harris.py

[Thomas S. Dye]

Aloha steko,

Stefano Costa <steko@iosa.it> writes:

> However, a more recent version of the program I described in 2007 can
> be found here
> https://gitorious.org/harris-matrix/harris-matrix and I know a similar
> approach has been used by the pyArchInit QGIS plugin.

The source code has been very helpful.  Thank you for pointing to the
most recent code.

Your code calls the acyclic method of pyGraphViz, which relies on the
GraphViz utility called acyclic. The man page for that utility says:

  acyclic is a filter that takes a directed graph as input and outputs a
  copy of the graph with sufficient edges reversed to make the graph
  acyclic.

In effect, the utility will take one or more stratigraphic observations
and silently invert them.  This is probably not what the user wants.

Also, the call to tred in your code precedes the removal of cycles.
Note that the man page for tred says:

  If a graph has cycles, its transitive reduction is not uniquely defined.
  In this case tred emits a warning.

If you change the order of the calls, like this:

        B=self.A.acyclic(copy=True) # make sure there are no cyclic relations
        C=B.tred(copy=True)  # remove redundant relations

then you will generate a uniquely defined transitive reduction, but will
still possibly invert stratigraphic observations without the user's
knowledge. 

Also, the problem of the graphs not looking nice enough for your
colleagues can be ameliorated somewhat by exporting to svg, rather than
png. I use the open-source Inkscape program to edit Harris Matrices
produced by GraphViz dot.  It works like a charm!

All the best,
Tom

-- 
T.S. Dye & Colleagues, Archaeologists
735 Bishop St, Suite 315, Honolulu, HI 96813
Tel: 808-529-0866, Fax: 808-529-0884
http://www.tsdye.com