NAGNews 96

Welcome to NAGNews

Issue 96, 08 April 2011

In this issue


Over 1000 people now trained by NAG's HPC Team


Part of NAGs Computational Science and Engineering (CSE) role with the UKs national supercomputing facility, HECToR, involves a dedicated team of HPC experts delivering a range of training courses to help the users optimize their numerical code for the supercomputer.

The HPC training programme covers topics ranging from getting started on the system to improving parallel application performance and scalability on distributed memory machines. Although offered primarily to HECToR users, the courses can be also be attended by the wider community. Training courses are held at NAG's offices in Oxford and Manchester, or on-site at universities. Attendance is free for HECToR users and for any other researchers whose work falls under the remit of the funding research councils EPSRC, NERC and BBSRC.

David Edmunds of Imperial College, who attended the recent course on Fortran 95, hosted at King's College London is the 1,000 person to be trained by the NAG HPC team.

David will receive a copy of the newly-published book 'Modern Fortran Explained - Numerical Mathematics & Scientific Computation' by Michael Metcalf, John Reid and NAGs Malcolm Cohen - three experts in the field, two of whom have actively contributed to Fortran 2008. This is a complete and authoritative description of Fortran in its modern form. It is intended for new and existing users of the language and for all those involved in scientific and numerical computing.

For full detail of upcoming courses see http://www.hector.ac.uk/cse/training/.


NAG Extends its Support for Calling its Numerical Libraries from Excel


NAGs technical experts have recently updated the "NAG and Excel" web page with new examples and VBA header files for the latest releases of the NAG Libraries. Now we support not only VBA6 but also Excel 2010 (64 bit) with VBA7.

For more details please visit http://www.nag.co.uk/numeric/nagandexcel.asp


Performance of Atmospheric Chemistry Simulations (GLOMAP/TOMCAT) Enhanced by NAG's HECToR Team


HPC experts from NAG, working under NAGs Computational Science and Engineering (CSE) support service for HECToR, the UKs national academic supercomputing facility, have restructured key parts of the GLOMAP-mode TOMCAT application and enhanced multicore performance allowing researchers to achieve four-fold reduction in runtimes, thus enabling new science and higher resolutions.

Dr Graham Mann (University of Leeds), who was the Principal Investigator of the project and also a major user of GLOMAP/TOMCAT, estimated that users of GLOMAP-mode consumed around 3.2 million AUs (allocation units) of XT4 resources over a 15 month period on HECToR with a notional cost of £44,000.

When extrapolated across future research using GLOMAP-mode TOMCAT on HECToR and other supercomputers, the optimizations could deliver significant cost savings. The key result for researchers however, is the reduction in CPU-time per model timestep, thus making it possible to do new science. Commenting on the dCSE success, Mann said "The optimization work by NAG means the code runs 15-20% faster in general and, importantly for us, delivered improved scaling to 256 cores on the XT4. This improved scaling means we can achieve significantly reduced runtime per timestep, thus enabling the investigation of scientific scenarios requiring many more timesteps". Mann also noted that "NAGs addition of another level of parallelism to the code via OpenMP will also enable more efficient use of the XT6 (Phase 2b) supercomputer, with its much greater number of cores per node."

Read the full story here http://www.nag.co.uk/Market/articles/glomap_tomcat_dcse_report.pdf.


Toward Adjoint Based Optimization in Computational Finance" Wins Financial Mathematics Student Prize


We are delighted to confirm this years winner is Niloofar Safiran of Software and Tools for Computational Engineering, RWTH Aachen University, Germany.

His paper, Toward Adjoint Based Optimization in Computational Finance discusses how Algorithmic Differentiation (AD) techniques may be applied to the LIBOR market model (LMM) in combination with NAG Optimization routines.

For more information about NAG and AD see http://www.www.nag.co.uk/doc/TechRep/pdf/tr5_10.pdf. Copies of Niloofar's paper will be available at the conference and via the NAG website after the conference. Advanced copies can obtained from nagmarketing@nag.co.uk.

A reminder of last years winning entry can be found on the NAG website http://www.nag.co.uk/about/student_awards/work.asp.


Two New Webinars


Two new 5 minute webinars have been recorded to show some more uses of the NAG Library and the functionality of routines.

The first demonstrates the use of the NAG Library to solve Global Optimization problems. The second shows the advantages of the reverse communication approach with a routine for finding roots of transcendental equations.

Both can be accessed via this web page: http://www.nag.co.uk/Market/webinars.asp.


Presentation Slides from NAG/NVIDIA GPU Conference - Practical Use of GPUs and Numerical Code in Finance


On 3rd March NAG and NVIDIA co-hosted "Practical Use of GPUs and Numerical Code in Finance" at the Illinois Technology Association in Chicago.

The event provided the opportunity to find out more about the advances in the use of GPUs for numerical work, and also provided an overview of NAG and NVIDIA's products. The presentation slides from the event feature below.

Overview of NAG in finance with brief demonstrations of use
Marcin Krzysztofik, NAG
The presentation included product demonstrations relevant to finance. Attendees gained an understanding of how NAGs mathematical and statistical software could be integrated into many different programs and environments, including Excel, MATLAB® (using the NAG Toolbox), C, C++, and C#.

An overview of GPU computing in financial services
John Ashley, NVIDIA
The presentation introduced NVIDIA and the basic concepts of GPU computing in the context of the Financial Services Industry. Areas where customers are successfully seeing dramatic performance improvements were highlighted.

