using NAG software - Numerical Algorithms Group
using NAG software - Numerical Algorithms Group
using NAG software - Numerical Algorithms Group
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
<strong>NAG</strong> Financial Mathematics Day<br />
MBS, University of Manchester<br />
John Holden<br />
Marcin Krzysztofik<br />
Craig Lucas<br />
Robert Tong<br />
David Sayers<br />
<strong>NAG</strong><br />
Rose Baker<br />
Salford University<br />
Michael Croucher,<br />
University of Manchester<br />
Experts in numerical algorithms<br />
and HPC services<br />
2 nd Feburary 2011
<strong>NAG</strong> Financial Mathematics Day: Part 1<br />
• Welcome<br />
<br />
John Holden<br />
• <strong>NAG</strong> for Mathematical Finance<br />
<br />
John Holden<br />
• <strong>NAG</strong> Software at the University of Manchester,<br />
User stories, <strong>NAG</strong> & Python<br />
<br />
Michael Croucher<br />
• Using the <strong>NAG</strong> C Library with C++ - getting<br />
started<br />
<br />
Marcin Krzysztofik<br />
<strong>Numerical</strong> Excellence<br />
2
<strong>NAG</strong> for Mathematical Finance<br />
John Holden<br />
Experts in numerical algorithms<br />
and HPC services
Agenda<br />
• <strong>NAG</strong> introduction<br />
• <strong>NAG</strong> and University of Manchester<br />
• <strong>NAG</strong>’s <strong>Numerical</strong> Libraries and Toolboxes<br />
• Recent Developments<br />
<br />
<br />
<br />
New Library Content (functionality)<br />
<strong>NAG</strong> Library for .NET<br />
<strong>NAG</strong> Routines for GPUs<br />
..<br />
• <strong>NAG</strong> Student Prize<br />
<strong>Numerical</strong> Excellence<br />
4
<strong>NAG</strong> Profile – 1<br />
What We Do<br />
•Mathematical, statistical, data analysis components<br />
•3D Visualization<br />
•High performance computing user support<br />
Users<br />
•Professional/ISV developers<br />
•Quants/analysts/researchers/modellers<br />
Strengths<br />
•Algorithmic content<br />
•Software engineering expertise<br />
5
<strong>NAG</strong> Profile – 2<br />
Customers<br />
•ISVs (finance, science, retail analytics, etc)<br />
•Finance<br />
•Pharmaceuticals & life sciences, Energy<br />
•Academia & Research<br />
Offices<br />
•UK (Oxford, Manchester)<br />
•US (Chicago)<br />
•Japan (Tokyo)<br />
•Greater China (Taipei)<br />
Origins<br />
•Started 1970 from six British universities<br />
•Commercial, Not-for-profit (no shareholders)<br />
•~100 staff, >50% technical, 25 PhDs in various disciplines<br />
6
<strong>NAG</strong> and University of Manchester<br />
The content of this slide has been changed…<br />
Please contact<br />
http://www.applications.itservices.manchester.ac.uk/<br />
Or contact <strong>NAG</strong> www.nag.co.uk / support@nag.co.uk<br />
<strong>Numerical</strong> Excellence<br />
7
Why Use <strong>NAG</strong> Maths Libraries and Toolboxes<br />
• Global reputation for quality – accuracy, reliability<br />
and robustness…<br />
• Extensively tested, supported and maintained code<br />
• Reduce development time<br />
• Concentrate on your key areas<br />
• Components<br />
<br />
<br />
Fit into your environment<br />
Simple interfaces to your favourite packages<br />
• Regular performance improvements!<br />
<strong>Numerical</strong> Excellence<br />
8
<strong>NAG</strong> development philosophy<br />
• First priority: accuracy<br />
• Second priority: performance<br />
<br />
How fast do you want the wrong answer<br />
