Issue 34 - 29 January 2004

In this issue:

  • Top story - IRIS Explorer: Collaborative Computational Steering on the Grid
  • In the News - 'A Sneak Peek at New Fortran Standards'
  • Sales Offer - Buy a NAG C Library or Fortran Library licence and get a Compiler free
  • General Information - 'And express it in numbers' - NAG Colloquium Presentation
  • Sales Offer - Upgrade discounts to GenStat' for Windows' 7th Edition
  • Tips and Hints - Using NAG's Data Mining Components
  • Product News - New implementations

For more information about any of the articles featured in this issue, please do not hesitate to contact us at nagnews@nag.co.uk


Top Story - IRIS Explorer: Collaborative Computational Steering on the Grid


Turning numbers into pictures is usually the easiest way of finding out what they mean. The art of creating figures or scenes from numerical data is called visualisation, and software packages for performing data visualisation have been with us for a long time. One such package is NAG's IRIS Explorer, which can be used to visualise data from a variety of different sources. Users of IRIS Explorer work with applications that are created from software modules by selecting and connecting them together via a simple point-and-click interface, which makes it easy to create and modify programs.

The need for visualisation as a means of understanding data becomes even more important as both the amount of data and computer power increases. In addition, the processes which produce data are often no longer static, but dynamic in nature. Thus, for example, whereas in the past a scientist would want to display the end result of a simulation, now there is an increasing requirement for the real-time display of intermediate results as well. This can be even more helpful if the scientist is able to control the simulation by altering its input whilst the simulation is running, since information gained from the intermediate results can be used to direct the future course of the calculation - a process commonly referred to as computational steering.

To read the entire article written by Jeremy Walton, Senior Visualisation Consultant at NAG please click here http://www.nag.co.uk/IndustryArticles/IRISExplorerArticle.pdf

For more information on NAG's IRIS Explorer please visit http://www.nag.co.uk/welcome_iec.asp or to arrange a free trial please email us at mailto:nagnews@nag.co.uk


In the News - 'A Sneak Peek At New Fortran Standards' - Should finance house quants really care?


Some of the more prominent quantitative analysts working for international banking institutions are strictly Fortran users now and have been for quite some time. There are compelling reasons why Fortran is likely to grow in relevance in the coming years, especially in light of the new standards being finalised by both the U.S and international standards committee charged with formalising Fortran 2003. Evolving Fortran standards also underline the reality that multi-language applications are the new norm, and for good reasons.

To read the entire article by Malcolm Cohen, Principal Technical Consultant at NAG, that was recently published in Financial Engineering News please click here.


Sales Offer - Buy a NAG C Library or Fortran Library licence and get a Compiler free


To celebrate the recent launch of the NAGWare f95 Compiler Release 5.0 we are offering you the chance to purchase the Fortran Library, Fortran 90 Library or C Library and get the corresponding compiler free* - contact us directly to discuss special offers on the NAG HPC Libraries.

The following link http://www.nag.co.uk/NAGNews/SalesOffer.asp shows a list of platform availability for the relevant numerical libraries and the corresponding compiler available. Please contact sales on +44 (0)1865 511245 or email us at mailto:nagnews@nag.co.uk for more information.

*This offer is valid on all orders of the following Fortran, Fortran 90 and C Libraries until 31st March 2004 and is limited to a single copy of a compiler.


General Information - 'And express it in numbers' - NAG Colloquium Presentation


The annual NAG Colloquium, Invited Lecture & AGM was held on 12th December 2003 at the Oxford University Computing Laboratory. The title of the Colloquium was 'Computational Mathematics for Measurement' and in this issue and the next four issues of NAGNews we will feature an abstract from each presentation given and a link to the presentation itself.

The first presentation entitled 'And express it in numbers' was given by Dr Seton Bennett, Deputy Director & Director of International Metrology at the National Physical Laboratory.

Reliable measurements are an essential requirement for industry and research, and they impact on the quality of our lives in a multitude of ways. The result of a measurement is usually expressed as a number times a unit, and our confidence in the result depends on a thorough understanding of both these elements. The SI (Syst'me International) of units is a coherent system based on seven well-defined base units, which by convention are regarded as dimensionally independent. The remaining units of the SI are derived in the form of products of powers of the base units according to the algebraic relationships between the quantities concerned.

The numerical result of a measurement must include a quantitative indication of the quality and reliability of the value given. Without this it is impossible to compare measurements made at different times and in different places, and it is therefore essential to have an accepted procedure for expressing the uncertainty of a measurement. Just as the SI brings coherence to measurement results, so the universal application of a rigorous method for expressing the uncertainty of these results allows them to be properly understood and correctly interpreted. This combination of a coherent system of units and a universally accepted convention for expressing the uncertainty associated with numerical results ensures that measurements are expressed in a complete and unambiguous way.


Sales Offer - Upgrade discounts to GenStat' for Windows' 7th Edition


Are you still using an old version of GenStat? If yes, we are delighted to celebrate the launch of GenStat for Windows 7th Edition by offering unsupported users up to 65% off the current list price and one year free support*.

The offer is as follows:

  • If you are an unsupported user of GenStat 5th Edition upgrade to GenStat 7th Edition and receive a 25% discount off the current list price*
  • If you are an unsupported user of GenStat 6th Edition upgrade to GenStat 7th Edition and receive a 65% discount off the current list price*

To upgrade please email us at mailto:nagnews@nag.co.uk or telephone +44 (0)1865 511245. Key features for the new release can be found on our website at http://www.nag.co.uk/stats/tt_soft.asp

* offer valid for upgrades up to and until 31 March 2004.


Tips & Hints - Using NAG's Data Mining Components (DMC)


Let us suppose that you have been given the task of predicting which customers will call upon a service: for example, it could be an insurance claim or buying a new product. The techniques available to help you in NAG DMC include generalized linear models (e.g., logistic regression, stepwise linear regression), radial basis functions (RBFs), multi-layer perceptrons (MLPs), decision trees and k-nearest neighbour discrimination. Each has particular strengths and weaknesses. One of the main differences is the way they divide up the data space to predict outcome categories using different decision boundaries. But which is most suitable for your needs?

In general, regression methods can handle very large data sets but uses a linear decision boundary method so will have limited flexibility unless you select suitable data transformations. RBFs and MLPs are more flexible in that they use complex non-linear decision boundaries but they are more complex to fit and so cannot handle as large a data set. Decision trees are best suited to data recorded on independent variables. Here the decision boundary is a series of linear dividers. The k-nearest neighbour method is flexible and simple. It does not rely on an underlying parametric model but it does require both training and predicting data to be available at the same time. Here the decision boundaries are local to the individual case.

So depending on what is an appropriate model for dividing up your data space and the amount of data that you are examining you can select the most appropriate technique. For more information on NAG DMC please visit http://www.nag.co.uk/numeric/DR/drdescription.asp or email us at mailto:nagnews@nag.co.uk


Product News - New implementations


NAG is committed to offering new implementations of its broad range of numerical and statistical software components and compilers and tools.

Since the last edition of NAGNews platform availability has increased for the NAGWare f95 Compiler, Release 5.0.

It is now also available for the following platforms:

  • Free BSD
  • Sun SPARC Solaris
  • Intel-32 Linux

For more information about the NAGWare f95 Compiler please click here http://www.nag.co.uk/nagware/np.asp


Managing your subscription


Please feel free to forward this newsletter to colleagues or to post it to your company's Intranet sites.

To subscribe, unsubscribe or learn more about managing your subscription, please see: http://www.nag.co.uk/NAGNews/index.asp.