matplotlib-users Mailing List for matplotlib

2011/2/22 Benjamin Root <ben...@ou...>:
>
> Admittedly, this isn't using matplotlib's hist() function because it only
> allows for one color per dataset.  However, you can use numpy's histogram
> function to get the bins and counts yourself, and then use bar() to make the
> bars.  bar() will allow you to color the bars individually.

Pylab hist() returns a list of patches so you can also change their properties.

Goyo

2011/2/22 Daniel Mader <dan...@go...>:
> Hi,
>
> there has been a similar question recently but I couldn't figure out
> if or how this is solved:
>
> I'd like to reduce the figure size so that I can add it to a LaTeX
> document without scaling (PDF output with LaTeX font rendering). For
> that, I need to adapt the font sizes, too.
>
> Unfortunately, the canvas is not properly scaled so that the axis
> labels and the possibly the tick marks are cut off.
>
> Is this a bug, feature, design flaw? How can I properly work around
> it, i.e. reduce the graph automatically for a given figsize/font size
> combination so that everything fits on the figure?

You'll have to understand how dimensions are calculated and then use
stuff like Figure.subplots_adjust.

http://matplotlib.sourceforge.net/api/figure_api.html#matplotlib.figure.Figure.subplots_adjust
http://matplotlib.sourceforge.net/api/figure_api.html#matplotlib.figure.SubplotParams
http://matplotlib.sourceforge.net/faq/howto_faq.html#automatically-make-room-for-tick-labels

As a fast and dirty trick you can pass big numbers to subplots_adjust
and then use bbox_inches='tight' in savefig.

Goyo

> 
> 
> On Wed, Feb 23, 2011 at 2:09 PM, bhargav vaidya <coo...@gm...> wrote:
> Hello Matplotlib Users.
> 
> I am trying to work my way out to interpolate a surface from polar coordinates (R,Theta,Z) to rectangular co-ordinates(X,Y,Z)
> Basically I am looking for an equivalent for POLAR_SURFACE.pro routine in IDL using matplotlib/Scipy?
> 
> http://idlastro.gsfc.nasa.gov/idl_html_help/POLAR_SURFACE.html
> 
> Does anyone of you have done that before or could tell me what functions can I use to get that?
> 
> Regards
> Bhargav Vaidya
> 
> 
> (Quick Note: technically, you have cylindrical coordinates because you have Z, not Phi)
> 
> While I am sure there are some more optimal methods, the straight-forward way is to do something like the following:
> 
> import numpy as np
> from scipy.interpolate import griddata
> 
> orig_x = R*np.cos(Theta)
> orig_y = R*np.sin(Theta)
> orig_z = Z
> 
> grid_x, grid_y, grid_z = np.mgrid[orig_x.min():orig_x.max():10j,
>                                                 orig_y.min():orig_y.max():10j,
>                                                 orig_z.min():orig_z.max():10j]
> 
> new_vals = griddata((orig_x, orig_y, orig_z), orig_vals, (grid_x,  grid_y, grid_z)
> 
> 
> After running this, you should have new_vals which should have the same shape as grid_x.  You can also specify the interpolation method used by griddata through the 'method' keyword argument.  The '10j' in the np.mgrid[] call merely specifies the number of points you want in that axis and can be changed to any other integer you want (just keep the 'j' there to make it work the way we want).
> 
> I hope that helps!
> Ben Root
> 



Hello Benjamin 

Thanks for the reply... I was a little vague in my first  question before...
Here is my specific problem

I have a 3D matrix say of density (rho [ r, theta , phi] ) 
where r, theta ,phi have (64,32,32) elements
theta and phi values are in radians.

I have to interpolate rho(r, theta) at some phi to rho(x,z) and rho(r, phi) at some theta to rho(x,y) ....
However I used the a different code then the one you mentioned .. I am using the scipy.ndimage.geometric_transform
I am attaching the figure obtained using a code below 


THE PART OF THE CODE TO CONVERT THE POLAR TO CARTESIAN IS GIVEN BELOW

I have the following problems :

1. I am not happy with the interpolation. This is partly because I am using pcolormesh for plotting data as imshow is not working for my irregularly spaced grid. I have also tried higher order  for the geometric -transform  but the difference is not that appreciable. Can I improve the interpolation is some way?

