The usage of computers for solving real-world scientific and engineering problems is becoming more and more important. Nevertheless, the current development practice for scientific software is rather primitive. One of the main reasons for this is the lack of good high-level tools. Most scientific software is still being developed the traditional way in Fortran, especially in application areas such as machine element analysis, where complex non-linear problems are the norm.

In this thesis we present a new approach to the development of software for scientific computing and a tool which supports this approach. A key idea is that mathematical models should be expressed in a high-level modeling language which allows the user to do this in a way which closely resembles how mathematics is written with pen and paper. To facilitate the structuring of complex mathematical models the modeling language should also support object-oriented concepts.

We have designed and implemented such a language, called ObjectMath, as well as a programming environment based on it. The system has been evaluated by using it in an industrial environment. We believe that the proposed approach, supported by appropriate tools, can improve productivity and quality, thus enabling engineers and scientists to solve problems which are too complex to handle with traditional tools.

The thesis consists of five papers (four of which have previously been published in the proceedings of international conferences) dealing with various aspects of the ObjectMath language and programming environment.