Lennart Johnsson's Panel

Application Specific PSE's

Introduction

* VanRosendale not here
* Panel is Numerical Methods/Parallelism background
* CATIA is an example PSE
* Note finite element businesses are very proprietary and not open

- ABACUS is open and this helps their success?

* Can you build a business case for PSE's

L Johnsson -- Contd I

* Why are many PSE's not out there?

- Is it just that hardware only just becoming available?

* Is Scalability important?
* What are enabling technologies
* What about testing and validation?
* Is nature of required PSE's too multidisciplinary for academics?

L Johnsson -- Contd II

* Who are users?

- Note CATIA set up for use by teams

* What should academia do

- Set up standards
- build prototypes, components
- establish test and validation cases and procedure

Randall Bramley

* Talked to users of (Mathworks) tools in research and teaching
* PSE must allow user to intervene

- at all levels of PSE
- Monolithic systems not good

* PSE's must be OPEN
* Symbolic manipulation leads to unreadable unmodifiable sometimes inefficient code

Randall Bramley II

* Users complain PSE's can't solve large problems
* More efficient standard interfaces between PSE's

- Microsoft OLE

* Let users build their own application specific PSE's in terms of Meta-PSE systems

Gallopoulos from Illinois

* Quicken and Tax Programs are well known PSE's
* PSE's should use language of user domain
* Compiler is a problem solving environment?

- Yes as say HPF is specialized in terms of supported data structures

* Illinois building MATLAB to F90 compiler

Marinescu from Purdue on PSE for Structural Biology

* 10-15 such groups nationwide
* Commercial use down the road
* Both computer science and biology issues
* Likes adaptive algorithms which are self scheduling and make use of whatever machines are available.
* Has system called Bond which is intelligent shell hiding details from user
* Socrates is a spreadsheet like GUI to define problem

Questions to Panel I

* Sherman(SCA) agrees that CORBA is not suitable for parallel implementations
* Sewell agrees users want to modify code produced by PSE's -- Protran from IMSL (Sewell's work) used to produce unreadable code.

* Build application specific PSE on top of general purpose PSE

Questions to Panel II

* Analogy to specialized versus general purpose hardware
* Can't start from scratch!
* Sherman: must be able to certify code -- can't modify code used in real world so open-ness can be a liability
* Are there two markets?

- Engineering -- can't modify
- Research -- must be able to modify
- Also Users versus Developers

* Note MATLAB and Maple combined succesfully!

Questions to Panel III

* Boeing trying to get more and more software from the outside except in aeronautics
* Boeing interested in PSE and agrees on legal issues brought up by Sherman
* Academia can only help set standards as in most cases industry is solving real and much larger problems
* Should tools for developing PSE's be in public domain?

- Certainly should be available but Lennart is worried about standard problem that no incentive for commercial activities

* John Rice agrees that PSE's need commercial support!
* MATLAB contrated with ? to produce sparse MATLAB. ? is NAG?