Current Computer Algebra Systems Research Articles

The Weierstrass–Durand–Kerner root finder is not generally convergent

Finding roots of univariate polynomials is one of the fundamental tasks of numerics, and there is still a wide gap between root finders that are well understood in theory and those that perform well in practice. We investigate the root-finding method of Weierstrass, also known as the Durand–Kerner-method: this is a root finder that tries to approximate all roots of a given polynomial in parallel. This method has been introduced 130 years ago and has since then a good reputation for finding all roots in practice except in obvious cases of symmetry. Nonetheless, very little is known about its global dynamics and convergence properties. We show that the Weierstrass method, like the well-known Newton method, is not generally convergent: there are open sets of polynomials p p of every degree d ≥ 3 d \ge 3 such that the dynamics of the Weierstrass method applied to p p exhibits attracting periodic orbits. Specifically, all polynomials sufficiently close to Z 3 + Z + 175 Z^3 + Z + 175 have attracting cycles of period 4 4 . Here, period 4 4 is minimal: we show that for cubic polynomials, there are no periodic orbits of length 2 2 or 3 3 that attract open sets of starting points. We also establish another convergence problem for the Weierstrass method: for almost every polynomial of degree d ≥ 3 d\ge 3 there are orbits that are defined for all iterates but converge to ∞ \infty ; this is a problem that does not occur for Newton’s method. Our results are obtained by first interpreting the original problem coming from numerical mathematics in terms of higher-dimensional complex dynamics, then phrasing the question in algebraic terms in such a way that we could finally answer it by applying methods from computer algebra. The main innovation here is the translation into an algebraic question, which is amenable to (exact) computational methods close to the limits of current computer algebra systems.

Multiplicative series, modular forms, and Mandelbrot polynomials

We say a power series ∑ n = 0 ∞ a n q n \sum _{n=0}^\infty a_n q^n is multiplicative if the sequence 1 , a 2 / a 1 , … , a n / a 1 , … 1,a_2/a_1,\ldots ,a_n/a_1,\ldots is so. In this paper, we consider multiplicative power series f f such that f 2 f^2 is also multiplicative. We find a number of examples for which f f is a rational function or a theta series and prove that the complete set of solutions is the locus of a (probably reducible) affine variety over C \mathbb {C} . The precise determination of this variety turns out to be a finite computational problem, but it seems to be beyond the reach of current computer algebra systems. The proof of the theorem depends on a bound on the logarithmic capacity of the Mandelbrot set.

Applying Computer Algebra Systems in Approximating the Trigonometric Functions

We propose numerical algorithms which can be integrated with modern computer algebra systems in a way that is easily implemented to approximate the sine and cosine functions with an arbitrary accuracy. Our approach is based on Taylor’s expansion about a point having a form of k p , k ∈ Z and p = π / 2 , and being chosen such that it is closest to the argument. A full error analysis, which takes advantage of current computer algebra systems in approximating π with a very high accuracy, of our proposed methods is provided. A numerical integration application is performed to demonstrate the use of algorithms. Numerical and graphical results are implemented by MAPLE.

Algorithmic Approach for Formal Fourier Series

The study of trigonometric series has started at the beginning of the nineteenth century. Joseph Fourier made the important observation that almost every function of a closed interval can be decomposed into the sum of sine and cosine functions. This technique to develop a function into a trigonometric series was published for the first time in 1822 by Joseph Fourier. The resulting series is nowadays called Fourier series. Since Fourier’s time, many different approaches to understand the concept of Fourier series have been discovered, each of which emphasizes different aspects of the topic. Some of the more powerful and elegant approaches are based on mathematical ideas and tools that were not available at the time Fourier completed his original work. Although the original motivation was to solve the heat equation for a metal plate, it later became obvious that the same technique could be applied to a wide variety of mathematical and physical problems and has many applications in electrical engineering, vibration analysis, acoustics, optics, signal treatment, image processing, etc. Despite the importance of Fourier series, the method used until now to compute them via computer algebra systems (CAS) is essentially based on the same principle as in Fourier’s time, i.e. by the evaluation of certain integrals. Unfortunately this technique is not completely successful for many functions. Although numeric values of the Fourier coefficients might be available, symbolic values are often not accessible. Modern CAS like Maple or Mathematica can compute such integrals in many cases for a given \({n \in \mathbb{Z}}\). However if one is interested in the Fourier coefficients for all \({n \in \mathbb{Z}}\), then n is considered as a given symbolic variable and such integrals can be computed only in few cases. In this paper we introduce an algorithmic approach to compute those Fourier coefficients, involving differential equations of a particular form, and recurrence equations. This approach extrapolates the computation of the Fourier series for functions for which the computation of Fourier coefficients via the definition is out of reach for current CAS.