• <strong>Algorithms</strong> chosen for<br />
<br />
<br />
<br />
<br />
<br />
usefulness<br />
robustness<br />
accuracy<br />
stability<br />
speed<br />
<strong>Numerical</strong> Excellence<br />
9
<strong>NAG</strong> Library and Toolbox Contents<br />
• Root Finding<br />
• Summation of Series<br />
• Quadrature<br />
• Ordinary Differential<br />
Equations<br />
• Partial Differential Equations<br />
• <strong>Numerical</strong> Differentiation<br />
• Integral Equations<br />
• Mesh Generation<br />
• Interpolation<br />
• Curve and Surface Fitting<br />
• Optimization<br />
• Approximations of Special<br />
Functions<br />
• Dense Linear Algebra<br />
• Sparse Linear Algebra<br />
• Correlation & Regression<br />
Analysis<br />
• Multivariate Methods<br />
• Analysis of Variance<br />
• Random Number Generators<br />
• Univariate Estimation<br />
• Nonparametric Statistics<br />
• Smoothing in Statistics<br />
• Contingency Table Analysis<br />
• Survival Analysis<br />
• Time Series Analysis<br />
• Operations Research<br />
<strong>Numerical</strong> Excellence<br />
10
<strong>NAG</strong> Data Mining Components<br />
• Data Cleaning<br />
<br />
<br />
Data Imputation<br />
Outlier Detection<br />
• Data Transformations<br />
<br />
<br />
Scaling Data<br />
Principal Component Analysis<br />
• Cluster Analysis<br />
<br />
<br />
k-means Clustering<br />
Hierarchical Clustering<br />
• Classification<br />
<br />
<br />
<br />
Classification Trees<br />
Generalised Linear Models<br />
Nearest Neighbours<br />
• Regression<br />
<br />
<br />
<br />
<br />
<br />
Regression Trees<br />
Linear Regression<br />
Multi-layer Perceptron Neural<br />
Networks<br />
Nearest Neighbours<br />
Radial Basis Function Models<br />
• Association Rules<br />
• Utility Functions<br />
<br />
To support the main functions<br />
and help with prototyping<br />
<strong>Numerical</strong> Excellence<br />
11
Partnerships with leading academics<br />
• University of Aachen<br />
• University of Birmingham<br />
• Illinois Institute of<br />
Technology<br />
• Goethe-Universitaet<br />
Frankfurt am Main<br />
• K U Leuven<br />
• University College<br />
London<br />
• University of Manchester<br />
• New York University<br />
• Osaka City University<br />
• University of Oxford<br />
• Universita di Roma I<br />
• Shanghai University<br />
• Stanford University<br />
• University of Strathclyde<br />
• National University of<br />
Singapore<br />
• University of Tennessee<br />
• Tsinghua University<br />
• University of Vienna<br />
• University of Warwick<br />
<strong>Numerical</strong> Excellence<br />
12
Don’t take our word for it….<br />
University Professors speeds up their optimisation…<br />
<strong>Numerical</strong> Excellence<br />
13
More world leading experts enjoy <strong>using</strong><br />
<strong>NAG</strong>
Who uses <strong>NAG</strong> – some examples<br />
Retail<br />
analytics<br />
• Embedded in ISV<br />
pricing &<br />
promotion<br />
optimization<br />
products<br />
Finance<br />
• 150+ major firms<br />
– mutual funds,<br />
investment<br />
managers, fixed<br />
income, hedge<br />
funds<br />
Energy<br />
• Exploration and<br />
production –<br />
seismic image<br />
processing<br />
Life Sciences<br />
• Pharmaceuticals<br />
– modeling and<br />
visualization<br />
15
Use of <strong>NAG</strong> Software in Finance<br />
• Portfolio analysis / Index tracking / Risk management<br />
<br />
