Hi David, On Fri, Jun 8, 2012 at 1:34 PM, david collins <antpuncher@gmail.com> wrote:

Hi, all-

What would it take to get the contour finder in yt to work on a projection?

(n.b., the method paper covers the contouring algorithm for anybody else out there: http://adsabs.harvard.edu/abs/2011ApJS..192....9T ) A new routine, unfortunately.... As it stands, the projection takes the yt QuadTree data structure and dumps it into a set of flat lists. On my agenda for the next N units of time (where N is undefined) is making the pixelization routines understand quadtrees so that this can be avoided, but that is off a ways. Clumpfinding on quadtrees is pretty easy since they have inherent spatial information, so that would help out with this. But, clumpfinding on 2D objects, even of variable resolution, is so easy as to be trivial -- particularly when they're all laid out like that. The very first pass, which would scale up to a surprisingly large fraction of datasets, would just be an N^2 check. We can apply smarter algorithms later, but if you want something done now, that'll do it. If you hop on to IRC early next week we can plow through this; there's even some code in the yt history that has this implemented already for 3D, which we could just drop down a dimension. -Matt

d.

-- Sent from my computer. _______________________________________________ yt-dev mailing list yt-dev@lists.spacepope.org http://lists.spacepope.org/listinfo.cgi/yt-dev-spacepope.org