Sylvester-Habicht Sequences and Fast Cauchy Index Computation

54Citations
Citations of this article
5Readers
Mendeley users who have this article in their library.

This article is free to access.

Abstract

In this paper we show how Schönhage's strategy for computing continued fractions (Schönhage 1971) can be combined with the theory of sub-resultants (Habicht, 1948; Collins, 1967; Brown, 1971; Brown and Traub, 1971; Loos, 1982; Gonzalez et al., 1990, 1994; Ducos, 1996; Ho and Yap, 1996; Lazard, 1998; Quitté, 1998) in order to compute the Cauchy index of a rational function or the signature of a non-singular Hankel matrix in a fast and also storage efficient way. Over the integers our algorithms have bit complexityO (M(d, σ) ·log(d)) withσ =O(dτ) where M(d,σ ) =O(dσ·log(dσ) ·loglog(dσ)) is Schönhage's bound for multiplication of integer polynomials of degrees bounded by d and bit size bounded by σ in the multi-tape Turing machine model (Schönhage, 1982). Thus our bound isO(d2τ·log(dτ) ·loglog(dτ) ·log(d)).As a byproduct of the necessary analysis we obtain a refinement of the Sub-resultant Theorem. We present a new exact divisibility for sub-resultants in the defective case which extends the formulæ for the non-defective situation in a natural way. We also prove that the size of coefficients in the ordinary remainder sequence is quadratic in d. © 2001 Academic Press.

Cite

CITATION STYLE

APA

Lickteig, T., & Roy, M. F. (2001). Sylvester-Habicht Sequences and Fast Cauchy Index Computation. Journal of Symbolic Computation, 31(3), 315–341. https://doi.org/10.1006/jsco.2000.0427

Register to see more suggestions

Mendeley helps you to discover research relevant for your work.

Already have an account?

Save time finding and organizing research with Mendeley

Sign up for free