Home Page

TutorialsDate and LocationTutorials will take place on August 6, 2000, immediately preceding the ISSAC 2000 conference, and they will be held at St Andrews University.Spaces in the tutorials are limited and will be assigned in the order that registrations are received. A tutorial will be held only if a minimum number of registrations is received. If a tutorial must be cancelled, registrants will be informed by July 18. Please refer to our cancellation and refund policy for more information. ScheduleTutorial 1: Martin Escardo, "Exact Numerical Computation", 9:0012:00. View abstract.Tutorial 2: Alexander Hulpke, "Using GAP", 13:0016:00. View abstract. Tutorial 3: Mohamed Omar Rayes, "Application of Computer Algebra in Mathematics Education", 16:3019:30. Sponsored by Texas Instruments. View abstract.
FeesPlease note that registration is required for all tutorials.
Tutorial 1  31 GBP (student rate: 22 GBP)
AbstractsTutorial 1: Exact Numerical Computation
Martin Escardo
This tutorial will present the theory and practice of exact numerical computation. The usual approach to numerical computation consists of approximating the set of real numbers by a large but finite set of real numbers, usually floatingpoint numbers. It is wellknown that, in the absence of careful numerical analysis, the obtained results tend to be unreliable due to the introduction, propagation and accumulation of roundoff errors. A simple example is given by iterates of the logistic map f(x)=4x(1x). Starting from, say, 0.671875, simple and double precision computations in the IEEE standard produce 0.934518 and 0.757154 as values of the 60th iterate, while the correct value rounded to six decimal digits is 0.315445. In commercially available computer algebra systems, one can specify an arbitrarily large precision for numerical computations. But, once chosen, the precision is fixed, and roundoff errors still occur. It is usually (but not always!) the case that a sufficiently large precision guarantees that the results will be correct to within a specified degree of accuracy. However, it is difficult or unfeasible to predict the necessary precision in advance. Moreover, the precision may be unacceptably large. Instead of approximating a number by a fixedlength finite numeral, one can realize a number by a "generator". A real number generator is a procedure that successively generates as many digits as one is willing to wait for (and, of course, as the resources of the computer allow). Thus, operations on numbers are realized by operations on number generators. For instance, addition is realized by a process that, given generators for numbers x and y, produces a generator for the number x+y. In this case, computations are usually performed by demand; if one wants 5 correct digits of y=g(f(x)), one may need 20 digits of the intermediate computation f(x), and 7 digits of the input x. In this way, roundoff errors are completely avoided. Exact numerical computation on the reals via potentially infinite representations was proposed by Wiedmer in 1980 and further investigated by a number of authors, including Boehm, Cartwright and Vuillemin. But the idea goes back to constructive mathematics, with the work of Brouwer in 1920 and Bishop in 1967, and recursion theory, with the work of Turing, Rice, Grzegorczyk, Kleene, Vesley, Pourel, Richards, Weihrauch and Kreitz, among others. In constructive analysis, the trichotomy law "x<a or x=a or x>a" fails. This is reflected in the fact that the (in)equality relations are undecidable when numbers are presented by generators. This is usually overcome by the constructive law "a<b => x<b or x>a". From a practical point of view, this means that algorithms for finiteprecision arithmetic usually don't give rise to algorithms for potentiallyinfiniteprecision arithmetic by a simple change of underlying representation of numbers.
Tutorial 2: Using GAPAlexander Hulpke
The aim of this tutorial is to give an introduction to the system GAP for mathematicians and computer scientists who want to use it in their own research, be it as a basis for implementations, or to just utilize the builtin functionality. GAP is a free system for computational discrete mathematics which has found use in many areas beyond its initial inception as system for group theory. Its interpretative nature and convenient list functions make it a comfortable environment for prototyping, while the implementation of timecritical subroutines in the Clanguage kernel maintains a satisfactory performance. Of particular interest might be its type system, which does not enforce rigid types, but aims to reflect the known mathematical (and implementational) properties of objects and seems to suit computer algebra very well. Topics include:
The tutorial will only assume prior usage of a computer and knowledge of (any) programming language, it will not assume familiarity with any particular area of mathematics. Tutorial 3: Application of Computer Algebra in
Mathematics Education
