English

Probabilistic Error Analysis for Inner Products

Numerical Analysis 2019-06-26 v1 Numerical Analysis

Abstract

Probabilistic models are proposed for bounding the forward error in the numerically computed inner product (dot product, scalar product) between of two real nn-vectors. We derive probabilistic perturbation bounds, as well as probabilistic roundoff error bounds for the sequential accumulation of the inner product. These bounds are non-asymptotic, explicit, and make minimal assumptions on perturbations and roundoffs. The perturbations are represented as independent, bounded, zero-mean random variables, and the probabilistic perturbation bound is based on Azuma's inequality. The roundoffs are also represented as bounded, zero-mean random variables. The first probabilistic bound assumes that the roundoffs are independent, while the second one does not. For the latter, we construct a Martingale that mirrors the sequential order of computations. Numerical experiments confirm that our bounds are more informative, often by several orders of magnitude, than traditional deterministic bounds -- even for small vector dimensions~nn and very stringent success probabilities. In particular the probabilistic roundoff error bounds are functions of n\sqrt{n} rather than~nn, thus giving a quantitative confirmation of Wilkinson's intuition. The paper concludes with a critical assessment of the probabilistic approach.

Keywords

Cite

@article{arxiv.1906.10465,
  title  = {Probabilistic Error Analysis for Inner Products},
  author = {Ilse C. F. Ipsen and Hua Zhou},
  journal= {arXiv preprint arXiv:1906.10465},
  year   = {2019}
}
R2 v1 2026-06-23T10:02:56.786Z