2. In the right panels of the plot above, I dont get the radial and vertical co-ordinate as the right cartesian values.. I think this can be solved if I give the right x y and z values to pcolormesh ... but I am not sure how to get the right values for them so that they are consistent with the transform.



import numpy as np
import scipy as S
import scipy.ndimage

def get_polar_plot(self,var,**kwargs):

		def r2theta(outcoords, inputshape, origin):
			xindex, yindex = outcoords
			x0, y0 = origin
			x = xindex - x0
			y = yindex - y0
			r = np.sqrt(x**2 + y**2)
			theta = np.arctan2(np.pi-y, x)
			theta_index = np.round((theta+ np.pi) * inputshape[1] / (1.015*np.pi))
			return (r,theta_index)

		def r2phi(outcoords, inputshape, origin):
			xindex, yindex = outcoords
			x0, y0 = origin
			x = xindex - x0
			y = yindex - y0
			r = np.sqrt(x**2 + y**2)
			phi = np.arctan2(y, x)
			phi_index = np.round((phi + np.pi) * inputshape[1] / (2.1*np.pi))
			return (r,phi_index)

		x1 = kwargs.get('x1') # This is Radial co-ordinate
		x2 = kwargs.get('x2') # This is Either Theta or Phi
		data_arr2d = var

		if kwargs.get('rtheta',False) == True:
			var_cart = S.ndimage.geometric_transform(data_arr2d, r2theta,order=1,output_shape = (data_arr2d.shape[0]*2, data_arr2d.shape[0]),extra_keywords = {'inputshape':data_arr2d.shape,'origin':(data_arr2d.shape[0],0)})

		if kwargs.get('rphi',False) == True:
			var_cart = S.ndimage.geometric_transform(data_arr2d, r2phi, order=1,output_shape = (data_arr2d.shape[0]*2, data_arr2d.shape[0]*2),extra_keywords = {'inputshape':data_arr2d.shape,'origin':(data_arr2d.shape[0],data_arr2d.shape[0])})

		return var_cart


Hope you will help me in this regard.

Regards
Bhargav

On Wed, Feb 23, 2011 at 6:56 PM, Jack Sankey <jac...@gm...> wrote:

> Hi everyone!
>
> Is there any backend-independent way to get and set the figure window
> coordinates and sizes (i.e. in pixels)? Currently when I make new plots they
> appear in random locations as determined by my operating system; during
> analysis when I'm making and destroying a lot of plot windows I want to be
> able to have them automatically appear in "good" locations (such as the
> upper left corner of my screen).
>
> Also, I am trying to find ways to get/set the z-order.
>
> Thanks in advance,
> Jack
>
>
There was someone who found that it was possible to add some additional
information to the creation of a figure window.  This information (at least
on Linux systems), could be taken advantage of by customized configurations
of the desktop manager.  I would be curious to see what has come of that
effort.

Ben Root

On Wed, Feb 23, 2011 at 10:52 PM, Mark Janikas <mja...@es...> wrote:

>  Hello All,
>
>
>
> Is there a way to add HTML links into output graphics?  In short, I would
> like to use something like the PLT.text command and have it link to an html
> page.  The url arg doesn’t seem to be what I am looking for.
>
>
>
> My final output medium is PDF… so I am using:
>
>
>
> from matplotlib.backends.backend_pdf import PdfPages as PDF
>
> import matplotlib.pyplot as PLT
>
>
>
>
>
> pdfOutput = PDF(fileName)
>
> PLT.text(“Some Text to Link”, url = “http:…”)
>
> …
>
> …
>
> PLT.savefig(pdfOutput, format='pdf')
>
> PLT.clf()
>
>
>
>
>
> Thanks so much for any advise!
>
>
>
> MJ
>
>
Looks like the url string is only used for SVG files.  If you could, please
file a feature request to have it added to the pdf backend:

https://sourceforge.net/tracker/?group_id=80706

Ben Root

Jason,

Try adding a pyqt property 'facecolor' to your widget.
(This way you can set it also through the designer.)

Add something like this to the widget's __init__ method:

figFacecolor = QtCore.pyqtProperty(str, getfigfacecolor,
                                setfigfacecolor, resetfigfacecolor)

And then define the three methods.

