nelson_p@apollo.HP.COM (Peter Nelson) (02/08/90)
So far, since this newsgroup's inception only 22 articles to it have appeared at this site. Is self-org-sys real new or just real quiet? I posted some comments about Autodesk's CA Lab here recently. After my disapointment with their offerings, especially in terms of the complexity of the information that can pass between adjacent cells, I decided to write my own. It is already about 75% done and has some novel features which I may describe in a future posting. Someone here recently posted a question about flocking behavior. The reference he was probably wondering about was "Flocks, Herds, and Schools: A Distributed Behavioral Model" by Craig Reynolds at Symbolics. The article was published in Computer Graphics, Vol 21, Number 4, July 1987 (SIGGRAPH Conference Proceedings) A notable thing about the article was that Reynolds was able to show that flocking behavior could be accounted for by *individual* boid's (simulated birds) choices and behavior. The existence or nature of the "flock" emerged from all these local effects. I am curious to learn more about such self-organizing-systems. The natural world is full of examples of how microscopic features or behavior produce macroscopic structures. The shape of my hand or the shape of a leaf is similarly the result of lots of individual cells responding individually to local effects (which is not to say that the "local" environment" to the cell may not be affected by chemicals emitted by cells a great distance away and dissusing through the tissue). Are there any general principals that may be applied to problems like this in simulations? If I wanted to create a CA program that would produce a particular shape, say a star or a triangle, is there any systematic way to go about it? ---Peter
demers@beowulf.ucsd.edu (David E Demers) (02/08/90)
In article <4880f240.20b6d@apollo.HP.COM> nelson_p@apollo.HP.COM (Peter Nelson) writes: > > to it have appeared at this site. Is self-org-sys real > new or just real quiet? > Hmm. We have had 135 here. I don't know anything about Usenet propogation, but someone can probably explain it. > behavior. The reference he was probably wondering about > was "Flocks, Herds, and Schools: A Distributed Behavioral > Model" by Craig Reynolds at Symbolics. The article was > published in Computer Graphics, Vol 21, Number 4, July 1987 > (SIGGRAPH Conference Proceedings) Also discussed in the proceedings of the Artificial Life conference from 1987 (Chris Langton, editor.) > I am curious to learn more about such self-organizing-systems. > If I wanted to create > a CA program that would produce a particular shape, say a > star or a triangle, is there any systematic way to go > about it? Lindenmayer systems are capable of generating structures greatly resembling real plants. See the Artificial Life proceedings for a good article plus lots of color plates of examples, as well as more references to this and other similar work. The second ALife conference is going on right now at Santa Fe, and I wish I were there... As I recall, one of the features of the Lindenmayer work is use of context-sensitive grammars for generation of strings which represent the structures. Thus branching may occur if a certain symbol appears in a particular substring only. Dave
honig@ics.uci.edu (David A. Honig) (02/13/90)
In article <4880f240.20b6d@apollo.HP.COM> nelson_p@apollo.HP.COM (Peter Nelson) writes: > I am curious to learn more about such self-organizing-systems. > The natural world is full of examples of how microscopic > features or behavior produce macroscopic structures. The > shape of my hand or the shape of a leaf is similarly the > result of lots of individual cells responding individually > to local effects (which is not to say that the "local" > environment" to the cell may not be affected by chemicals > emitted by cells a great distance away and dissusing through > the tissue). > > Are there any general principals that may be applied to > problems like this in simulations? If I wanted to create > a CA program that would produce a particular shape, say a > star or a triangle, is there any systematic way to go > about it? > > ---Peter Another interesting self-organizing system created solely by local action is a free marketplace. Another interesting self-organizing system that can be interpreted as evolutionary is the erosion of land: the formation of valleys between ridges with the valley streams merging to form tree structures rooted at the ocean. Let me make the analogy explicit: Evolution requires: 1. heritable variation, ie, as the state of a system (usually the gene pool of a species) progresses through time, new states resemble previous ones but not entirely. 2. evaluation and differential reproduction (natural selection). Take the positions and orientations of rocks to be the state of a geological system. Look at how that state changes over time. 'Stable' substates tend to exist for longer amounts of time. Weathering processes perform natural selection on the positions of rocks and dirt. An example of a stable substate is a flat rock lying on its broad side; an unstable state would be that rock on edge. Random perturbations such as rain drive the search for stable configurations much as mutation (and crossover) drives the search for novel genotypes. Note that individuals within the population (rocks on the same hillside) interact, and influence each other's evolution. For instance, two rocks may together block a stream and thus slow their tumbling down the hill. Local forces involving the local slope of the land, gravity, and water flow result in mountains with globally self-similar ridges and streams (both ridges and rivers form tree structures, the latter draining the former). -- David A. Honig "Live Free Or Don't"