Instructor:
Dr. Matthew Conroy
Office hours and email
Important Dates:
First project:
Proposal due: January 24
Project due: February 7
Presentations: February 10,12,14
Second project:
Proposal due: February 21
Project due: March 6
Presentations: March 9,11,13
February 10, 2020
Here is a sampling of topics people in past quarters have investigated in their second projects.
The purpose of this list is to help you come up with Project Two ideas.
It is okay to work on project topics that others have done: everyone
is different, and the projects always come out different.
- Graphs
-
MDS
- food consumption by country
- food exports by U.S. state
- visual perception of lipstick color
- tech companies
- Fortune 500 companies
- Wikipedia (distance between article)
- baseball/basketball/soccer players
- U.S. universities
- U.S. presidential inauguration speeches
- Species in U.S. national parks
- Cities
- Countries
- Monte Carlo
- Aeroplane Chess
- Durak
- Snakes and Ladders
- BS (card game)
- Battleship strategy
- Baseball card game
- Starbucks queuing
- Ultimate Tic-Tac-Toe
- Escaping a forest
- Traffic analysis
- Photosynthesis
- Markov
- novelists
- song lyrics
- folk songs
- baseball
- Uno (game) strategies
- breast cancer progression
- Florida Keys coral reefs
-
Markov chain + Monte Carlo
February 10, 2020
Some comments on the Milgram/Travers paper assignment.
- I hoped you enjoyed reading the paper. I definitely got the impression that at least a number of you did. Thank you for all of your thoughts.
- The paper is pretty long, and it is easy to miss what the main point of it is (the writing could be better in this regard).
The most misinterpreted part was this:".The empirical technique of this research has two major
contributions to make to the developmentof that theory. First, it sets an upper bound on the minimum number of intermediaries required to link widely separated Americans." (page 441).
The way I read this is that they are saying that the technique used in the experiment allows one to get an upper bound between two people: the starting person and the
ending person: if the chain starts with person A and ends with person B with x intermediaries, then the number of intermediaries required to link A and B is less than or equal to x.
They are not claiming that they have figured out a numerical upper bound that applies to all pairs of Americans. Certainly they were very aware how far they were from being able
to support such a claim.
- Quite a few people suggested that the study should have included incentives (i.e., money given to people who participated in the study).
I don't know how common this was in the 1960s. It seems like having incentives could lead to people "gaming" the system, and sending
the envelope around to friends (not sending it toward the target) in order to get the money.
-
Some people suggested that the researchers could have included a list of the target person's acquaintances in the packet,
to make it easier for people to connect to them. This sounds like not a bad idea to me, though, of course, that would
require getting those acquaintances OKs.
- Some people suggested it was a failure of the study to use non-random samples (e.g., stock holders in Nebraska).
It seems that the authors were really just trying out this experimental method (i.e., the chain of letters thing), and
so were not so concerned with getting a number at all. If they had used a more random set of starters,
and the results were very poor, they would not have known what caused the failure: was it the experimental method
itself, or was it the set of starters? I think with the study they did, they could conclude that the
experimental method is okay, and then they, or others, could go on using the method to study different populations, etc.
- I thought it was interesting that people were critical of the study for its size.
There are always limitations on size, and it is always the case that a larger, more diverse sample would be better.
As an initial experimental study, I don't know that this sample size was too small for the conclusions that they were hoping to make.
Some students suggested scaling up the study by a factor of 100 or even 10000.
I'm not sure folks thought through the issues associated with such massively scaled work.
-
A number of people said that, since we have the internet, participants in a similar study done today could just look
up the target person on the internet, find their address, and send the package directly to them.
I don't know why people think that participants would so flagrantly violate the rules of the study:
there are no points awarded for being the person to send the envelope to the target, so I don't know
why students thought people would "game" the system in this way.
Also, this was possible in 1969. In those days (until fairly recently, in fact), the telephone
companies published phone books which contained phone numbers, and, in most cases,
addresses of everyone in a city with a phone. They books were available in libraries,
so that, for instance, one could go into a Omaha, Nebraska, library and look at a
Boston phone book to find information about a person in Boston. The internet makes this
easier, but it was still common in the 1960s. In fact, in the 1960s, telephone
operators were common, and you could call a Boston operator from anywhere and ask
them for this sort of information.
- One student made the interesting observation that the graph of acquaintances is not static, and is always
changing: people meet people every day, people die every day, people lose touch every day.
So the study of this graph is could involve more than just the study of it at a particular moment in time,
but over a period of time, or even over all time. Did people have "acquaintances" before
they had language? So many interesting questions!
January 4, 2020
Welcome to Math 381 A, Winter quarter 2020.
For those students looking for an add code, I do not overload my courses.
If there are opening in the first week, you can add yourself to the course.
No adds will be allowed after one week (i.e., starting 10/2/19). There are no add codes.
The first writing assignment will be posted soon.
You will need to install lpsolve to solve LPs in this course.
You can start here.
The download site is here.
You may also want to read the information in the right-hand column about installing
lpsolve on Macs or PCs.
For Macs, if you are running the latest OS, you may need to install lpsolve using
HomeBrew, which is a very useful tool in itself.
Once you have installed Homebrew, you can install lpsolve with the commands
brew tap brewsci/science
brew install lp_solve
executed one at a time on the command line in Terminal.