Trade-offs in the design of cross-disciplinary software systems
As researchers we are often faced with the difficult and demanding task of preparing models, and their computer implementations, for decision making, or, more recently, for integrated assessment. Such assessment often involves large scale problems, where the decisions to be made can deeply affect the environment, the social context and the economic background of regions and even nations. Yet, we face the grim reality that a model is a focused representation of the world, and it is always a result of several compromises in terms of details and structure, leading to trade-offs in terms of complexity, flexibility and performance. This trade-off becomes an essential design property. We often wish our models to be as simple as possible, balancing transparency, understandability and level of detail. Now, we are involved in the SEAMLESS project, an EU FP6 Integrated Project, aims at generating an integrated framework of computer models. This framework can be used for assessment of how future alternative agricultural and environmental polices affect sustainable development in Europe. Thus, we are designing a cross disciplinary software system to deal with different simulation domains. In this, we need to take care of many differences between the different modeling societies.In this article we describe all the risks we have identified as associated to our architecture centric approach and how we dealt with them. This article describes the design of the modeling framework for SEAMLESS.
T. van der Wal, R. Knapen, M. Svensson, I. N. Athanasiadis, A. E. Rizzoli, Trade-offs in the design of cross-disciplinary software systems, MODSIM 2005 Intl Congress on Modelling and Simulation, pg. 732-737, 2005, Modelling and Simulation Society of Australia and New Zealand.
You might also enjoy (View all publications)
- Introducing digital twins to agriculture
- Location-specific vs location-agnostic machine learning metamodels for predicting pasture nitrogen response rate
- Machine learning for large-scale crop yield forecasting