Smooth parameter must be a (not-necessarily integer) number of pixels

[drphilmarshall]

Seems to be in physical units at the moment.

[dfm]

I don't quite understand your comment... it should be in units of pixels already. The main annoyance is that currently it must be the same in every dimension.

[drphilmarshall]

Huh - that's good! But it must be some other behaviour I'm seeing then,
perhaps that the kwargs aren't getting passed though properly. Did you see
my attempt to upgrade the Dustin and my Mayacamas notebook? Feel free to
close this, apologies for jumping to conclusions!

On Mon, Apr 6, 2015 at 9:21 PM, Dan Foreman-Mackey <notifications@github.com

wrote:

I don't quite understand your comment... it should be in units of pixels
already. The main annoyance is that currently it must be the same in every
dimension.

—
Reply to this email directly or view it on GitHub
#45 (comment).

[dfm]

Can you post an example here where you're not getting the desired behavior?

[drphilmarshall]

Yeah - check out this notebook. It seems not to want to obey my smoothing and contour filling instructions!

[drphilmarshall]

Oh, and here's my forked and tweaked version of the demo.py script, where I
seem to have gotten the idea that the smoothing length was in physical
rather than image units...

[dfm]

In the notebook, it looks like you're not importing the updated version at all!

[dfm]

Make sure that you've installed the git master version somewhere where the notebook can find it. You'll also need to restart the kernel.

[drphilmarshall]

Thanks Dan - as predicted, I had an old pip-installed triangle_plot that
was dominating my python path. With this removed, the triangle plots are
now up to date in the straight line notebook. There's some odd behaviour
with the filled contours, see what you think of the with-scatter results
plot, in the s dimension.
I feel like I already solved this in CornerPlotter...

On Mon, Apr 6, 2015 at 10:07 PM, Dan Foreman-Mackey <
notifications@github.com> wrote:

Make sure that you've installed the git master version somewhere where the
notebook can find it. You'll also need to restart the kernel.

—
Reply to this email directly or view it on GitHub
#45 (comment).

[dfm]

If you find the method you use to deal with cases like this, let me know cause I've never been able to find anything.

[dfm]

Never mind! I figured it out. Try pulling from master and making that plot again.

[drphilmarshall]

Ooh! I see what you did :-)
That almost works - just need to connect the dots the right way, now (see
the second triangle plot in this notebook

On Tue, Apr 7, 2015 at 11:10 AM, Dan Foreman-Mackey <
notifications@github.com> wrote:

Never mind! I figured it out. Try pulling from master and making that plot
again.

—
Reply to this email directly or view it on GitHub
#45 (comment).

[dfm]

How about now?

[drphilmarshall]

That's fixed it! Nice work Daniel - thank you! :-)

On Tue, Apr 7, 2015 at 6:36 PM, Dan Foreman-Mackey <notifications@github.com

wrote:

How about now?

—
Reply to this email directly or view it on GitHub
#45 (comment).