Team Arénaire

Overall Objectives
Scientific Foundations
Application Domains
New Results
Contracts and Grants with Industry
Other Grants and Activities

Section: Other Grants and Activities

International Initiatives

Contributions to Standardization Bodies (ANSI-IEEE 754R and ISO/IEC JTC1/SC22/WG21)

The Department of Development and Industrial Relations (DirDRI) at Inria has supported our participation to the ongoing revision of the IEEE Standard for Binary Floating-Point Arithmetic (ANSI-IEEE 754). Since the first visit of M. Daumas, we have taken many opportunities to raise the impact of formal results from our project and lately from the Spaces project that was also supported in 2005. In particular, correctly rounded floating-point elementary functions are now considered for inclusion in the next revision of the IEEE-754 standard. G. Melquiond participated to the meeting of the revision committee in June 2005. Extended scientific reports are available on the intranet of DirDRI.

The challenges encountered when developing the Boost library made clear how an interval arithmetic library and the C++ language have to interact. As a consequence, G. Melquiond wrote, in collaboration with H. Brönnimann (Polytechnic U. Brooklyn, NY USA) and S. Pion (Géométrica team, Sophia Antipolis), a proposal  [50] for integrating interval arithmetic to the C++ standard library which was submitted to the C++ Standards committee (ISO/IEC JTC1/SC22/WG21). H. Brönnimann and S. Pion participated to the Fall meeting of the revision committee in 2005.

LinBox Initiative

Keywords : generic software library, matrix computation, sparse or structured matrix, exact arithmetic, finite field, rational number.

Participant : G. Villard.

LinBox is an ongoing collaborative research project for efficient algorithms and a software library in exact linear algebra (see § 5.6 and § 6.7 ). About thirty researchers from nine institutions in Canada, the USA and France are participating— .

Grant of the Japanese Society for the Promotion of Sciences

Keywords : automatic differentiation.

Participant : N. Revol.

N. Revol obtained a grant of the Japanese Society for the Promotion of Sciences for a short stay in Japan, to collaborate with Prof. K. Kubota, Chuo Univ., Tokyo, on automatic differentiation (postponed due to the pregnancy of N. Revol).

Certifications of Properties of Floating-Point Arithmetic (CNRS-NASA)

Keywords : floating-point, interval arithmetic, formal method, PVS, Coq.

Participants : S. Boldo, F. Cháves, M. Daumas, G. Melquiond.

CNRS PICS 2533 on ``certifications of properties and uses of floating-point arithmetic'' supports our collaboration with the National Institute of Aerospace in Hampton, Virginia. It also involves the École Polytechnique (G. Dowek) and the University of California at Berkeley (W. Kahan). French funding is matched on a mission basis by a Research Cooperative Agreement awarded by NASA Langley Research Center to NIA.

Funding started in Fall 2004 with the visit of Professor Kahan (1989 ACM Turing Award) in Arénaire project. He animated a series of seminal workshops.

In 2005, this collaboration with NIA participated to a long post-doctoral visit of S. Boldo, and a short visit of F. Cháves and F. Kirchner ( École Polytechnique ). In the meantime, G. Melquiond worked with colleagues from University of California at Berkeley and presented a seminar at Intel in Portland.

Collaboration ATIPS–LIP

Keywords : hardware arithmetic operator, cryptography, hash function, efficient implementation.

Participants : J.-L. Beuchat, A. Tisserand, N. Veyrat-Charvillon.

J.-L. Beuchat and A. Tisserand have been invited one month (June 2005) in the ATIPS laboratory, University of Calgary, Canada, to work on the efficient implementation of arithmetic operators (dedicated modular arithmetic support in current architecture and FPGA implementations).

R. Glabb (ATIPS) has been invited one month (September 2005) in Lyon and N. Veyrat-Charvillon has been invited one month (October 2005) in Calgary on the implementation of an efficient multi-mode operator computing all algorithms of the SHA-2 family of hash functions into a single architecture.


Logo Inria