Propagation of Roundoff Errors in Finite Precision Computations: A Semantics Approach

  • Matthieu Martel
Conference paper
Part of the Lecture Notes in Computer Science book series (LNCS, volume 2305)


We introduce a concrete semantics for floating-point operations which describes the propagation of roundoff errors throughout a computation. This semantics is used to assert the correctness of an abstract interpretation which can be straightforwardly derived from it. In our model, every elementary operation introduces a new first order error term, which is later combined with other error terms, yielding higher order error terms. The semantics is parameterized by the maximal order of error to be examined and verifies whether higher order errors actually are negligible. We consider also coarser semantics computing the contribution, to the final error, of the errors due to some intermediate computations.


Numerical Precision Abstract Interpretation Floating-point Arithmetic IEEE Standard 754 


  1. 1.
    F. Chaitin-Chatelin and V. Frayssé. Lectures on Finite Precision Computations. SIAM, 1996.Google Scholar
  2. 2.
    F. Chaitin-Chatelin and E. Traviesas. Precise, a toolbox for assessing the quality of numerical methods and software. In IMACS World Congress on Scientific Computation, Modelling and Applied Mathematics, 2000.Google Scholar
  3. 3.
    P. Cousot and R. Cousot. Abstract interpretation: A unified lattice model for static analysis of programs by construction of approximations of fixed points. Principles of Programming Languages 4, pages 238–252, 1977.Google Scholar
  4. 4.
    P. Cousot and R. Cousot. Abstract interpretation frameworks. Journal of Logic and Symbolic Computation, 2(4):511–547, 1992.zbMATHMathSciNetGoogle Scholar
  5. 5.
    M. Daumas and J. M. Muller, editors. Qualité des Calculs sur Ordinateur. Masson, 1997.Google Scholar
  6. 6.
    D. Goldberg. What every computer scientist should know about floating-point arithmetic. ACM Computing Surveys, 23(1), 1991.Google Scholar
  7. 7.
    E. Goubault. Static analyses of the precision of floating-point operations. In Static Analysis Symposium, SAS’01, number 2126 in LNCS. Springer-Verlag, 2001.Google Scholar
  8. 8.
    E. Goubault, M. Martel, and S. Putot. Concrete and abstract semantics of fp operations. Technical Report DRT/LIST/DTSI/SLA/LSL/01-058, CEA, 2001.Google Scholar
  9. 9.
    E. Goubault, M. Martel, and S. Putot. Asserting the precision of floating-point computations: a simple abstract interpreter. In ESOP’02, 2002.Google Scholar
  10. 10.
    J. R. Hauser. Handling floating-point exceptions in numeric programs. ACM Transactions on Programming Languages and Systems, 18(2), 1996.Google Scholar
  11. 11.
    W. Kahan. The improbability of probabilistic error analyses for numerical computations. Technical report, Berkeley University, 1991.Google Scholar
  12. 12.
    W. Kahan. Lecture notes on the status of IEEE standard 754 for binary floatingpoint arithmetic. Technical report, Berkeley University, 1996.Google Scholar
  13. 13.
    D. Knuth. The Art of Computer Programming-Seminumerical Algorithms. Addison Wesley, 1973.Google Scholar
  14. 14.
    P. Langlois and F. Nativel. Improving automatic reduction of round-off errors. In IMACS World Congress on Scientific Computation, Modelling and Applied Mathematics, volume 2, 1997.Google Scholar
  15. 15.
    V. Lefevre, J.M. Muller, and A. Tisserand. Toward correctly rounded transcendentals. IEEE Transactions on Computers, 47(11), 1998.Google Scholar
  16. 16.
    M. Pichat and J. Vignes. The numerical study of unstable fixed points in a chaotic dynamical system. In IMACS World Congress on Scientific Computation, Modelling and Applied Mathematics,, volume 2, 1997.Google Scholar
  17. 17.
    J. Vignes. A survey of the CESTAC method. In Proceedings of Real Numbers and Computer Conference, 1996.Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 2002

Authors and Affiliations

  • Matthieu Martel
    • 1
  1. 1.CEA - Recherche Technologique LIST-DTSI-SLAGif-Sur-Yvette CedexFrance

Personalised recommendations