swarm-support
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Swarm-Support] Multiple Simulation Runs


From: Paul Johnson
Subject: Re: [Swarm-Support] Multiple Simulation Runs
Date: Mon, 13 Dec 2010 14:27:59 -0600
User-agent: Mozilla/5.0 (X11; U; Linux x86_64; en-US; rv:1.9.2.13) Gecko/20101208 Lightning/1.0b2 Thunderbird/3.1.7

Greetings

On 12/11/2010 08:44 AM, address@hidden wrote:
> Dear all,
> 
> I am an economics student, new to swarm, and currently doing my diploma 
> thesis. I have got some Problems with Java/swarm. Maybe someone can help me?
> 
> I want to do multiple (40 or so) simulations in batch mode, without 
> restarting the programme after each run. The Problem is that I get a maximum 
> of 397.000 simulation iterations. That is enough for each run, since I only 
> need 300.000. But somehow there is a problem with the multiple run design, I 
> have. After 90.000 iterations of the second run the simulation is aborted.
> The error message says something about “malloc”.
> I suppose there is some problem with the memory?
> If I put the swarm initialization in the for-loop for the iteration it breaks 
> down immediately after the first run. Is there some method of finilizing 
> swarm or clear all memory or so in order to prevent that?
> I would be very thankful for your help, I am working on that for 4 days now 
> and have no clue what else I could try…
> 

Swarm does need to shut down cleanly at the end of each simulation run.  Then 
you run it again.

In the most recent edition of the Swarm ObjC Tutorial, I put quite a bit of 
effort in documenting how you can run things
over and over with various parameters.  look for the batchBug sections.

The crash you are seeing with "malloc" is indeed memory, it means probably that 
you have freed some memory and then
tried to access it, or that you have a loop that goes out of range, or such.  
So finding out why the crash happens is
your first priority.  Figure out what simulation conditions cause that one run 
to die, then run interactively under gdb
so you can see where the crash happens.

After your program "runs perfectly" :), that is the time to script it up to run 
over and over again.

pj


> Thank you very much in advance!
> 
> Best Regards
> 
> Johannes Nienhaus
> 

-- 
Paul E. Johnson                       email: address@hidden
Professor, Political Science          http://pj.freefaculty.org
Assoc. Director, Center for Research Methods and Data Analysis
1541 Lilac Lane, Rm 504
University of Kansas                  Office: (785) 864-9086
Lawrence, Kansas 66045-3129           FAX: (785) 864-5700


reply via email to

[Prev in Thread] Current Thread [Next in Thread]