/users/rlr/RePast/Demos-3/SIR1_DS/Readme.txt
--------------------------------------------

This is a simple  SIR (susceptible-infected-recovered) agent-based model.

This copy uses a dynamic schedule of events (DES) vs
the discrete time (synchronous) version in SIR1 from
which this was dervied.

To make a copy:

/appl/repast/Demos-3/bin/copyProject.pl \
    -S/users/rlr/RePast/Demos-3 -XSIR1_DS  \
    -R/users/YOURLOGIN/ -PSIR1_DS

where YOURLOGIN is your login, and you might have 
       -R/users/YOURLOGIN/RePast
if you want to put the model in a directory named RePast/ .

Then you can 
  cd SIR1_DS
  ant compile
You can then get a further description of the model with:
  bin/batchrun.sh --help
and then run it with
  ant run
  ant run -Dargs="D=2"
  bin/batchrun.sh nA=100 ncps=2 T=50
Look at the report.* file (eg report.00 is the default)
to see the results.

See the file
  /users/rlr/RePast/Demos-3/SIR1_DS/notes-050410.txt
for additional information.

----------------------------------------------------------------

SIRS1_DS -- simple agent-based epid model.

Agents are in one of 3 states:
    Susceptible   Infected  Recovered 
We start with all in state S except 1.

The basic dynamics:

We want agents to schedule events of these types:

- contactAnotherAgent
  On contact if one is infected and the other is not,
  then with probInfInfectsS, change the other to infected, 
  and schedule a becomeRecovered event.  at time 
      timeInfected + t
  where t = Normal( infectionDuration, infectionDuration/4.0 )
  
- becomeRecovered 
  Set state to Recovered

We would like on average numContactsPerStep (cps) per agent, where
a "step" is a Repast unit of time (e.g., interpreted as a day).
To get started, Model-buildSchedule  schedules contactEvent's for
each agent, the time for each being drawn from a 
uniform random distribution U[0,1].
Then, on each contact event, the instigator agent 
schedules itself for a future event at a time drawn
from, say, a normal distribution with mean 1, and some variance.
(In this model, lets just make the variance = 1 for simplicity.)

NB: Be sure that the time added to current time to
    get future events is > 0, so that the events
    are scheduled in the future!

The Model's contactEvent(a,b) method decides if
one agent infects another, and if so, tells that agent
to execute its infected() method.
The Agent-infected() method sets its state and schedules
a RecoverEvent.   The Agent-recover() just changes state.


We also want to write stats to report file each step (day).
Stats written each t include:
  #Newly Infected at that t
  total S, I, R at that t

One could of course add other events, eg,
  - becomeInfectious
  - becomeDormant
  - die
If there are die events, one has to be careful to remove
events from the schedule for those that die, which means agents
need a ListArray instance variable that contains pointers
to all future events they are to participate in.

(Note we could also have an ArrayList of pastEvents,
so that we would have a history of all events the agent
participated in during the entire run.)

Here are parameters (alias) = default values:

rDebug (D) = 0
seed (S) = 1234567
stopT (T) = 100
initialAgentsFileName (iAFN) = 
initialParametersFileName (iPFN) = 
infectionDuration (infDur) = 10
numAgents (nA) = 16
numContactsPerStep (ncps) = 1
outputDirName (oDN) = ./
probInfInfectsS (prIIS) = 0.05
reportFrequency (rF) = 1
reportFileName (rFN) = report
runNumber (rN) = 0
saveRunEndState (sRES) = 0
worldSizeX (wsX) = 40
worldSizeY (wsY) = 40