NAGnews 139

Posted on
5 May 2016

In this issue:

NAG Fortran Compiler - new functionality announced

The latest version of the NAG Fortran Compiler, Release 6.1, is now available. The Compiler is renowned for its checking capabilities and detailed error reporting. Release 6.1 has extensive support for both legacy and modern Fortran features, and also supports parallel programming with OpenMP.

New functionality at this release includes:

  • More Fortran 2008 features, specifically:
    • The improved parallelizable do construct, DO CONCURRENT, is supported
    • Bessel functions
    • COMPILER_VERSION and COMPILER_OPTIONS inquiry functions
  • OpenMP 3.1 – fully supported
  • More Fortran 2003 features including:
    • Length type parameters for derived types
    • Rounding-mode descriptors in Input/Output statements
  • New Tool: precision unifier for converting code to use some standard numeric intrinsic functions and to have uniform single, double or quad precision
  • New polish tool options
  • Improved operation of the interface-block generator

The NAG Fortran Compiler, Release 6.1 is available for Linux 32-bit and Linux 64-bit and Apple Mac. The Microsoft Windows version of the Compiler (NAG Fortran Builder) will follow soon.

Many customers will be entitled to use the latest version of the Compiler as part of their NAG software licence agreement. If you’re not sure about your eligibility email us and we’ll check for you.

Multidimensional Improvements to the NAG Riemann Solvers

The following is taken from a recent NAG Blog post by Kevin Olsen, NAG HPC Software Developer.

The NAG Library contains routines for solving the partial differential equations specific to compressible, ideal fluid flow. These equations are generally written in conservation law form where the conserved quantities are mass density, momentum and total energy of the fluid.  This set of equations can be solved using a finite volume technique that considers each conserved variable as a volume average over a finite volume (typically a small cube) and sums the fluxes (flow rates per unit area) computed at the faces surrounding the volume to get the total rate of change of a particular variable for that volume.

Several methods exist to solve this set of coupled equations (e.g. Flux Corrected Transport, ENO and WENO schemes, etc.). I focus here on the Godunov method where, in its simplest form, the fluxes are computed by solving a Riemann problem at the interface between two cells in the computational mesh. In such methods, appropriately limited (I do not address limiting procedures here, but see, for example [Toro, 1999]) states left and right of a cell interface are computed. These states are assumed to be constant, discontinuous states, i.e. a Riemann problem. The Riemann problem does have an exact solution which can be used to finally compute the fluxes at the cell interface. The NAG Library possesses a routine for computing the exact solution and returning the fluxes to the user in one spatial dimension (D03PX). The NAG Library also possesses three other approximate "Riemann Solvers": Roe’s solver (D03PU), Osher’s solver (D03PV), and the HLL solver (D03PW). Descriptions of all these algorithms can be found in [Toro, 1999]. To make these routines more generally useful they should be able to handle the full set of three dimensional equations and this blog discusses the modifications to these routines to accomplish this.

Read Kevin’s blog post here.

Students at The University of Hong Kong receive their NAG Scholarships

As part of NAG’s Academic Outreach Programme NAG is delighted to fund three Scholarships in Computational Statistics at The University of Hong Kong.

The three most recent winners of the Scholarships were awarded their prizes at the Master of Statistics Graduation Dinner. Congratulations to the highly deserving Students Haoyun Chen, Jiawen Li, and Johnathan Sze Nok Lo on your outstanding work. We look forward to hearing of your future successes.

Haoyun Chen

Johnathan Sze Nok Lo

Jiawen Li

The NAG Scholarships in Computational Statistics are awarded to final-year Master of Statistics candidates who have declared either the Risk Management theme or Statistical Informatics theme, based on their examination results in the theme.

The Code Contributors - future proof your algorithmic code with the NAG Library

Collaboration is at the heart of NAG and the NAG Library, with hundreds of algorithms having been contributed by people all over the world. Last year we interviewed some past and present ‘code contributors’ to learn about the process and what it means to them.

The interviews feature in this piece, published in Scientific Computing. We continue to publish snippets from the interviews to highlight this important element of the NAG Library and encourage future contribution and collaborations.

Here we ask some of our code contributors what they see as the key benefit for the end user from accessing their methods via the NAG Library.

Rebecca Killick, Lecturer in Statistics, University of Lancaster
"Accessing my work via the NAG Library gives it a ‘seal of approval’ and assures people that the code is accurate and correct."

Maurice Cox, National Physical Laboratory, UK
"Key benefits are robustness of the Chebyshev and B-spline bases, and the ability to have polynomials degrees when necessary in the hundreds and an unlimited number of spline knots. Backward error-analyses exist for most of my contributed code: the solution provided is the exact solution of a closely neighbouring problem."

Fred Hickernell, Professor, Illinois Institute of Technology
"The code is highly reliable. The NAG Library contains some of the best algorithms available."

If you’d like to learn more about contributing code to the NAG Library we would be delighted to hear from you. A list of NAG’s many contributors can be seen here – would you like to belong to this group? Email us with thoughts or questions 

The interviews feature in this piece.

Algorithmic Differentiation Presentations from QuanTech Conference

NAG recently attended the WBS QuanTech Conference in London. AAD (Adjoint Algorithmic Differentiation) dominated many of the talks there. NAG’s Jacques Du Toit and Collaborator Uwe Naumann headlined with presentations showing the results of their latest research which will feed into future versions of NAG’s AD software tool, dco/c++. For those not able to attend you can view their talks here.

If you’re new to AAD for Computational Finance, review our popular paper which includes C++ source code examples.

Out & About with NAG

Come and see us at various conferences and events over the next few months.