mistakes. Fine points of the method are described using exercises in which codes solving Burgers' and the Navier-Stokes equations are verified. Additional topics include verification of codes with automatic time-stepping options, hard-wired boundary conditions, artificial dissipation, shock-capturing, and non-ordered approximations. Related topics such as code validation and solution verification are briefly discussed to show how the method of manufactured solutions fits within the Verification & Validation framework.

The course is aimed at computational scientists and engineers, including managers responsible for code development. Code developers, code testers, and analysts working with software that solves partial differential equations will come away with a working knowledge of the method of manufactured solutions and how it can be used to ferret out hard to find coding mistakes and provide a convincing demonstration that the equations are being correctly solved.

The course assumes familiarity with numerical methods for PDE's, computer programming, and some knowledge of an applied discipline such as fluid flow, thermal science, or computational mechanics.

The course may be taken in conjunction with or independently of the pre-congress short course "Verification and Validation in Computational Mechanics" to be held on Sunday, July 27th, 2003. Those interested in the broader issues of Verification & Validation are directed towards the pre-congress short course, while those mainly interested in the Method of Manufactured Solutions as a practical method for code verification should consider the Thursday course.

Registration Fee $175

The registration fee includes the following:

  • Copy of the book "Verification of Computer Codes in Compuational Science and Engineering," 2003, by Patrick Knupp and Kambiz Salari. [An $80 value]
  • All of the viewgraphs presented at the short course.
  • Continental Breakfast and Coffee Break
  • Registration for this Short Course will be available in April 2003.

Instructor

Dr. Patrick Knupp has 23 years of research, development, and applications experience in computational science. He is the author of the MWT3D code for free-surface/subsurface groundwater flow and numerous papers concerning groundwater flow and transport. He is well-known for his basic research in numerical grid generation, being the co-author of the book "The Fundamentals of Grid Generation" and over three-dozen papers on mesh generation. He has developed and applied the Method of Manufactured Solutions to various codes and problems over the last ten years and is the co-author of the book "Verification of Computer Codes in Computational Science and Engineering. He is presently a Principle Member of the Technical Staff at Sandia National Laboratories.

Dr. Patrick Knupp
Sandia National Laboratories

Email:pknupp@sandia.gov


Technical Program

- Plenary Lectures
- Minisymposia
    - Accepted Minisymposia
    - Minisymposium Proposal Form
- Technical Paper

    - Abstract Submission Form
    - Abstract Formats and Templates
- Short Courses
    - Verification & Validatio - Oberkampf
    - Mesh Generation and Automated Simulation - Owen
    - Verification of Computer Codes in ComputationalScience & Engineering - Knupp
    - Multiscale Multiphysics Computational Solid Mechanics - Fish

 

Last Revised: 3/24/03 4:58 PM       
Comments to Content Provider
7th USNCCM Site Map