[Swarm-Support] HDF5 hdfview works, Swarm usage mysterious (was Re: New Code: Interpolator...

[Paul E Johnson]

Marcus gave a good tip with this note.

I just downloaded and tried hdfview. It works as advertised. A graphical 
view of hdf5 formatted data.

As far as Swarm is concerned, we have pretty good data output strategies 
with the changes Marcus put into EZGraph for Swarm-2.2.  I think it is 
no trouble at all to run a model and watch the time plots on the screen 
and at the same time save the data into an hdf5 file.  For plots of 
summary indicators, this works perfectly well. For R, we have the HDF5 
library addon that Marcus made, and also in the R list Robert Gentleman 
has appeared with another library which also supports HDF5 data. I did 
not try the latter yet, but the former works fine.

I wish I knew more about creating an hdf5 data set and getting elements 
out of it and into Swarm. I've got the lisp archiving approach figured 
out pretty well, partly because you can look at a file and see something 
understandable:

(list

 (cons 'bfParams
  (make-instance 'BFParams
  #:numfcasts 60

...

And after a bit of trial and error, one can make that work. (You can see
example in the "simpleBatchBug3" tutorial I just put in, for a snap, 
look here:
http://lark.cc.ku.edu/~pauljohn/ps909/Agent-Based_Models/tutorial-2.2-pre3.tar.gz
)

HDF5 is more mysterious, and I've made a few frustrated efforts to 
understand it, but did not yet succeed.

But it is certainly on my list of things to do.  IF other people have 
working example code, please let me/us know.



Marcus G. Daniels wrote:
> Steve Railsback wrote:
>
> Well, considering that Excel stops at 65536 rows x 256 columns, or 
> Quattro at 1 million rows, that falls considerably short of the 350 
> million we need.  And we need considerably more-automated data cleaning 
> techniques, as well.
> There is a spreadsheet-like interface for HDF5, either in the form of 
> the HDF5 data viewer (from the HDF5 site), or with the HDF5 plugin for R.
>
> In any case, it's just a question of choosing the right tool for the job.

--
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