The elements can be cell-centered (data dep connections) in DX.  Some
operations require node-based data (e.g., Isosurface), but Post will do the
interpolation for you.

I was thinking of data at vertices, which is why I suggested more
generically decomposing to tets.  But in your case, trying cubes is
reasonable.

Yes, represent different cell types as a group.  But if the types are the
same, you should use a multigrid instead of a simple group.



Steve Ettorre <[EMAIL PROTECTED]>@opendx.watson.ibm.com on 08/20/2000
05:12:43 PM

Please respond to [email protected]

Sent by:  [EMAIL PROTECTED]


To:   [email protected]
cc:
Subject:  Re: [opendx-users] supported element types



Thanks for the response. Two followup questions -

- I am dealing with cell centered data (results are computed at the
centroid
of the volume cell). In view of this, for pyramids and prisms it seems more
natural to think of them as degenerate cubes (view a prism as a cube with 2
duplicate nodes and a pyramid as a cube with 4 duplicate nodes). This way
the
data are still residing at the cell centroid. Does this seem like a
reasonable approach?

- Is it best to represent the collections of different cell types as
different fields within a group?

TIA,
-S.



Reply via email to