Scientific computing with algebraic and generative abstractions for geophysical problems

Computational modeling using partial differential equations (PDEs) is becoming an important tool for industry and research. The traditional methods for developing such models are time consuming and expensive. Newer methods exploting object oriented and ge nerative techniques are being developed, but there is a considerable sceptisiscm in the user communities against the new generation of abstraction based software. One reason for this hesitation is the need for run-time speed, and it is generally assumed t hat using higher levels of abstraction in the code entails a high run-time overhead. In this project we will do a side by side comparison of geophysical solver software developed using traditional methods and software developed using algebraic and gener ative SAGA technologies. An important research topic is then to develop the SAGA tool set to achieve comparable run-time properties while maintaing the SAGA development effectiveness.