Re: Using HDF5 to create an R data frame

[Marcus G. Daniels]

>>>>> "B" == Benedikt Stefansson <address@hidden> writes:

B> The harder issue is how to do statistics over lists of objects with
B> instance variables. I have not been able to do that
B> successfully. Problem is to make statistical methods act on
B> instance variables of objects instead of more common
B> array/matrix/dataframe.

My hope/intent was that it would be possible to structure a model in
such a way that lists with objects which were candidates for statistics
would be serialized `shallow' so that one could run R routines 
on the dataframes.  For example, this should work with the 
collection.hdf5 file in the previous example:

  mean(myCollection$value)

I can see that converting the complicated trees produced by deep
serialization into a suitable form for routines like `mean' could be
annoying (it's not impossible, or even hard `given the task': someone
could write utility routines for R to do common transformations).  I'm
just not sure that this is a practical problem, if it can be
accomodated in a Swarm model design.  For example, if the goal is to
output data for loading into R (not to load and save exact agent
state), then one can cheat by writing serialization routines that skip
over component objects.

                  ==================================
   Swarm-Support is for discussion of the technical details of the day
   to day usage of Swarm.  For list administration needs (esp.
   [un]subscribing), please send a message to <address@hidden>
   with "help" in the body of the message.