 |
| Titel |
Object-oriented numerics with FOSS: comparing PyPy & NumPy, GCC/Clang & Bitz++ and Gfortran |
| VerfasserIn |
Dorota Jarecka, Sylwester Arabas, Maciej Fijalkowski, Anna Jaruga, Davide Del Vento |
| Konferenz |
EGU General Assembly 2013
|
| Medientyp |
Artikel
|
| Sprache |
Englisch
|
| Digitales Dokument |
PDF |
| Erschienen |
In: GRA - Volume 15 (2013) |
| Datensatznummer |
250083503
|
|
|
|
|
|
| Zusammenfassung |
| Employment of object-oriented programming (OOP) techniques may help to improve code readability, and hence its auditability and maintainability - both being arguably crucial for scientific software.
OOP offers, in particular, the possibility to reproduce in the program code the mathematical "blackboard abstractions" used in the literature. There exist a number of free and open-source tools allowing to obtain this goal without sacrificing performance.
An OOP implementation of the MPDATA advection algorithm used as a core of weather, ocean and climate modelling systems will serve as an example for briefly highlighting some relevant recent FOSS developments including:
- NumPy support in the PyPy just-in-time compiler of Python.
- the Blitz++ library coupled with the C++11 support in GCC and Clang;
- support for OOP constructs from Fortran 2003/2008 in GFortran;
A brief overview of other performance-related packages for Python like Numba and Cython will be also given.
This poster will describe and extends key findings presented in http://arxiv.org/abs/1301.1334 |
|
|
|
|
|
|