RE: [Swarm-Support] batchswarm problem - getActivity

[Corinne Coen]

Marcus,
Sorry to be confusing.  Originally I was seeing crashes with segment faults.
But using gdb I was able to correct that problem.  But now I have different
model performance with the GUI and in batch mode.

CSN is an ivar that is set in Firm.m on line 190.
         CSN = getUniformD( 0, 1);

If you run ./rivalry and read the report, you will see a line that reads 
@CSN's@   0.30    0.13   0.08

If you run in batch mode, that line always reads 
@CSN's@   0.0    0.0    0.0

CSN is not set and does not work.  How can this be?
Corinne


-----Original Message-----
From: address@hidden [mailto:address@hidden On Behalf
Of Marcus G. Daniels
Sent: Friday, March 28, 2008 12:09 AM
To: Corinne Coen
Cc: Swarm Support
Subject: Re: [Swarm-Support] batchswarm problem - getActivity

Hi Corinne,
> To Marcus and Paul.
> Attached is the BatchSwarm.m file that I am using.
I tried the model you sent me.
I observe that in your parameter control says that the simulation should 
stop at time 50.  
So I put in a printf in -stepReport like this:

printf ("%ld\n", getCurrentTime ());

Then I run the code.  What I observe is 1 to 50 printed out and then an 
expected stop.

Do you expect something else to happen?  You mention that:

 > "CSN is the variable that "works" in the GUI.

..but I don't understand how you define that.   I don't see CSN on a 
probe display in the immediate GUI ProbeDisplay.  If I probe the 
ModelSwarm in the ObserverSwarm and then get a Firm, I see a fixed CSN 
ivar there; it stays constant in the GUI version of the model is 
running.  I feel like I must have missed an e-mail on the high level 
problem.  I understood it was just a conspicuous lack of anything happening.

Marcus
_______________________________________________
Support mailing list
address@hidden
http://www.swarm.org/mailman/listinfo/support