Next: Finding Minimal Polynomials Up: Integer Relation Algorithms Previous: Rational Inputs

## The PSLQ Algorithm

The first integer relation algorithm working reliably on vectors of non-trivial length was discovered by Ferguson and Forcade [26]. After a number of improvements, in 1991 a new algorithm, known as the PSLQ algorithm, was developed by Ferguson [24]. More recently a much simpler formulation of this algorithm was introduced  [25]. The papers [24] and [25] are available on David Bailey's WWW page at

http://science.nas.nasa.gov/Groups/AAA/db.webpage/dbailey.html

Because this algorithm employs a numerically stable matrix reduction procedure, it is free from the numerical difficulties that plague other integer relation algorithms. Furthermore, its stability allows for an efficient implementation with lower run times on average than other algorithms currently in use.

A distinguishing feature of the PSLQ algorithm is that, besides recovering an integer relation (if one exists), it can also produce a negative (non-existence) result of the form if a relation exists, then its Euclidean norm is greater than M (an explicit numerical bound)''. One can show, on using a working precision that is only slightly higher than that of the input data, that these bound results obtained from computer runs are reliable. Results of this type, supporting conjectures about transcendence of certain constants, are to be found in [2,4]. (See also Section 2.3 on minimal polynomials.) For example, Bailey and Ferguson [4] used PSLQ to establish, that if Euler's constant satisfies an integer polynomial of degree 50 or less, then the Euclidean norm of the coefficients must exceed .

In the context of these exclusion bounds it is important to realize that, even for very similar'' numbers, the Euclidean norms of their relations can differ greatly. For example, if x satisfies an integer polynomial of degree d, norm sand k is an integer, then kx satisfies an integer polynomial whose norm may become as high as roughly |k|ds. The lesson here is that, if we can, we should normalize'' our input so that the resulting integer relation is of low Euclidean norm.

Next: Finding Minimal Polynomials Up: Integer Relation Algorithms Previous: Rational Inputs
Agnes Szanto
2000-05-10