Automation Algorithms

 This appendix evaluates optimization algorithms to provide a background to solver selection in Streamline. The discussion is largely independent of present Streamline: it clarifies options not just for application tailoring as discussed in the thesis, but also for later revisions and wholly different adaptive systems. Understanding of the option space is needed, because optimization models are perpetually amenable to refinement and even relatively minor changes to a model may cause it to fall outside a specific problem class, necessitating the replacement of the solver algorithm. We treat this topic in such detail because, even though decision support is fairly novel in systems design today, we expect it to become increasingly commonplace.