NAGnews 149

In this issue:


SMP enabled NAG C Library now available at Mark 26

The NAG C Library, Mark 26 is now optimized for use on Symmetric Multi-Processors (SMP) systems. This is the first time that NAG has enabled C Library routines for SMP, and we look forward to hearing of the performance gains that it brings to existing and new users of the Library.

Existing NAG Library users might be entitled to use the NAG C Library for SMP as part of their licence agreement. Do get in touch if you are interested in using it and we'll check your eligibility.

Mark 26 of the NAG Library brings a host of new functionality to our users: new NAG Optimization Modelling Suite including Interior Point Method for Nonlinear Optimization and Semidefinite Programming. Learn more here.


Latest NAG Student Winner Shines a Light on the Calibration of Stochastic Local Volatility Models

It was great to meet one of our latest Student Prize winners at the QuanTech London conference recently. Maartyn Wyns who studies at the University of Antwerp, won a free pass to QuanTech for his work on 'A Finite Volume - Alternating Direction Implicit Approach for the Calibration of Stochastic Local Volatility Models'. View his paper and poster.


Calibration of stochastic local volatility (SLV) models to their underlying local volatility model is often performed by numerically solving a two-dimensional non-linear forward Kolmogorov equation. We propose a novel finite volume (FV) discretization in the numerical solution of general 1D and 2D forward Kolmogorov equations. The FV method does not require a transformation of the PDE. This constitutes a main advantage in the calibration of SLV models as the pertinent PDE coefficients are often nonsmooth. Moreover, the FV discretization has the crucial property that the total numerical mass is conserved. Applying the FV discretization in the calibration of SLV models yields a non-linear system of ODEs. Numerical time stepping is performed by the Hundsdorfer-Verwer ADI scheme to increase the computational efficiency. The non-linearity in the system of ODEs is handled by introducing an inner iteration. Ample numerical experiments are presented that illustrate the effectiveness of the calibration procedure.

Image removed.


Webinar: Improving Application Performance on the Intel Xeon Phi

Webinar: Improving Application Performance on the Intel Xeon Phi Processor
7 day webinar series (2 hour sessions) 3-11 July 2017

This series of 2 hour theory and practical webinars delivered over 7 days will teach the fundamental skills needed to achieve optimum performance on the Intel® Xeon Phi™ Processor.

NAG HPC Engineers will show how to achieve application performance gains on the Intel Xeon Phi Processor through the use of OpenMP; this entails fully utilizing all cores as well as efficient use of its SIMD vectorization capabilities.

By the end of this course, attendees will know the Intel Xeon Phi Processor and what applications can best leverage it. They will also know how to use OpenMP to utilize multicore parallelism as well as vectorization, and they will know how to further optimize already-parallel applications to even more efficiently utilize the Intel Xeon Phi Processor and maximize performance. More information and register.


Leslie Fox Blue Plaque Unveiling Ceremony and The Leslie Fox Prize for Numerical Analysis 2017

Professor Leslie Fox played a pivotal role in establishing the Numerical Algorithms Group (NAG) and supported our organization throughout his lifetime. While he was Director of the University of Oxford Computing Laboratory and Professor of Numerical Analysis, he and some of his fellow Oxford associates collaborated with others from Manchester University and NAG Founder Director, Dr Brian Ford from Nottingham University, to establish NAG.

Two important events are taking place over the next couple months that honour Professor Fox. The first is the 18th Leslie Fox Prize Meeting at Strathclyde University on the 26th June 2017. Then on the 18th July 2017 a Blue Plaque in honour of Professor Fox will be unveiled at Dewsbury Railway Station at 11.30am.

Read more about our association with Professor Fox here.


Featured Technical Poster

High Performance Tape-Free Adjoint AD for C++11 - Introducing NAG dco/map, a cross-platform, accelerator ready AAD tool

Making Adjoints of C++ Codes - the C++ language is so complex that no AD compilers can handle it. To get an adjoint, we must write it by hand or use an operator overloading AAD tool. Handwritten adjoints are tedious to write and maintain, while operator overloading tools have a tape which must be managed. The tape is particularly problematic on accelerators since each thread (or potentially each AVX lane) needs its own. On GPUs this is just impractical.

View the poster.


Out & About with NAG

Stop press: We have a number of significantly reduced passes for the Fixed Income Conference in Florence this October 2017. Contact us if you would like to receive one. These passes will be allocated on a first come, first served basis.

Training Courses & Webinars

Webinar: Improving Application Performance on the Intel Xeon Phi Processor - 7 day webinar series (2 hour sessions)
3-11 July 2017

Workshop: Fortran Modernization Workshop, Universitat Politecnica de Catalunya
24-26 July 2017

Webinar: Getting Performance from OpenMP on NUMA Architectures
3 August 2017

NAG-TACC Institute 'HPC for Managers' Summer School
11-15 September 2017

Exhibitions, Conferences and Trade Shows

ARPM (Advanced Risk and Portfolio Management) Bootcamp
14-19 August 2017

SEG International Exposition & 8th Annual Meeting
24-29 September 2017