Michael Murphy's Computer Systems Project Proposal
Latimer, period 5. 9/16/05.
MASON
Oct. 5 : prototype 1 - movement is adjacent neighborhoods rather than random.
Try Moore neighborhood vs von Neuman
Oct. 7 - prepared writeup for prototype 1 - see
prototype 1 with screenshots
Text version:
My First Modification to an Agent Based Modeling Program
The original model allowed for the unsatisfied agents in Schelling's
neighborhood to move to any space on the board. This resulted in
highly sorted and segregated neighborhoods as minorities would quickly
move out of their environment and be replaced by a member of the local
majority. This setup allows for neighborhoods to become completely
filled with members of the local majority as majority members can move
to the center of the neighborhood, not being bound by local movement
restrictions. The following screenshots of Steps amounts 1, 2, 50, and 51
show this principle.
Note how the population becomes noticeably segregated between steps 1 and 2.
Note how the neighborhoods are completely defined by steps 50 and 51.
The happy agents fill completely segregated neighborhoods, and the
unhappy agents move randomly in the badlands on the fringe of the
neighborhoods as there are no more open ideal locations.
My modification of Schelling's neighborhood allows the agents only to
move to open spaces in their von Neuman adjacent spaces surroundings.
This lack of mobility resulted in a population that never quite became
well sorted and segregated into populations. A stable neighborhood
setup never really generated because same color agents couldn't congregate
across distances to form localized majority populations. The
neighborhoods in my modification often have spaces in the middle as the
outer members have no reason to move in and those outside the neighborhood
are unable to move within it due to their movement restrictions.
Opposite color agents also often get stuck in the middle of a neighborhood
where they are the stark minority, as they become literally trapped in
the center of the neighborhood again due to their handicapped movement.
These stark contrasts to the original setup are easily seen in the
sample states from steps 1, 2, 50, and 51 of the modified model.
In the extension, we can see how the agents do become more segregated in
step 2 than in step 1. We can also see, however, that step 2 is much
more similar to step 1 than than step 2 in the unmodified version was
similar to its previous step. The localized movement allows for much
more easily predictable movement and population patterns.
The population is fairly segregated by steps 50 and 51, but we can see how
the population is much more spread out, taking up most of the available
terrain. There are no badlands because the populations do not completely
fill the highly desirable inner sections of the neighborhoods. We can
see instances of the spaces in the middle of neighborhoods as well as
minorities being trapped within majority environments.
Oct. 14 - absent
Nov. 4 - Website
"Yes" competition, college board epidmiology scholars
Dec. 14 Current Status Report
Right now, I am working on getting the bugs out of my first version of my
second project for the computer systems lab. My Classroom class appears to
be working, but I deleted my classroomVirus class in frustration and I am
going to program it again with good style, in hopes that it will work this
time. The Student class is also a bit buggy, but I am hoping that if I create
another classroomVirus then everything will fall into place. I am desperately
trying to get this working by Friday so I can spend that period with Dr. Cammer,
getting the ball rolling really fast on finding the actually data that I need
to make my project a viable option for the Young Epidemiologists Scholarship
that I plan to enter.