Bayesian Filter
---------------

Project: build a spam filter as in Hackers and Painters.
Tell students at the beginning of the semester to start
saving legitimate email and spam, maybe merge everybody's
together and agree not to read it.

Generate a training set and a testing set; have a contest
to see who does best on the testing set.


Determinism
http://en.wikipedia.org/wiki/Determinism




Complexity digest

http://www.comdig.com/index.php


Whaddawe make of this?  http://www.themoderngeneralist.com


Types and Forms of Emergence
Authors: Jochen Fromm
Subj-class: Adaptation and Self-Organizing Systems;
Pattern Formation and Solitons

    The knowledge of the different types of emergence is essential if
we want to understand and master complex systems in science and
engineering, respectively. This paper specifies a universal taxonomy
and comprehensive classification of the major types and forms of
emergence in Multi-Agent Systems, from simple types of intentional and
predictable emergence in machines to more complex forms of weak,
multiple and strong emergence.

cm/fromm05types.pdf
http://arxiv.org/abs/nlin.AO/0506028



Buchanan, Nexus
---------------

Graph algorithms: Mark Buchanan Chapter 2

1) graph representation

2) graph vizualization (see Gato)

3) random graphs (what's the Erdos result?)

4) generating synthetic small world graphs.


Article about small world networks for searching in peer-to-peer
http://www.mathaware.org/mam/04/essays/smallworld.html


Buchanan, page 86:  WWW is small because a few nodes have many
edges, not because some edges are non-local, so there is more
than one way to be a small network...

Page 94, interesting example of a failed explanatory model.
A predictive model can be right for the wrong reason?  What
does this mean for an explanatory model?


Project: evaluate the ability of a set of nodes to carry out a DDOS
attack by max flow calculation.  How does the behavior depend on
network architecture?

Pages 144, 145, sequence of models.  What is the underlying argument
of each?  What was the reaction?

Page 158, is Landau's idea a form of anti-reductionism?

Massive fallacy on page 194 -- the test of "trickle down" is not
whether it increases equality, but whether it makes the poorest
people better off in absolute terms.

(In the context of finite resources, relative wealth does matter,
since the price gets bid up by whoever has the most wealth.)

page 206, interesting hypothesis about Silicon Valley and Rte 128

page 207, claim about changes in physics are really changes in
science, but Buchanan is physics-centric, "new way of doing science"


Laughlin and Pines: tough read, maybe I can get a physicist to
come explain it, but the last paragraph is great.










Bayesian update: scene from Groundhog Day with Chris Elliot.
Larry: He might be ok. [explosion] Well, no, probably not now.


What does IMDB mean when they say "Bayesian estimate"?
http://www.imdb.com/chart/female



Example where a little analysis goes a long way:

Problem 183 from Math Horizons April 2004

Pirates on a plank.



Cell Simulation

http://flash.uchicago.edu/~emonet/biology/agentcell/

Time based grading?  Effort based (combo of time and effectiveness)?

Teach bayes' theorem to half class, make them teach other half.


Preliminary Outline of Topics for Computational Modeling
Allen Downey
Fall 2005


0) Themes

Challenges to the principle of determinism:

   (present state + laws = future state)
   (effects have causes)
	   
   quantum, chaos, Wolfram's thesis, SOC

Computation as a third kind of science.

Algorithmic breakthroughs that lead to scientific breakthroughs
	    (what we know is what we can compute)

Algorithmic breakthroughs that lead to design breakthroughs
	    (what we build is what we can analyze)

Critical evaluation of "new kind of science" claims.
	 (revolution or just another crackpot theory?)

A new kind of engineering

  design for analyzability

  black box abstraction for controlling/limiting interactions

  reductionism for debugging

  new kind of engineering: evolving circuits/code to meet specs
  
  "debugging" by automated testing and random mutation/recombination




1) Survey of fields where computation has created new ideas/paradigms

Finite automata (Game of life and Wolfram's stuff)

Galaxy models

4-color theorem, automatic theorem provers

Unstable aircraft

Gehry's designs

Bioinformatics, human genome project

Ising model of magnetism

Chaos, nonlinear maps

Self-organized criticality

Long-tailed distributions, self-similar network traffic,
long-range dependence

Emergence (turtles, termites, traffic jams)

Small world phenomena (scale free networks, etc)


2) Computational techniques for scientific exploration

Finite automata

Discrete event simulation (priority queue implementation)

Monte Carlo simulation

      random number generation

Vizualization / exporatory data analysis.

Programming as a tool for understanding (frames and transformations)

Bayesian statistics

Fuzzy logic

Analysis of algorithms



3) Projects

birth order

income inequality

locality of reference in library checkouts

sand pile models

StarLogo models (termites)

Langton's ant (turmites)

infectious disease model