Inequalities, assessment and computer algebra

The goal of this paper is to examine single variable real inequalities that arise as tutorial problems and to examine the extent to which current computer algebra systems (CAS) can (1) automatically solve such problems and (2) determine whether students’ own answers to such problems are correct. We review how inequalities arise in contemporary curricula. We consider the formal mathematical processes by which such inequalities are solved, and we consider the notation and syntax through which solutions are expressed. We review the extent to which current CAS can accurately solve these inequalities, and the form given to the solutions by the designers of this software. Finally, we discuss the functionality needed to deal with students’ answers, i.e. to establish equivalence (or otherwise) of expressions representing unions of intervals. We find that while contemporary CAS accurately solve inequalities there is a wide variety of notation used.

Complexity and algorithms for computing Voronoi cells of lattices

In this paper we are concerned with finding the vertices of the Voronoi cell of a Euclidean lattice. Given a basis of a lattice, we prove that computing the number of vertices is a #P-hard problem. On the other hand we describe an algorithm for this problem which is especially suited for low dimensional (say dimensions at most 12) and for highly-symmetric lattices. We use our implementation, which drastically outperforms those of current computer algebra systems, to find the vertices of Voronoi cells and quantizer constants of some prominent lattices.

Automated and readable simplification of trigonometric expressions

Automated simplification of trigonometric expressions is an important problem that hasn’t been completely solved by current computer algebra systems. This paper presents a number of unique prescriptions for the ordering of some trigonometric transformation rules, which have been derived by observing how human experts follow their intuitive rules. We have implemented the procedure in Lisp because of its suitability for formula manipulations and rule-based reasoning systems. Consequently, it can simplify many trigonometric expressions which are even difficult to do by hand, and it achieves much better results for many hard problems than any of Maple, Mathematica, and Maxima do.

Multi‐precision Laplace transform inversion

AbstractFor the numerical inversion of Laplace transforms we suggest to use multi‐precision computing with the level of precision determined by the algorithm. We present two such procedures. The Gaver–Wynn–Rho (GWR) algorithm is based on a special sequence acceleration of the Gaver functionals and requires the evaluation of the transform only on the real line. The fixed Talbot (FT) method is based on the deformation of the contour of the Bromwich inversion integral and requires complex arithmetic. Both GWR and FT have only one free parameter: M, which is the number of terms in the summation. Both algorithms provide increasing accuracy as M increases and can be realized in a few lines using current Computer Algebra Systems. Copyright © 2004 John Wiley & Sons, Ltd.

Differentiation and Integration of Indefinite Summations with Respect to Indexed Variables — Some Rules and Applications

In this paper we present some rules for the differentiation and integration of expressions involving indefinite summations with respect to indexed variables which have not yet been taken into account of current computer algebra systems. These rules, together with several others, have been implemented in MACSYMA and MAPLE as a toolkit for manipulating indefinite summations. We discuss some implementation issues and report our experiments with a set of typical examples. The present work is motivated by our investigation in the computer-aided analysis and derivation of artificial neural systems. The application of our rules to this subject is briefly explained.

An Algorithm for Exact Division

Current computer algebra systems use the quotient-remainder algorithm for division of long integers even when it is known in advance that the remainder is zero. We propose an algorithm which computes the quotient of two long integers in this particular situation, starting from the least-significant digits of the operands. This algorithm is particularly efficient when the radix is a prime number or a power of 2. The computing time of this new algorithm is smaller than the computing time of the classical division algorithm. If the length of the result is much smaller than the lengths of the inputs, then the speed-up may be quite significant, as it is confirmed by practical experiments. Most importantly, however, the new algorithm is better suited for systolic parallelization in a "least-significant digit first" pipelined manner, and therefore it is suitable for aggregation with other systolic algorithms for the arithmetic of long integers and long rationals. We also present applications of this algorithm in integer GCD computation and in division modulo a power of 2.