[vtk-developers] vtkIntegrateAttributes for point cells

[Biddiscombe, John A.]

Inside vtkIntegrateAttributes is a large switch statement which sends the cell by cell integration to the relevant part of the code.

I would like to do a summation of values over particles, and the integrate attributes filter already does the book keeping for all the cell/point arrays. I first made a subclass to handle a simple particle summation, but quickly found that there are a number of internal private variable (class within vtkIntegrateAttributes) that I don't have access to.

Adding another switch option for particles was trivial, simply call the relevant integrate function with a weight of 1, instead of previously computed area or volume and hey presto. The summation is performed.

I'd like to modify the documentation to say something along the lines of "When this filter operates on data containing points/particles, the integration assumes a unit weighting per particle and does not consider area or volume of cells..." and so on.

If I commit this change, a dataset which contains mixed cells may return a different result than it currently does, since previously, particles or points were simply skipped and contributed zero to the summation. I believe that anyone using the filter with mixed topology is running a risk already, so ...I plan on committing this change and will add a warning to make sure that if mixed topologies are encountered the user is informed. If anyone has strong feelings about this, then please say so.

yours

JB

-- 
John Biddiscombe,                            email:biddisco @ cscs.ch
http://www.cscs.ch/
CSCS, Swiss National Supercomputing Centre  | Tel:  +41 (91) 610.82.07
Via Cantonale, 6928 Manno, Switzerland      | Fax:  +41 (91) 610.82.82