Courses/CS 461/Winter 2006/Mark Luntzel/Week 2

From CSWiki

a1)

For the Random Balls model, an emergent property would be the trace of the path of the center of mass.

a2)

For the Ramdom Balls model (Chemistry & Physics -> Mechanics -> Random Balls), the balls move in straight lines and bounce off the edges but not against each other, so that the total center of mass changes as the balls move randomly.

a3)

it is evident that the number of agents (balls) in the model will greatly affect the emergent chacteristic. There isn't a whiole lot to say about this one, as there are not many parameters and the properties of the agents are rather limited. the balls move in a straight line and bounce off the sides of the "billiard table." Consequently the center of mass is changed .The less balls one puts on the table means that the center of mass will move accordingly, and more balls restrict the movement of the center of mass.

This works because of physics center of mass properties, where the center of mass of any number of objects can be calculated in a system.

b1)

For the Slime model (Biology -> Slime), an emergent property would be the tendency for the turtles to congragate despite not having any sort of heirarchical "leader" to group behind.

b2)

The agents (turtles) each drop "pheromones" which are meant to attract other agents, but the attractors dissipate over a fixed amount of time.

b3)

The less agents there are means that the chances of them "smelling" the other attractors is severely reduced. At around 50 agents, the agents will congregate randomly into several groups. At the maximum 400 agents in the model, this grouping happens faster (intuitively and experimentally) and the sub-blobs will merge if they are close enough to each other.

The angle of the agents "sniffing" can be adjusted. Increasing this value makes the agents group into much smaller blobs and faster, as they can quickly pick up more of the attractor.

the amount of wiggle can be adjusted, making the agents follow random paths until they hit a certain gradient level of the attractor, before they group. some are inevitably left wandering as they do not come into contact with the sufficient level of the attractor.

these rules when taken together allow the agents to group, without a primary attractor.

c1)

In the Cooperation model (Biology -> Evolution -> Cooperation), the emergent property would be that cooperative agents, rather than gereedy agents, are more successful in reproducing

c2)

in the default settings for this model, the agents each have an equal probability that they will be either cooperatiove or greedy. each have a reproduction cost as well as a metabolism, stride-lengh, and reproduction-threshold (has enough energy to reproduce), which all affect how successful each will be.

c3)

it is difficult to tell exactly why the emergent phenomena arises. I suspect that since the agents are cooperative, they're giving the other cooperative agents more to eat and so they have more energy and can reprduce more often. perhaps a somewhat flawed comparison would be socialism.

what is interesting is when you set the chances that the agents will be cooperative down to about 0.10, they do alright against the numerous greedys for a while but inevitably they cannot find enough grass to eat and die off. what the greedy cows eat then must give rise to systemic mad cow disease. very interesting!