Heya,
If anyone has some sample code for a flock of birds I may take a crack at that. It is also a behavoir I think would be very interesting to use.
The difference between a "normal" flocking algo and the PSO as follows:
#1. As Jamie mentioned the PSO is trying to solve problems, not just look interesting. It is in the class of evelutionary algos that find solutions to problems, this one based on the diversity of its populations and "fitness"
#2. You CANNOT change the target of the PSO as it is evolving. Well you can, but the population will not nessesarily find the new target. What tends to happen is you get a cluster of particles gravitating around the original target, and a new set stuck around the new one. The system never optimizes.
So if you want to "flock" to a fixed target, wait for optimization (all particles end up on the same point), then define a new target and re-randomize the population than the PSO may work for you.
If there are any PSO experts on the list I would be interested in newer PSO algos that:
A. Can deal with moving targets B. Will be able to find multiple solutions (particles optimize on multiple points)
B>
Jamie Bullock wrote:
The PSO (Particle Swarm Optimizer) object by Ben Bogart kind of does this. It can be found in the CVS under externals/bbogart/pso. It is really designed for solving optimisation problems, but it is based on the biological model of a flocking behaviour. You set a target as an n- dimensional vector and n solutions will "flock" through the space following their neighbours and the optimal solution.
Jamie
On Thu, 2005-05-05 at 19:42 -0400, Thomas Ouellet Fredericks wrote:
I would like to know if anyone has any suggestions on how to get a "flock of birds" behavior in Pure Data on XP? No the graphics, just the positions.
Tom
PD-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
PD-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list