Re: [Repast-interest] scheduler in simphony

[Michel N. <mic...@gm...>]

Miles,

I changed  *if* ( k==weight) by *if* ( k>=weight).
It functions correctly.
Thank's
Best regrards.
Michel


On 2/26/07, Miles Parker <mil...@gm...> wrote:
>
>
>
> cc'ing to list..
>
>
> I think there are simply logic issues w/ code. With a quick look I think
> that the issue is that you are grabbing nextNode every period and testing
> k==weight (fragile, k>=weight prob. better). So anytime you grab a series of
> nextNode in a way that k is allowed to be > then the node distance then you
> never have this condition met.
>
>  On Feb 26, 2007, at 10:10 AM, Michel Nabaa wrote:
>
>  Miles,
> Thank you for your help,
> I will post my "little" experience on the r interest.
> I think a missunderstood was happend.
> My model runs correctly for 30 steps and the vehicules don't move after.
> I don't know why the scheduler works correctly for 30 steps and doesn't do
>
> that correctly after. In fact, after 30 steps (31,32,33.......), my
> vehicules do not continue any more
>  to move form a node to an adjacent another !!!???
>
> Regards,
>
> Michel
>
>
>
> On 2/26/07, Miles Parker <mil...@gm...> wrote:
> >
> >
> >
> > Excellent -- please post experience to r-interest? The _more_ stable
> > version is in the next couple of weeks..
> >
> >  On Feb 26, 2007, at 7:03 AM, Michel Nabaa wrote:
> >
> >  Hi ,
> >
> > long k;
> > *
> >
> > public
> > *Vehicule(){
> >
> > k=0;
> >
> > }
> >
> >
> >
> > @ScheduledMethod
> > (start = 1,interval = 1)
> > *public* *void* step() {
> > k = k+1;
> >
> > Context context = ContextUtils.*getContext*(
> > *this*);
> >
> > ContinuousSpace space = (ContinuousSpace) context.getProjection(
> > "Continuous Space");
> >
> > Network network = (Network) context.getProjection(
> > "A Network");
> >
> > Node nextNode = (Node)network.getRandomAdjacent(
> > node);
> >
> > weight = getEdgeWeight(network,node,nextNode);
> >
> > // weight = 5;
> >
> > NdPoint pt = space.getLocation(nextNode);
> >
> >
> > *if* ( k==weight )
> >
> > {
> >
> >
> > space.moveTo(*this*,0,(*float*)pt.getY(),( *float*)(pt.getZ()+Math.*
> > random*()));
> >
> > *this*.node = nextNode;
> >
> > k=0;
> >
> >
> >
> >
> >
> > }
> >
> >
> >
> >
> >
> >
> >
> >
> > }
> > my model runs correctly for 30 steps and the vehicules stop after.
> > If i do weight = constant (5 for ex), my model runs correctly.
> > Thank's for your help.
> > Moreover, i wanna know, please when the simphony stable version will be
> > available?
> >
> > Michel
> >
> >
> >
> > On 2/23/07, Miles Parker <mil...@gm... > wrote:
> > >
> > >
> > >
> > > Hi Michel,
> > >
> > >
> > > Yes, personally I would use Repast J if my work was critical and
> > > dependent on it. I think we made it pretty clear that the release version is
> > > very early alpha. We should have a newer version real soon now, but I would
> > > not plan on it solving your current issues or to be frank having much better
> > > docs. That said, if you continue with Simphony, I think you would be best
> > > off using the approach described by Nick and myself, e.g.
> > >
> > >
> > >
> > > Another way to do this would be to take a state approach, where every
> > > agent is always say "IDLE" or "MOVING" and then you have a scheduled method
> > > activated every period that checks movement status, updates some count and
> > > when it reaches the time between nodes, moves the agent. This is sort of the
> > > time-step/rules based approach ( e.g. a lo of social science
> > > ABMs/Ascape) and is simpler but less efficient for cases where there may be
> > > long durations between nodes. (The schedule approach doesn't have to "visit"
> > > every agent every tick.)
> > >
> > >
> > >
> > >
> > >  On Feb 23, 2007, at 12:45 PM, Michel Nabaa wrote:
> > >
> > >  I tried your method but it does not function even when i look in the
> > > API doc.
> > > you developped many classes for scheduling, but i don't know if i
