HPC Problem Solving Environments
The aim of this research is to formulate and develop HPC infrastructure, including scalable libraries of legacy partial differential equations (PDE) software, kernels for parallel reuse PDE methodologies, new parallel PDE libraries, problem solving environments, and underlying theories needed to support multidisciplinary physical prototyping in the classroom, laboratory, desk, and factory.
An object-oriented approach is used with multiple levels. At the top level are large complex objects like PDE PROBLEM, NUMERICAL PDE PROBLEM, SPECIFICATIONS, and SOLUTION. At the bottom are utiliy objects like BLAS (Basic Linear Algebra Subroutines) to manipulate vectors and arrays. Each object has substantial identifying information about its type, capabilities, properties, and interface requirements. This allows the structure to be open ended.
Slide presentations from the review meeting at Purdue University in November, 1994:
Recent Panels & Presentations on HPC Problem Solving Environments:
The //ELLPACK WWW Server is now up and running! Click here to find out more about it.
Quad Chart describing our research for 1996 and beyond.
The image map below allows you to explore the various aspects of this project by following different links. Click on a colored box to link to the page(s) that describe that task. For work that is in progress or has not yet been placed on the web, the link will take you back to our top level page.
For further information contact pellpack@cs.purdue.edu