/users/rlr/Courses/ICPSR-02/SimpleModel1/notes.txt
--------------------------------------------------

A simple model to show some basic features of Swarm,
and to act as a starting template for more complicated models.

This model has one type of Agent, living in a 2D (torus) world.
In this version, each agent:
  - wanders around randomly, by trying to move one cell over
    in a randomly selected direction.  But if the selected cell
    is occupied, the agent stays where it is.
  - picks one of its (up to) 8 neighbors at random,
    and plays one round of a PD with that other agent.
    Each agent has a value i, which is its probability of cooperating.
    So if each have i=1, they both cooperate; each each has i=0,
    they both defect.  After they play (using the usual 5,3,1,0 payoffs)
    they each increment their wealth by the resulting scores.

Thus the Agents have:
 ID     - arbitrary identifier
 x,y    - location in grid
 i      - probability they cooperate
 wealth - cummulative score
 numberOfMeetings - a count of times they have met (and played) some other

The ModelSwarm-buildObjects method is where agents
are created and placed in the world and on the agentList.
The ModelSwarm-step method is where agents are activated,
i.e., sent the step message themselves.
The agentList is shuffled each ModelSwarm-step, so they are
activated in a new random order each step.

The Agent-step method currrently just calls
the Agent-makeRandomMove method to try to move, and then
it calls the Agent-getRandomNeighbor to see pick a neighbor to play,
and then if there is another to play, it calls the
Agent-play method to actually play.
Note both agents have their wealth and numberOfMeetings incremented
when they play.

Values written to the report file (and plotted on GUI) are:
 - average wealth per meeting
 - average wealth per time step
 
From this simple program, there are many directions one could go.
For example:

- Add an adaptive mechanism, along the lines of Epsteins
  demographic PD.
- All the agents to play multiple rounds when they meet.
  This would most likely involve a new space of possible strategies.
- Giving the agents the ability to refuse to play, e.g.,
  based on reputation or tags.
- Give them the ability to move in non-random ways,
  e.g., if they can see agents with good reputations, or particular tags,
  or some other feature of agents (or the world).

Of course one could also change the agents so they
are not playing PD at all, and have them interact in some other ways.



====================================================================
To establish a copy of this in your home directory:

   cd ~
   mkdir SimpleModel1
   cd SimpleModel1
   cp /users/rlr/Courses/ICPSR-02/SimpleModel1/*.h .
   cp /users/rlr/Courses/ICPSR-02/SimpleModel1/*.m .
   cp /users/rlr/Courses/ICPSR-02/SimpleModel1/Makefile .
   cp /users/rlr/Courses/ICPSR-02/SimpleModel1/*.txt .

Then to compile and link it enter:

   make

If there are no errors, enter
  ll
and you should see something like:

-rw-r--r--    1 rlr      car          1069 Jul 24 23:40 Agent.h
-rw-r--r--    1 rlr      car          5556 Jul 24 23:41 Agent.m
-rw-r--r--    1 rlr      car         60120 Jul 24 23:41 Agent.o
-rw-r--r--    1 rlr      car           649 Jul 24 17:59 BatchSwarm.h
-rw-r--r--    1 rlr      car          3246 Jul 24 18:00 BatchSwarm.m
-rw-r--r--    1 rlr      car         57064 Jul 24 22:49 BatchSwarm.o
-rw-r--r--    1 rlr      car          1521 Jul 24 17:42 main.m
-rw-r--r--    1 rlr      car         51164 Jul 24 23:40 main.o
-rw-r--r--    1 rlr      car          2081 Jul 24 22:20 Makefile
-rw-r--r--    1 rlr      car           736 Jul 24 22:44 ModelSwarm.h
-rw-r--r--    1 rlr      car          7452 Jul 24 22:51 ModelSwarm.m
-rw-r--r--    1 rlr      car         65776 Jul 24 22:52 ModelSwarm.o
-rw-r--r--    1 rlr      car          1459 Jul 24 23:23 notes.txt
-rw-r--r--    1 rlr      car           818 Jul 24 18:01 ObserverSwarm.h
-rw-r--r--    1 rlr      car         10033 Jul 24 23:12 ObserverSwarm.m
-rw-r--r--    1 rlr      car         73200 Jul 24 23:12 ObserverSwarm.o
-rw-r--r--    1 rlr      car           366 Jul 24 22:20 Parameters.h
-rw-r--r--    1 rlr      car          3051 Jul 24 22:35 Parameters.m
-rw-r--r--    1 rlr      car         53596 Jul 24 22:40 Parameters.o
-rwxr-xr-x    1 rlr      car          6505 Jul 24 23:41 sm1

You can then run the model

  ./sm1 &

The usual Swarm widgets should pop up.
As with Tipping, Axelcult, etc, you can
change parameters on the run command:

  ./sm1 -DnumAgents=5 &

or run in batch mode

  ./sm1 -DstopT=100 -b &

To see a list of parameters:

  ./sm1 -Dh

============================================================================