Re: [Swarm-Support] Asynchronous schedule

[Derek Farren]

Hello Dinis.

I am also new to Swarm, and was in your shoes a few months ago.

Here is what I would do:

1) I would create a class variable that counters the timesteps... lets call 
it "counter", and a class variable that saves the time where the last event 
happened, lets call it "lastCounter".
2) you need a method on your agent class that giver you the time for the 
next event, taken from an exponential distribution. Lets call the method 
"getNextEventTime" and the class variable that it gives "nextEventTime".
3) create a setter method on your agent class to "refresh" the variable 
nextEventTime only if its time to do it. It should look like this:

-setNextEventTime
{
  if (counter-lastCounter == nextEventTime)
  {
    nextEventTime = [anAgent getNextEventTime];
    lastCounter = counter;
  }
return self;
}

I hope this helps you.

Derek

----- Original Message ----- 
From: <address@hidden>
To: <address@hidden>
Sent: Friday, April 15, 2005 2:07 PM
Subject: [Swarm-Support] Asynchronous schedule


> Hello everyone,
>
> I'm trying to build a Swarm (java version) model with timesteps dependant
> on the state of things at different times. Basically i want to draw the
> time to the next event from an exponential distribution, choose my event,
> and then schedule it. This is what I've done in a mean field simulation in
> fortran but I'm not quite sure that this can be done for multiple
> individuals. I suppose so but I don't know exactly how.
>
> I suppose each individual should have a method that, when called, inserts
> the time to the next event in the schedule, and at that time, the method
> gets called againd to determine the time to the next event. But then...
> how to schedule a message for that agent only? I suppose it has something
> to do with indexes but I also don't know how to use those. Not a lot of
> information around for Java too... Maybe I sould have gone for objective C
> but everyone kept telling me that it wasn't used anymore, Java was
> everywhere, blah blah blah...
>
> I'm new to programming and Swarm altogether, and I've spent the last month
> struggling until I finally got something close to what I wanted working. I
> hope I haven't reached a dead-end after all that I had to learn this past
> month just to get one lousy algorithm working! :-)
>
> Anyway Swarm looks cool even if just for fun, and there seems to be a lot
> of enthusiasts of ABM, but right now I needed a serious algorithm for
> serious work. Hope someone can/will help me... A future Swarm citation in
> a scientific paper could be in hand if I get this done! ;-)
>
> Dinis Gökaydin
>
> Theoretical Epidemiology Group - IGC (www.igc.gulbenkian.pt/sites/ggomes)
>
> Oeiras - Portugal
>
>
> _______________________________________________
> Support mailing list
> address@hidden
> http://www.swarm.org/mailman/listinfo/support