Hi Kathy,
On Thu, Dec 5, 2013 at 4:22 PM, Kathy Eastwood kathy.eastwood@nau.edu wrote:
Hi, Matt
I am still struggling with this issue of creating non-square plots. I can now create an FRB, and get plots from it, but I cannot seem to get anything except square plots. And because my matplotlib experience is "limited", to say the least, I'm not sure exactly what the square plot represents.
To be honest, I always struggle a bit when I have to do matplotlib operations like this -- so I completely understand! On the plus side, what it looks like is that you're at the point where you can manually manipulate the matplotlib plots pretty in depth.
So, here is one of the SlicePlots from before, 1 kpc wide by 2 kpc tall (even though I would like it to be even taller, and even though it cuts off the side of the plot window).
from yt.mods import * %matplotlib inline import matplotlib.pyplot as plt import matplotlib.colorbar as cb from matplotlib.colors import LogNorm
import numpy as np
pf = load("Strat_Box_hdf5_plt_cnt_0064") slc = SlicePlot(pf, 'x', 'Density',center='c',width=((1,"kpc"),(2.0,"kpc"))) slc.annotate_title('This is a Density plot') slc.save('test_2kpc.png') slc.show()
and here is the plot: http://i.imgur.com/wW64TYV.png
So, with the ultimate goal of creating an even taller plot with an FRB, and using your suggestions, I tried using an FRB, which is now working mostly
sl = pf.h.slice(2, pf.domain_center[2],fields=["Density","Temperature"] ) frb = sl.to_frb( (1.0, 'kpc'), (2048,1024),height=(2.0, 'kpc')) fig = plt.figure() ax = fig.add_axes([0.1,0.1,0.85,0.85])
I think this might be where it's making the square plot. The add_axes call accepts a rectangle definition which is defined as [left, bottom, width, height]. So the axes you're describing will in fact be square. I think you want it to have an aspect ratio of 2.0 and you'll want the figure to be that way too. So you'll need to change the figure call:
plt.figure((width, height))
(where width and height are in inches, which is typically pixel_size / dpi, where dpi is default 100) and then try with the usual axes call you had above. The axes width, height are both relative, so I think for a non-unitary aspect ratio figure, they can remain the same here.
Here's some documentation: http://matplotlib.org/api/figure_api.html#matplotlib.figure.Figure.add_axes
I think there are examples of this on matplotlib's website, but I wasn't able to find them off-hand.
You might also have luck doing this interactively, so that you can quickly iterate on the commands -- I think that the only ones that are going to be necessary to experiment with are fig = ... , ax = ..., ax.imshow and fig.savefig.
Hope that helps!
-Matt
ax.imshow(frb["Density"], origin="lower", norm = LogNorm(), clim = (1.0e-25,1.0e-22), cmap = "spectral")
fig.savefig("test_dens.png")
However, even given my current ("limited") matplotlib skills and doing a stretch with clim the best I could, this doesn't look right to me -- it seems to be square, and missing the parts above and below the galactic plane: http://i.imgur.com/OGQiHUh.png
Any suggestions?
thanks in advance, and for all your work on yt... kathy
On Wed, Nov 27, 2013 at 7:06 AM, Matthew Turk matthewturk@gmail.com wrote:
Hi Kathy,
Sorry you're having trouble!
On Tue, Nov 26, 2013 at 1:21 PM, Kathy Eastwood kathy.eastwood@nau.edu wrote:
Dear (very helpful) yt folk:
I have been struggling with trying to make a slice plot of some FLASH data that is an odd size; the volume is 1 kpc square in x and y, and 40 kpc tall in z. The grid sizes vary widely, as expected. I have two questions: one, I would like to understand the odd behavior that I get with regular SlicePLots, and I would like some help using the FixedResolutionBuffer, which I *suspect* is the way I want to go, but I have not been able to make it work. I am using the newest version of yt (upgraded this morning!) on Mac OS 10.7.5.
The SlicePlot weirdnesses: from yt.mods import * %matplotlib inline import matplotlib.pyplot as plt import matplotlib.image as mpimg import matplotlib.colorbar as cb import numpy as np pf = load("Strat_Box_hdf5_plt_cnt_0064") #First try a 20 kpc tall box slc = SlicePlot(pf, 'x', 'Density',center='c',width=((1,"kpc"),(20.0,"kpc"))) slc.annotate_title('This is a Density plot') slc.save('test_20kpc.png') slc.show() #Please see plot at http://i.imgur.com/N65CNfU.png
Whoa! That's pretty squished.
