[yt-users] Problem with annotate_contour() callback on OffAxisProjectionPlots

[Nathan Goldbaum]

This commit puts together the plumbing necessary to get Matt's proposed fix
to proceed without throwing AttributeErrors:

https://bitbucket.org/ngoldbaum/yt/commits/3b5b687182b1a51fb2aa46d879dc338c76b46ac4

No guarantees of correctness - in particular I think you'll need to take
care to make sure the orientations are what you expect by specifying a
north vector.

Cameron if those changes work for you can you bundle my and Matt's changes
into a pull request?

-Nathan


On Tue, Apr 1, 2014 at 4:20 PM, Matthew Turk <matthewturk at gmail.com> wrote:

> Hi Cameron,
>
> On Tue, Apr 1, 2014 at 4:51 PM, Matthew Turk <matthewturk at gmail.com>
> wrote:
> >
> > On Apr 1, 2014 4:42 PM, "Cameron Hummels" <chummels at gmail.com> wrote:
> >>
> >> Sorry about that--I wasn't totally clear in what I wanted to get out of
> >> this.
> >>
> >> I wanted the contours in test4 to be identical to the contours in test3,
> >> since they're both created using annotate_contour() with the same
> >> data_source.  Yes, there are contours in test4.png, but they are
> >> significantly different from those in test3.  Does this make sense?
> >>
> >> If you look at the contours for test1 and test2, they match, despite the
> >> fact the underlying data being plotted change.
> >>
> >
> > Ah, I understand now.
> >
> > So, two things. I have used real data (much easier than random) and the
> > contours do not match as well as I would like. However, they cannot match
> > exactly - you can see this in the random pf by looking at how the off
> axis
> > projections are covered with what look like stencil outlines of cells.
> These
> > won't go away, regardless of what we do with the off six projection,
> because
> > of how the data is created. The on-axis is produced by pixelizing a
> sparse
> > grid of points, which is then fed into a Delaunay triangulation before
> > contouring the result. The off axis is already densely packed, which
> means
> > the interpolation process becomes nearest neighbor.
> >
> > As for the other issue, I believe there is a transposition error combined
> > with the fact that linearly interpolating inside a cell produces
> different
> > results during the projecting process. I think you had proposed an
> > adjustment to the axis order for OAP objects; did it make it in?
>
> Ah!  I think I found it.  I was doing something wrong in my script; by
> setting north_vector and so on, I got identical results for the
> different images, *except* that one was clearly weighted when it
> shouldn't be.  I tracked this down and have pushed a partial fix here:
>
>
> https://bitbucket.org/MatthewTurk/yt/commits/1e4ca6c3f7da6a6a7857660690c2207dab934652
>
> Unfortunately, I was unable to finish it because I don't know how the
> CallbackWrapper works completely, and it is necessary.  Here we need
> to regenerate the FRB from a different data_source, whereas by default
> the contours will use the extant FRB.  Nathan, do you know how we
> might fix this?  Can we just retain a reference to the window_plot in
> the wrapper?
>
> -Matt
>
> >
> > Matt
> >
> >>
> >> On Tue, Apr 1, 2014 at 1:32 PM, Matthew Turk <matthewturk at gmail.com>
> >> wrote:
> >>>
> >>> Hi Cameron,
> >>>
> >>> On Tue, Apr 1, 2014 at 4:11 PM, Cameron Hummels <chummels at gmail.com>
> >>> wrote:
> >>> > Hey everyone,
> >>> >
> >>> > I have an unusual problem that I cannot seem to work out.  I'm trying
> >>> > to
> >>> > generate a set of contours on an OffAxisProjectionPlot using the
> >>> > underlying
> >>> > data from another OffAxisProjectionPlot (i.e. a different
> data_source).
> >>> > I
> >>> > can get this working A-OK when I just use the ProjectionPlot class,
> but
> >>> > when
> >>> > I attempt to use the OffAxisProjectionPlot, I encounter problems.
> >>> >
> >>> > I made a small script demonstrating this problem, as well as the
> >>> > success
> >>> > when this method is applied to a ProjectionPlot object.  Any
> >>> > suggestions?
> >>> >
> >>> > Note, this is yt 2.x dev.  Here is the script:
> >>> >
> >>> > http://paste.yt-project.org/show/4459/
> >>> >
> >>> > Cameron
> >>>
> >>> My reading of your script is that you want "test4.png" to have
> >>> contours on it, right?  I just ran this script on 61e6b84f875c, which
> >>> is near tip of 2.x, and I got this image for test4.png:
> >>>
> >>> http://i.imgur.com/hfycSN9.png
> >>>
> >>> Is this not what you wanted?
> >>>
> >>> -Matt
> >>>
> >>> >
> >>> > --
> >>> > Cameron Hummels
> >>> > Postdoctoral Researcher
> >>> > Steward Observatory
> >>> > University of Arizona
> >>> > http://chummels.org
> >>> >
> >>> > _______________________________________________
> >>> > yt-users mailing list
> >>> > yt-users at lists.spacepope.org
> >>> > http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org
> >>> >
> >>> _______________________________________________
> >>> yt-users mailing list
> >>> yt-users at lists.spacepope.org
> >>> http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org
> >>
> >>
> >>
> >>
> >> --
> >> Cameron Hummels
> >> Postdoctoral Researcher
> >> Steward Observatory
> >> University of Arizona
> >> http://chummels.org
> >>
> >> _______________________________________________
> >> yt-users mailing list
> >> yt-users at lists.spacepope.org
> >> http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org
> >>
> _______________________________________________
> yt-users mailing list
> yt-users at lists.spacepope.org
> http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.spacepope.org/pipermail/yt-users-spacepope.org/attachments/20140401/1926ca72/attachment.html>