Vinsamlegast notið þetta auðkenni þegar þið vitnið til verksins eða tengið í það: http://hdl.handle.net/1946/32235
One of the main problems when dealing with simulations in a large environment is the amount of resources required to compute all of the events that are happening. Since it is too costly to simulate them all, the developer must consider ways to reduce the computation at the expense of the player experience, by giving up some details or stopping the simulation of all the processes that are not observable.
The aim of the project is to develop an AI system that can determine which are the most important elements of a simulation and then compute new models of the simula- tion at higher levels of abstraction. Given a fully detailed model of an environment, our solution can automatically generate an abstract model that can be used to approximate the detailed model, but at lower computational cost.