#Now try a 5 kpc tall plot slc = SlicePlot(pf, 'x', 'Density',center='c',width=((1,"kpc"),(5.0,"kpc"))) slc.annotate_title('This is a Density plot') slc.save('test_5kpc.png') slc.show()
#Please see plot at http://i.imgur.com/L1YORNK.png ; note that it seems to filling #a different part of the plotting "window"
#Now try a 2 kpc tall plot slc = SlicePlot(pf, 'x', 'Density',center='c',width=((1,"kpc"),(2.0,"kpc"))) slc.annotate_title('This is a Density plot') slc.save('test_2kpc.png') slc.show()
#Please see plot at http://i.imgur.com/Bq1WaIP.png ; note that it seems to filling #even MORE of the plotting "window", but that the density labels were cut off a bit.
Okay, so I *think* I might have an idea what's going on here. I've replicated the problem here, but I don't have any ideas how to fix it. It's basically just that (automatically) coming up with the right aspect ratio for the *figure* is tricky when the image is not the right aspect ratio. I think Devin Silvia, Nathan Goldbaum and Kacper Kowalik have all worked hard on this but not had a lot of success. It also looks to me like the FRB may have a uniform aspect ratio (800x800) which may be contributing to the problem. So, on to FRBs ...
OK, now the part that doesn't work. After reading, I decided that an frb must be the way to go, so I tried frb = FixedResolutionBuffer(slc, (0.0, 1.0, 0.0, 1.0), (1024,1024)) plt.imshow(frb['Density']) plt.savefig('test_figure.png')
and I get the following error. Note: I also tried slc.to_frb and that didn't work either.
AttributeError Traceback (most recent call last) <ipython-input-17-e68719d58ae4> in <module>() ----> 1 frb = FixedResolutionBuffer(slc, (0.0, 1.0, 0.0, 1.0), (1024,1024)) 2 plt.imshow(frb['Density']) 3 plt.savefig('test_figure.png') 4 5
/Users/kde/yt-x86_64/src/yt-hg/yt/visualization/fixed_resolution.pyc in __init__(self, data_source, bounds, buff_size, antialias, periodic) 89 self.antialias = antialias 90 self.data = {} ---> 91 self.axis = data_source.axis 92 self.periodic = periodic 93
AttributeError: 'SlicePlot' object has no attribute 'axis'
Ah, this one we can fix. Here what the FRB wants is actually the slice itself, not the slice plot. There are two ways to do this. The first is to grab the *existing* slice object:
sl = slc.data_source
and feed sl into a FixedResolutionBuffer creation. But I think, since you're going to avoid all of the slice plot machinery altogether, it'd probably make more sense to take control of it with the data object machinery directly. So you can create a slice, which is a data source, like this:
sl = pf.h.slice(2, pf.domain_center[2])
What I'm doing here is saying axis 2 (z) and at the z-th coordinate of the domain center. Now if we query sl, we get all of the data points; this isn't what we want, since they are all different resolutions, so we use the FRB. sl objects have a handy way of making FRBs:
sl.to_frb( ...
this takes a width (which can be a tuple), a resolution, and optionally a height. help(sl.to_frb) will show the arguments and an example. Now the FRB can be queried like you would a data object, and it gives back (N,N) arrays. Hope that helps, and let us know!
-Matt
PS Those are some gorgeous simulation results, by the way.
THanks in advance for any help. cheers kathy
-- Kathy DeGioia Eastwood, Ph.D. Professor of Physics and Astronomy Northern Arizona University Flagstaff, AZ 86011-6010 Ph: 928-523-7159 FX: 928-523-1371 Kathy.Eastwood@nau.edu deliveries: 602 S. Humphreys St., Bldg 19 Rm 209
yt-users mailing list yt-users@lists.spacepope.org http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org
-- Kathy DeGioia Eastwood, Ph.D. Professor of Physics and Astronomy Northern Arizona University Flagstaff, AZ 86011-6010 Ph: 928-523-7159 FX: 928-523-1371 Kathy.Eastwood@nau.edu deliveries: 602 S. Humphreys St., Bldg 19 Rm 209