< back to main site

Publications

Testing algorithms in standards and METROS.

Barker, R M; Cox, M G; Harris, P M; Smith, I M (2003) Testing algorithms in standards and METROS. NPL Report. CMSC 18/03

[img] Text
cmsc18.pdf

Download (480kB)

Abstract

Many mathematical and statistical problems that arise in metrology can be posed in such a way that they possess a unique correct solution. However, algorithms in standards or software libraries, such as METROS, may only provide an approximate solution or the solution to a nearby problem. An approximate or nearby problem is often introduced so as to give a problem that is (more) tractable, i.e., a problem that can be solved or is easier to solve and/or is quicker to solve. Choosing to solve an approximate problem is one reason why software may not deliver correct results. Another reason is that the implementation of software for solving the approximate problem is poor or faulty. Algorithm testing is concerned with understanding the effect of solving an approximate problem by measuring the departure of the solution so obtained from the solution to the correct problem.
A framework is presented for describing the activities of algorithm testing and numerical software testing, and how these activities contribute to the (overall) validation of the software implementation of an algorithm for achieving a given computational aim. A number of approaches are indicated, based on the methodologies of numerical software testing, for undertaking algorithm testing. The application of the methodologies to a particular case study, viz., the problem of fitting a Gaussian peak function to measurement data, is discussed. The application of the methodologies to additional case studies will be the subject of future work.

Item Type: Report/Guide (NPL Report)
NPL Report No.: CMSC 18/03
Subjects: Mathematics and Scientific Computing
Mathematics and Scientific Computing > Software Engineering
Last Modified: 02 Feb 2018 13:16
URI: http://eprintspublications.npl.co.uk/id/eprint/2708

Actions (login required)

View Item View Item