Hi Jeroen, hello Net!

>> Jeroen Bergmans writes:
 > On Thursday 10 April 2003 12:16, Lloyd A Treinish wrote:
 [...]
 >> You can also do operations based upon your own attribute to
 >> select members that you or do not want to process.

 > Yes, but what do I use to delete (or ignore) for example all member
 > fields with a "level" attribute less than, say, 3 ? Note that the
 > number of these fields is not known a priori, and can be quite
 > large.

If I interpret Lloyd correctly, he suggested adding an extra attribute
to each of your fields. Something like

object 42 class field
  component "positions" value xxx
  component "connections" value xxx
  component "invalid connections" value xxx
  component "data" value xxx
  ...
  attribute "refinement level" value xxx

You can then evaluate the "refinement level" attribute in your DX
program.

BTW: you can also put things like simulation time and integration step
into the dx file using attributes. I usually do so to make annotations
of e.g. time to graphics.

 > Another thing I noted is that when I use "Post" to convert my
 > connection dependent data to position dependent data, the invalid
 > connections component is not converted to an invalid positions
 > component, this generates some strange interpolation values at
 > borders between grids. How should I "translate" the invalid
 > connection component?

IMHO Post is crap. At least it did nothing useful for me. If you
really need position-dependent data, try to re-interpret your cell
averages as point values located at the cell centres.

The effects you observe might also occur because there is no way for
DX to interpolate across the boundaries: no connections there to
interpolate in.

Hope this helps
-Frank

Reply via email to