The Numerical Mathematics Consortium, a group of industry and academic experts working to define a standard for mathematical functions applicable to numerical algorithms, announced the fourth draft release of the standard document. A highlight of the release is a new scorecard framework for organizing and presenting functions covered by the standard. This approach helps engineers, scientists and educators assess the quality and expected functionality of math function implementations on various computational platforms by calling out implementation-specific concerns. The latest draft revision, which includes examples of the new scorecard framework, is presented through an interactive Web interface to encourage input and feedback from domain experts.
“The Numerical Math Consortium’s effort is significant because the group is taking a principled approach to defining a standard for computational mathematics, building from a solid foundation of mathematical theory to promote implementations that are both consistent and complete,” said Dr. Alan Edelman, professor of applied mathematics at MIT and member of the Numerical Math Consortium. “With the new wiki-style presentation, the consortium is expanding to provide domain experts the opportunity to contribute and add value to the standardization process.”
The new scorecard framework outlines assessment categories such as speed, resource requirements and accuracy to give users an overview of how functions should behave given implementation on specific computing systems. This information offers guidance to users on how to implement, evaluate and apply standard-compliant functions. To initiate this framework, the consortium has developed example scorecards for the sine and QR decomposition functions.
To encourage participation from domain experts in the development of the scorecards and the standard itself, the Numerical Math Consortium is presenting the latest draft standard in a dynamic, wiki-style format. This new presentation style makes it possible for users to provide comments and feedback in real time. It also helps the consortium integrate material that addresses new technologies in a timely manner.
Additional revisions to the fourth draft standard include clarifications to the sections on scope, goals and principles and guidelines for participation. The consortium also has added a new chapter that defines the behaviors of computational math and the issues that may arise.
Readers can review the latest draft revision of the standard presented in the wiki-style format and with examples of the new scorecard-based approach. Readers also can view background material related to the standardization effort and learn about how organizations and individuals can get involved in the consortium.
About the Numerical Mathematics Consortium
The Numerical Mathematics Consortium is a nonprofit organization comprising vendors and individuals in industry and academia committed to establishing an open mathematical semantics standard for numerical algorithm development and reuse. Established in 2005 by founding members that include INRIA, Maplesoft, National Instruments (Nasdaq: NATI) and PTC, the Numerical Mathematics Consortium is focused on reducing the overall cost of numerical algorithm development and increasing reuse in both industry and academia for application areas such as embedded design, industrial control and scientific research.