HELP! * GREEN=global GREY=local Full HTML for

GLOBAL foilset Recommendations for Research in Problem Solving Environments

Given by Geoffrey C. Fox at Arpa PI Meeting on 14 February 1996. Foils prepared 6 May 1996
Abstract * Foil Index for this file See also color IMAGE

This was a summary of some conclusions of a workshop at Purdue Sept. 25-27,1995 and presented at ARPA PI meeting at San Antonio, February 14,1996
The recommendations fall into three broad areas:
Basic Research in Architecture of and Technologies for Problem Solving Environments
  • These are both PSE components and the technologies/methodologies to build PSE's
Promotion of Interactions between academic, commercial, government computer science and user communities
Construction of (rapid) prototypes of "complete" (useable) PSE's for defense applications

Table of Contents for full HTML of Recommendations for Research in Problem Solving Environments

1 Recommendations for New and Enhanced Activities in Problem Solving Environments
Arpa PI Meeting
San Antonio Texas February 14,1996
2 Abstract of Recommendations for Problem Solving Environments (PSE's)
3 Recommendation: Develop Architectures and Frameworks for PSE's - I
4 Recommendation: Develop Architectures and Frameworks for PSE's - II
5 Recommendation: Develop Needed Component Technologies
6 Recommendation: Encourage Multi-Community Interactions
7 Recommendation: Develop Prototype PSE's

This table of Contents Abstract

HELP! * GREEN=global GREY=local HTML version of GLOBAL Foils prepared 6 May 1996

Foil 1 Recommendations for New and Enhanced Activities in Problem Solving Environments
Arpa PI Meeting
San Antonio Texas February 14,1996

From Recommendations for Research in Problem Solving Environments Arpa PI Meeting -- 14 February 1996. * See also color IMAGE
Full HTML Index
Geoffrey Fox
NPAC
Syracuse University
111 College Place
Syracuse NY 13244-4100
HELP! * GREEN=global GREY=local HTML version of GLOBAL Foils prepared 6 May 1996

Foil 2 Abstract of Recommendations for Problem Solving Environments (PSE's)

From Recommendations for Research in Problem Solving Environments Arpa PI Meeting -- 14 February 1996. * See also color IMAGE
Full HTML Index
This was a summary of some conclusions of a workshop at Purdue Sept. 25-27,1995 and presented at ARPA PI meeting at San Antonio, February 14,1996
The recommendations fall into three broad areas:
Basic Research in Architecture of and Technologies for Problem Solving Environments
  • These are both PSE components and the technologies/methodologies to build PSE's
Promotion of Interactions between academic, commercial, government computer science and user communities
Construction of (rapid) prototypes of "complete" (useable) PSE's for defense applications
HELP! * GREEN=global GREY=local HTML version of GLOBAL Foils prepared 6 May 1996

Foil 3 Recommendation: Develop Architectures and Frameworks for PSE's - I

From Recommendations for Research in Problem Solving Environments Arpa PI Meeting -- 14 February 1996. * See also color IMAGE
Full HTML Index
Develop the requirements for both PSE's and their component technologies
  • Suggest focus on scientific and engineering computation with component modules such as computational geometry, grid generation, particle dynamics and sparse matrix solvers
  • what is role of particular component technologies
Architecture should allow a distributed software development approach with synergistic contributions from loosely coupled groups
Take very successful library software and algorithm activity and augment with study of abstract library structure.
  • This must allow "plug and play" of new algorithms or data structures in a way that does not exclude key technologies
HELP! * GREEN=global GREY=local HTML version of GLOBAL Foils prepared 6 May 1996

Foil 4 Recommendation: Develop Architectures and Frameworks for PSE's - II

From Recommendations for Research in Problem Solving Environments Arpa PI Meeting -- 14 February 1996. * See also color IMAGE
Full HTML Index
Focus on the "glue" for building PSE's as well as the "bricks" -- a paradigm shift
This glue implies interfaces and data exchange formats for numerical and symbolic modules needed as PSE components
Glue includes/implies the software (engineering) approach for developing PSE's
The defined architecture must support identified features of PSE's and their development
  • What are the critical features of scientific PSE's that distinguish them from Quicken (a well-known financial PSE) and say a command and control system which could include say a weather simulation as a sub-PSE
Glueing diverse pieces requires second recommendation of crossdisciplinary interactions
HELP! * GREEN=global GREY=local HTML version of GLOBAL Foils prepared 6 May 1996

Foil 5 Recommendation: Develop Needed Component Technologies

From Recommendations for Research in Problem Solving Environments Arpa PI Meeting -- 14 February 1996. * See also color IMAGE
Full HTML Index
Computational Geometry and Grid generation
Collaborative Computing linked to configuration controlled databases with humans and computers in the loop
Multi-level object technology for reusable, maintainable, robust software which is performance portable -- focus on complete PSE and not just computational kernels
Mechanisms for validating correctness of numerical code
Metacomputing -- PSE's are naturally set up as cross-country (institution) distributed computing systems
Wrapper technology for legacy Systems (Rockwell tells me next DoD aircraft needs 10,000 separate programs to be run!)
Large scale knowledge base systems for particular domains -- real engineers must use these PSE's
As well as normal scalable algorithms, software, filesystems, databases etc.
HELP! * GREEN=global GREY=local HTML version of GLOBAL Foils prepared 6 May 1996

Foil 6 Recommendation: Encourage Multi-Community Interactions

From Recommendations for Research in Problem Solving Environments Arpa PI Meeting -- 14 February 1996. * See also color IMAGE
Full HTML Index
Many researchers (e.g. Grand Challenge collaborations) are developing independently what are essentially PSE's for physics, chemistry, engineering ....
  • Most do not know they are developing a PSE and that by using such a framework, they could (re)use work of others.
Need to gather these communities -- academia, government, industry, computer technology, application development -- together to establish framework for development of standards, exchange experiences (what works and what doesn't) and requirements
This discussion should include libraries, interfaces, data-structures, representation, middleware and PSE components
HELP! * GREEN=global GREY=local HTML version of GLOBAL Foils prepared 6 May 1996

Foil 7 Recommendation: Develop Prototype PSE's

From Recommendations for Research in Problem Solving Environments Arpa PI Meeting -- 14 February 1996. * See also color IMAGE
Full HTML Index
Of course all this basic research needs to be tested and evaluated!
We should create some complete scientific and/or engineering prototype PSE's
This should be multi-disciplinary on both computer technology side (algorithms, software, user-interface, artificial intelligence, database, collaboration etc.) and on application side
Choose an area where PSE adds value to an established computational area rather than proposing a new solution
  • So we test PSE approach and not role of computation in a particular area
Set-up as microcosm of proposed general case with distributed loosely coupled development as well "entries" in key components
Northeast Parallel Architectures Center, Syracuse University, npac@npac.syr.edu

If you have any comments about this server, send e-mail to webmaster@npac.syr.edu.

Page produced by wwwfoil on Tue Feb 18 1997