Advances in use of GPUs for financial engineering
Robert Tong, NAG
The presentation gave details of the latest work that is being done to implement NAG methods on GPU processors. It provided the opportunity to see some of the latest performance data that has resulted from this work.


A Sad Passing.....


The death of Heather Wilkinson in February brought to an end the contact between NAG and the Wilkinsons. They contributed so much to the spirit and foundations of NAG.

The idea for what became the Numerical Algorithms Group came in a bath in Bunny, Nottinghamshire in February 1970. But the seminal seed, at least in part, goes back to the visit of Jim Wilkinson to Nottingham University in May 1968. And Jims linear algebra was the back bone of the early NAG Libraries. He was our contributor, mentor and great supporter.

After a brilliant Cambridge student career Jim during WW2 latterly worked at Fort Halstead on problems in the thermodynamics of explosions (involving hyperbolic pdes). There he met Heather Ware, the best mathematician in her year (1943) in London University. Romance blossomed and Jim would cycle with Heather from her home to the Fort. One day he suddenly started pedalling so fast that she could not keep up with him. Arriving several minutes after Jim "she found him, in some embarrassment, explaining that busy on a problem he had forgotten she was there!" Despite this they married in 1945 and lived happily together for over 40 years, until Jim died of a heart attack in 1986.

They were marvellous hosts and loved giving dinner parties. Heather was a great cook and Jim a wine connoisseur and cellar-keeper. When Jim died the wine in their house was rumoured to be worth more than twice the value of their house in Teddington, London. At one dinner party Jim introduced Sven Hammarling and me, which is how Sven came to work at NAG.

Heather opened and named Wilkinson House for us at Jordan Hill in July 1989 and maintained continuing interest and involvement with us for the rest of her life. David, their son, has gifted some of Jims mathematics and computing books to NAG in memory of his parents. We have so much for which to thank Jim, and Heather.

Brian Ford


Recent Blog Posts


Important Events in the HPC Calendar by Andrew Jones
Calling NAG Routines from IDL


Out & About with NAG


INFORMS Conference on Business Analytics & Operations Research
10th-12th April 2011, Chicago
This event is important to those in the area of analytics, operations research, or management science, or just want to know how these time-proven applications and processes can improve your business bottom line. We are exhibiting this year, so please visit us at stand 33 if you are attending.

British Applied Maths Colloquium
11-13 April 2011, University of Birmingham
This event is widely acknowledged as one of the most significant Applied Mathematics conferences in the UK. NAG are pleased to sponsor this event. We will be participating within the Numerical Analysis Mini symposium scheduled for the 11th April, as well as the Software Workshop on the 12th April. We are also exhibiting, so please come and visit us if you are attending.

Global Derivatives and Risk Management
12-15 April 2011, Paris
Global Derivatives & Risk Management is the longest running and largest derivatives & risk management conference in the World. Established in 1994, the conference has built up a reputation as the one must-attend event in the industry. It is the only financial engineering conference to cover such a wealth of derivative asset classes including equity; credit; interest rates; commodities; FX; volatility; inflation and even catastrophe derivatives all under one roof. We are delighted to be sponsoring this event again this year. We will be at our exhibition stand, and look forward to welcoming you, to discuss any requirements you may have. HECToR Training Courses by NAGs HPC Team.

The Practical Use Of GPUs To Improve The Performance Of Numerical Applications
3 March 2011, Illinois Technology Association, Chicago
NAG and NVIDIA are co-hosting a technology event which will detail the use of numerical code on GPGPUs. NAG routines on GPUs will be described to help those who are making use of Graphical Processors to increase the performance of their applications. Full details will be published shortly, however space will be limited. If you'd like to reserve a place please email us.

NAG 'direct award' prize for best financial math project using NAG
Would you like to win a free pass to the prestigious finance event Global Derivatives Trading and Risk Management 2011? If you're a student studying mathematical finance you can enter the NAG 'direct award' prize by sending through details of your mathematical project. Email us for more information.
The most recent winner of a NAG 'direct award' prize is Nicoletta Gabrielli, University of Rome, Masters Student. To learn more about Nicoletta's work visit http://www.nag.co.uk/about/student_awards/work.asp.

One element of NAG's CSE role in HECToR is the provision of HPC related training courses. The courses are provided free of charge to HECToR users and UK academics whose work is covered by the remit of one of the participating research councils (EPSRC, NERC and BBSRC). Others people may attend on payment of a course fee. Please see the HECTOR eligibility page for more details

Here are the courses being held over the next few months.

Multicore
5 April 2011, Imperial College London

Transitioning to the Cray XE6
6 April, NAG's Manchester Office
13 April, NAG's Oxford Office

Co-Array Fortran
7 April, NAG's Manchester Office
13 April, NAG's Oxford Office

For more information on any of the above events visit NAG's ‘Out & About’ webpage


New NAG product implementations


The NAG Toolbox for MATLAB, Mark 22 is now also available for the following platform:

  • x86-31 Linux 32-bit for MATLAB R2010b
  • x86-64 Linux 64-bit for MATLAB R2010b

The NAG C Library, Mark 9 is now also available for the following platform:

  • x86-32- Linux 32 bit using the gcc 4.1.3 compiler

The NAG Library for SMP & Multicore, Mark 22 is now also available for the following platform:

  • IBM Power 6 AIX64 using the XLF v13.1.0.3 compiler

NAGNews - Past Issues


We provide an online archive of past issues of NAGNews. For editions prior to these, please contact us.