> > > schedule
> > > the operations in my context class as in repastJ, you haven't any demo
> > > about this... because, with annotations, i can't schedule my actions
> > > dynamically.
> > > I find simphony very good, but not documented until the moment.
> > > I don't know what i must do!!!  i prepare my phd thesis and i am
> > > interested to
> > > repast especially simphony version. I help that it will be more stable
> > > and documented as soon as possible.
> > > Is it a good idea to return to repast j (many users advises me to do
> > > that) ??
> > > is there a good graph library as in simphony?
> > > Can i do the same as in repast j  in simphony??
> > > simphony will stay alpha1 until when?????
> > >
> > > I need your help, please.
> > > I lose much time on this.
> > >
> > > Michel
> > >
> > >
> > > On 2/23/07, Miles Parker <mil...@gm... > wrote:
> > > >
> > > >
> > > >
> > > > Michel,
> > > >
> > > >
> > > > Java annotations are static so you would never be able to do a
> > > > dynamic schedule exactly like this. In that sense annotations are really
> > > > just a convenience. Also, don't you really want your start time to be the
> > > > time that the move occurs (t + weight) and duration to be 1?
> > > >
> > > >
> > > > You can look throguh the api, and perhaps someone can suggest some
> > > > example code or a better approach, but there are lots of ways to do this,
> > > > for example something like..
> > > >
> > > >
> > > > final Vehicle v = ...
> > > > ISchedule s = RunEnvironment.getInstance().getCurrentSchedule();
> > > > ScheduleParameters p = ScheduleParameters.createOneTime(
> > > > s.getTickCount());
> > > > s.schedule(p, new IAction() {
> > > > public void execute() {
> > > > v.moveTo([targetNode]);
> > > > }
> > > > });
> > > >
> > > >
> > > > wanring, I haven't actually compiled or tested this..
> > > >
> > > >
> > > > Another way to do this would be to take a state approach, where
> > > > every agent is always say "IDLE" or "MOVING" and then you have a scheduled
> > > > method activated every period that checks movement status, updates some
> > > > count and when it reaches the time between nodes, moves the agent. This is
> > > > sort of the time-step/rules based approach ( e.g. a lo of social
> > > > science ABMs/Ascape) and is simpler but less efficient for cases where there
> > > > may be long durations between nodes. (The schedule approach doesn't have to
> > > > "visit" every agent every tick.)
> > > >
> > > >
> > > >
> > > >  On Feb 23, 2007, at 5:06 AM, Michel Nabaa wrote:
> > > >
> > > >  hi all,
> > > >
> > > > in my model, i have two projections : network and continuous space
> > > > and two kinds of agents:
> > > > 1 node agents
> > > > 2 vehicule agents.
> > > >
> > > > the weight on a edge between two nodes corresponds to time for
> > > > moving between these nodes.
> > > > A vehicule agent must move from a node to other node with respect to
> > > > the weight of the edge.
> > > > When i do
> > > >
> > > > @ScheduledMethod
> > > > (duration = weight,start = 1, interval = 1), i obtain this error :
> > > >
> > > > The value for annotation attribute ScheduledMethod.duration must be
> > > > a constant expression.
> > > >
> > > > Any one can help me, please?
> > > >
> > > > Regards
> > > >
> > > > Michel
> > > >
> > > >
> > > >
> > > > -------------------------------------------------------------------------
> > > > Take Surveys. Earn Cash. Influence the Future of IT
> > > > Join SourceForge.net's Techsay panel and you'll get the chance to
> > > > share your
> > > > opinions on IT & business topics through brief surveys-and earn cash
> > > >
> > > > http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV_______________________________________________
> > > > Repast-interest mailing list
> > > > Rep...@li...
> > > > https://lists.sourceforge.net/lists/listinfo/repast-interest
> > > >
> > > >
> > > >
> > > >
> > >
> > >
> > >
> >
> >
>
>