CTADEL

CTADEL: Code-generation Tool for Applications based on Differential Equations using high-level Language specifications. After having implemented and evaluated several parallel versions of a forecast routine of an operational numerical weather forecast system, the idea emerged to automatically generate efficient versions of this routine from a high-level description consisting of the partial differential equations that govern the atmosphere. This was a challenging idea to be applied in numerical weather forecasting, because it was not clear whether it would be possible to generate codes that are comparable in efficiency to the “best” hand-written codes. Furthermore, the development of a code generation system for this task requires a combination of skills in the diverse areas of artificial intelligence, computer algebra systems, parallel programming, compilers, numerical mathematics, and computational physics. The first prototype version of the CTADEL system “Code-generation Tool for Applications based on Differential Equations using high-level Language specifications” demonstrated an improved performance of the generated codes compared to the hand-written codes. This inspired me to fully redesign the system in order to create a system for solving a wider scope of problems.

This software is also peer reviewed by journal TOMS.