Optimization , linear algebra, copulas…<br />
• Calibration / Economic Scenario modelling<br />
<br />
Optimization<br />
• Derivative pricing<br />
<br />
PDEs, RNGs, multivariate normal, …<br />
• Fixed Income/ Asset management / Portfolio<br />
Immunization<br />
<br />
Operations research<br />
• Data analysis<br />
<br />
Time series, GARCH, principal component analysis, data smoothing,<br />
…<br />
• Monte Carlo simulation<br />
<br />
• ……<br />
RNGs<br />
<strong>Numerical</strong> Excellence<br />
16
<strong>NAG</strong> Libraries Ease of Integration<br />
• C++ (various)<br />
• C# / .NET<br />
• CUDA<br />
• OpenCL<br />
• Visual Basic<br />
• Java<br />
• Borland Delphi<br />
• Python<br />
• F#<br />
• …<br />
• and more<br />
• Excel<br />
• MATLAB<br />
• SciLab<br />
• Octave<br />
• Mathematica<br />
• Maple<br />
• LabVIEW<br />
• R and S-Plus<br />
• SAS<br />
• …<br />
• and more<br />
<strong>Numerical</strong> Excellence<br />
17
Recent Developments from <strong>NAG</strong><br />
• Recent Developments<br />
<br />
New Library Content (functionality)<br />
• My colleagues will cover the following<br />
<br />
<br />
<br />
<br />
<strong>NAG</strong> Libraries and Excel – Marcin<br />
<strong>NAG</strong> Library for .NET – Marcin<br />
<strong>NAG</strong> Toolbox for MATLAB - Craig<br />
<strong>NAG</strong> Routines for GPUs - Robert<br />
<strong>Numerical</strong> Excellence<br />
18
<strong>NAG</strong> Library & Toolbox – New at Mark 22/09<br />
•Global optimization<br />
•Nearest Correlation<br />
Matrix<br />
•Partial Least Squares<br />
Regression Analysis<br />
•Prediction intervals for<br />
fitted models<br />
•Option Pricing<br />
•more Copulas<br />
•EVTs<br />
•more Linear Mixed Effect<br />
Regression<br />
•Fast quantile selection<br />
routine<br />
•Wavelets<br />
•Adoption of LAPACK 3.1<br />
•New RNGs<br />
Scrambled Seq for QMC<br />
Mersenne Twister<br />
Sobol Sequence generator (50,000 d)<br />
<strong>Numerical</strong> Excellence 19<br />
19
<strong>NAG</strong> Financial Maths Student Prizes<br />
•For “projects” <strong>using</strong> <strong>NAG</strong> <strong>software</strong><br />
• Submit a description of the project, its goals and the<br />
use of <strong>NAG</strong> <strong>software</strong><br />
• The Prize<br />
<br />
“Free attendance of finance conference including<br />
contribution to flights and accommodation”<br />
<strong>Numerical</strong> Excellence<br />
21
Competition Entry<br />
Send to: nagmarketing@nag.co.uk<br />
As sponsors of a wide variety of Mathematical<br />
Finance conferences and events we are able to<br />
offer the following prizes:<br />
•Global Derivatives, Trading & Risk Management<br />
April 2011, Paris<br />
•Risk Europe, Belgium<br />
•Quant Europe, London<br />
Other conferences will be announced as prizes are secured.<br />
<strong>Numerical</strong> Excellence<br />
22
Summary<br />
• <strong>NAG</strong> for Quality, World-leading <strong>Numerical</strong><br />
Software Components<br />
<br />
<br />
<br />
accurate, reliable, robust<br />
extensively tested, supported and maintained code<br />
updated for new architectures and new algorithms<br />
<strong>Numerical</strong> Excellence<br />
23
<strong>NAG</strong> Contacts<br />
www.nag.co.uk<br />
Technical Support and Help<br />
support@nag.co.uk<br />
<strong>NAG</strong>News<br />
http://www.nag.com/<strong>NAG</strong>News/Index.asp<br />
<strong>Numerical</strong> Excellence<br />
24