The main activities in the reporting period have been the development of a parallel optimizer coupled with a CFD flow solver, for use within the field of optimum aerodynamic design. BAe provided a basic, Genetic Algorithm based, optimization code to solve this problem.
Investigation on methods for integrating parallel library (hill-climbing) optimization software into the GA environment took a proportion of this activity. Options for suitable sequential optimization software have been reviewed. Current NAG library software and techniques based on work by Virginia Torczon will form the initial starting point.
The interface software between the FRONTIER project multi-objective GA based optimizer and the (parallel) NAG library optimization routines (which will produced in the project), has been specified. Convexity testing algorithms, which will be applied to the data generated by the GA in order to determine at which point to switch to a hill-climber, are being developed.
A FRONTIER application to act as a benchmark that can be used for the testing during the development of the three optimizer routines has been delivered to the University of Manchester.