Nadia

I haven't seen this done before so I don't know if there's a standard
way. The idea seems to be to take some points which are real data,
create a random variable for each point with the points' position as
the mean, then choose some number of points from each distribution to
create some new points clustered around the original data. Some
examples online seem to use uniform distributions and Poisson
distributions or mixtures of these (uniform for the x-variable and
Poisson for the y). If my take on this is correct, you can use
scipy.stats to do this - an example is in the attached file which
creates Gaussian distributions for each of the x and y coordinates
then creates an equal number of new points for each of the seed
points. The online examples I saw seem to choose random numbers of new
points for each seed point. I didn't bother trying to cover all the
possibilities. Hopefully this is helpful,

Gary R.

On Thu, Feb 24, 2011 at 3:04 PM, Uri Laserson <las...@mi...> wrote:
> Hi all,
> I am interested in jittering points in a plot.  I searched the forum, but I
> am amazed at the dearth of results on the topic.  I am referring to
> something like this:
> http://goo.gl/Db47s
> or
> http://goo.gl/BjIZt
>
> Is there a standard way people do this with MPL?
> Thanks!
> Uri
> ...................................................................................
> Uri Laserson
> Graduate Student, Biomedical Engineering
> Harvard-MIT Division of Health Sciences and Technology
> M +1 917 742 8019
> las...@mi...

Hello All,

Is there a way to add HTML links into output graphics?  In short, I would like to use something like the PLT.text command and have it link to an html page.  The url arg doesn't seem to be what I am looking for.

My final output medium is PDF... so I am using:

from matplotlib.backends.backend_pdf import PdfPages as PDF
import matplotlib.pyplot as PLT


pdfOutput = PDF(fileName)
PLT.text("Some Text to Link", url = "http:...")
...
...
PLT.savefig(pdfOutput, format='pdf')
PLT.clf()


Thanks so much for any advise!

MJ

Mark Janikas
Product Developer
ESRI, Geoprocessing
380 New York St.
Redlands, CA 92373
909-793-2853 (2563)
mja...@es...<mailto:mja...@es...>

Hi all,

I am interested in jittering points in a plot.  I searched the forum, but I
am amazed at the dearth of results on the topic.  I am referring to
something like this:
http://goo.gl/Db47s
or
http://goo.gl/BjIZt

Is there a standard way people do this with MPL?

Thanks!
Uri

...................................................................................
Uri Laserson
Graduate Student, Biomedical Engineering
Harvard-MIT Division of Health Sciences and Technology
M +1 917 742 8019
las...@mi...

Hi everyone!

Is there any backend-independent way to get and set the figure window
coordinates and sizes (i.e. in pixels)? Currently when I make new plots they
appear in random locations as determined by my operating system; during
analysis when I'm making and destroying a lot of plot windows I want to be
able to have them automatically appear in "good" locations (such as the
upper left corner of my screen).

Also, I am trying to find ways to get/set the z-order.

Thanks in advance,
Jack

On Wed, Feb 23, 2011 at 4:52 PM, Rene Breton <sup...@gm...>wrote:

> Thanks much for the quick answer Ben!
>
> At least now I know it doesn't exist, yet. I might dig into the source
> code to change the color as you suggest. As a workaround, maybe I will
> simply tweak the .ps output in Adobe Illustrator; should be easy enough
> to remove such thing.
>

Another issue may be the tickmarks, which are made separately in another
segment of code.

>
> By the way, I'm very glad you (and people) took over the mplot3d. I
> really think that 3D plotting was the number one missing thing in
> Matplotlib and Python more generally. (mayavi2 is way to complicated and
> clunky for what most people need)
>
>
Thank you for your vote of confidence.  This is exactly the intent of
mplot3d, which is to allow for the possibility of basic 3d plotting features
without the need to go "all-in" to a major 3d plotting system.  We are still
a long way from what I envision for mplot3d.


> If I may make another suggestion, besides getting more control over the
> axes spines, it would be to allow the axes rendering to "mix" multiple
> plot objects. Let's say one plots a sphere and a cylinder (like a rod)
> that goes through the sphere and get the part of the cylinder that lies
> inside the sphere to be hidden. Of course, I realize it might not be
> trivial to do, depending exactly how plots are rendered.
>
>
You are not the first to request this, and you will not be the last.
Unfortunately, it is an intractable problem that can never be completely
solved by the 2d rendering framework employed by matplotlib.  If/When
matplotlib gains an opengl backend, this problem will be revisited.

Please do continue to feel free to continue commenting on mplot3d and how
you think it can be better.  And, as always, patches are welcome!

Ben Root