Re: [Swarm-Support] Model actualization (II)

[David Camacho Trujillo]

Dear Paul Johnson,


As always, many thank for your response.
I have take care of this calls to “(double) getParameterX” but this problem was
still there. Then I have introduce some statements to “getFloatparamX” at the
beginning of the method “newGeneration” in orther to actualize this values.
This apparently work, but not really.
First it annoy me because I suppose it has not to be necessary.
Second, some how, the impact of the changes under the GUI version is not as in
the pass (when I was changing this on the “modelSwarm” GUI), and it is terrible
in the case of the “batch mode” because after the change of this parameters, the
evaluate method return an “NAN” message. (I guess is internally the same problem
on the GUI version).
And, yes, definitively whether this problem is easy to solve or not, you have
your t-shirt for sure! ;-)
(I hope you don´t have this one!)
(Now I start to prepare the .tar to send it to you, but many comments are in
spanish so...let me translate them! Yes, I know, a terrible habit, but I have do
it before I really learn how to program!) 

David








Mensaje citado por: Paul Johnson <address@hidden>:

> I think there is probably some slippage between the GUI and your GA 
> class and the Parameter object.  My first thought was that the GA code 
> itself was not taking the steps necessary to check with the arguments 
> object to find out the newest values of the parameters.  I'd put some 
> fprintf(stderr,"") diagnostics in to see if the arguments class really 
> does have the value from the GUI and then that the GA asks for the 
> newest values.
> 
> However, you say it works for integers, but not doubles, and that 
> perplexes me. You say there are ints and doubles side-by-side in a 
> particular object and gui changes in an int work, but gui changes in a 
> float do not?  That's a neat trick.  Is it possible you mistakenly used 
> "getInt" instead of "getDouble" to retrieve values from arguments?  That
> 
> might break floating point values.
> 
> I have no other guesses.
> 
> If you send your code in a tarball to me directly, including a README 
> problem explaining the trouble, I might spot it pretty easily. If it is 
> a hard one to fix and takes me more than a couple of hours, I'm going to
> 
> ask  you for a t-shirt from your university/institute. That's the 
> standard fee these days. I'm accumulating some really neat t-shirts from
> 
> various schools around the world.
> 
> David Camacho Trujillo wrote:
> > Dear Swarm community,
> > 
> > I have actualized my model (It consist of a GA developed by J.J.
> Merello)
> > following the new tutorial for the 2.2 version. Therefore, now I don´t
> establish
> > the model´s parameters on the “ModelSwarm” but on the “Parameters”
> Object. 
> > Almost every runs fine, but now, if I change the values of float (or
> double)
> > parameters, they change on the widget (under GUI mode) but the
> modelSwarm does
> > not really actualize them. This is not the case of the “integer”
> values, which
> > are properly actualized. (Same problem under batch mode)
> > 
> > I have notice that the methods on “Parameters” are different for
> integers or
> > doubles, and I am using specific methods to do so “setParameterX” but
> still, it
> > does not work properly and I don´t have an idea on what could be
> wrong.
> > Did you have some?
> > 
> > (Iam working under Suse 9.2)
> > David
> > 
> > _______________________________________________
> > Support mailing list
> > address@hidden
> > http://www.swarm.org/mailman/listinfo/support
> 
> 
> -- 
> Paul E. Johnson                       email: address@hidden
> Dept. of Political Science            http://lark.cc.ku.edu/~pauljohn
> 1541 Lilac Lane, Rm 504
> University of Kansas                  Office: (785) 864-9086
> Lawrence, Kansas 66044-3177           FAX: (785) 864-5700
> _______________________________________________
> Support mailing list
> address@hidden
> http://www.swarm.org/mailman/listinfo/support