Wide-Area Distributed Problem Solving (DIPSO)

The DIPSO project is aiming to create a Grid-enabled environment for supporting the early stages of complex problem solving. The environment will provide tools to facilitate decision-makers to better formulate their problems through the utilisation of a range of search and optimisation techniques. Such techniques explore a variety of high quality solutions prior to the decision-maker committing to any one in particular. The environment will consist of a workflow engine, to enable the decision-maker to specify the problem solving process. Each stage within this process will be supported via a toolbox containing machine learning algorithms and data input/output components. The Triana workflow engine is being extended to support this.

The environment is to be evaluated on a series of test functions in the first stage. Subsequently, two design studies will be undertaken with the industrial partners. SEA Group will be providing a parametric model related to the conceptual design of remotely controlled sub-sea vehicles. Design parameters may include engine type, vehicle dimensions, power supply etc. Solutions will be evaluated against performance across possible missions that the described vehicles may be required to undertake. The objective is to identify the compromise design that best satisfies perceived requirements. EVOTEC OAI, the other industrial partner, will be providing empiric data sets relating to various criteria in a bioinformatics design application. Response surfaces relating to such data sets will be generated by the tools provided and subsequently passed to the search and optimization processes. High quality solutions will be returned to the decision-maker to assist in establishing appropriate direction in terms of further in-house design processing.