Friday, April 16, 2010

Simulating Social Behaviour

For a long time I have been thinking of simulating life like creations. A-life seems to be an upcoming topic in Computer Science and elsewhere. I admit I havent read about it much. I have read a little bit about complex systems though. After attending a lot of CS courses and doing projects on AI/ ML and random processes as well as optimization (as part of EE ), I have developed a basic understanding that all AI / ML problems are essentially optimization problems. Each one a unique and a unique way to solve each. Quite Interesting.

What is more interesting is when we see our own life and try to see it in the aspect of an optimization problem. However our own life is not standalone but is influenced by that of others and is (figuratively) a sum total of all the states that are influencing our own state.

Now this is getting a bit vague.

To put it straight forward way I want to design a simulation of artificial life. Where I can spawn a number of individual entities and assign a separate optimization problem to each. Where the goal of each individual entity is to optimize its own problem. Now the deciding factor can be either CPU time, where entities compete for CPU time. But I guess this has already been tried. I think I want to relate them by their optimization functions not just to the environment but by each other also.

Obviously I need to formulate the optimization problem first and than hope that it can actually be solved :).
chow for now !