matplotlib-devel Mailing List for matplotlib

On 2012年08月29日 10:07 AM, David Raymond wrote:
>
> I think that I have found a serious bug in vector plotting (quiver)
> when the "angle=xy" option is used on a plot with a large aspect
> ratio. It is my understanding that with this option, the vector with
> tail at (x,y) would have its head at (x + vx,y + vy); at least this is
> what it says in the documentation.
Not a bug.
Note this in the documentation:
 *scale_units*: *None*, or any of the *units* options.
 For example, if *scale_units* is 'inches', *scale* is 2.0, and
 ``(u,v) = (1,0)``, then the vector will be 0.5 inches long.
 If *scale_units* is 'width', then the vector will be half the width
 of the axes.
 If *scale_units* is 'x' then the vector will be 0.5 x-axis
 units. To plot vectors in the x-y plane, with u and v having
 the same units as x and y, use
 "angles='xy', scale_units='xy', scale=1".
Try this modified call to quiver:
q = plt.quiver(X, Y, vx, vy, angles='xy', scale_units='xy')
Eric

On 2012年08月29日 10:07 AM, David Raymond wrote:
>
> I think that I have found a serious bug in vector plotting (quiver)
I think you are correct, but a quick look at the code has not yet 
revealed what is going wrong. I will look into it.
Eric
> when the "angle=xy" option is used on a plot with a large aspect
> ratio. It is my understanding that with this option, the vector with
> tail at (x,y) would have its head at (x + vx,y + vy); at least this is
> what it says in the documentation.
>
> The problem is best illustrated by the following example:
>
> ------------
>
> #!/usr/bin/python
> #
> from numpy import *
> import matplotlib.pyplot as plt
>
> lx = 100.
> ly = 10.
> pi = 3.14159
> kx = pi/lx
> ky = pi/ly
> nx = 21
> ny = 21
> x = linspace(0,lx,nx)
> y = linspace(0,ly,ny)
> (X,Y) = meshgrid(x,y)
>
> # this is a streamfunction
> psi = sin(kx*X)*sin(ky*Y)
>
> # these are the velocity components derived from the streamfunction
> vx = ky*sin(kx*X)*cos(ky*Y)
> vy = -kx*cos(kx*X)*sin(ky*Y)
>
> # plot the x velocity
> cx = plt.contourf(x,y,vx)
> b = plt.colorbar(cx, orientation='vertical')
> tl = plt.title("vx")
> xlab = plt.xlabel("x")
> ylab = plt.ylabel("y")
> plt.show()
>
> # plot the y velocity
> cy = plt.contourf(x,y,vy)
> b = plt.colorbar(cy, orientation='vertical')
> tl = plt.title("vy")
> xlab = plt.xlabel("x")
> ylab = plt.ylabel("y")
> plt.show()
>
> # plot the streamfunction and the velocity vectors using the angles=xy option
> cp = plt.contour(x,y,psi)
> q = plt.quiver(X,Y,vx,vy,angles='xy')
> tl = plt.title("psi contours, (vx,vy) vectors")
> xlab = plt.xlabel("x")
> ylab = plt.ylabel("y")
> plt.show()
>
> ------------------------------------
>
> The contour plot of vy, the y component of the vector, clearly shows
> that vy is non-zero at y = 5. However, the vector plot has it zero
> along this line. Interestingly, vx appears to be represented correctly
> on the x = 50 line. The magnitude of vectors should be inversely
> proportional to the spacing of streamfunction contours, and this is
> manifestly untrue in the vector plot.
>
> Please tell me if I am doing something stupid.
>
> Dave
>
> PS: I am using matplotlib 1.1.1 and numpy 1.6.2 on Arch linux. (Yes,
> I have to set the #!/bin/python line in the example to #!/bin/python2
> on Arch, since Arch has python3 as default!)
>

I think that I have found a serious bug in vector plotting (quiver)
when the "angle=xy" option is used on a plot with a large aspect
ratio. It is my understanding that with this option, the vector with
tail at (x,y) would have its head at (x + vx,y + vy); at least this is
what it says in the documentation.
The problem is best illustrated by the following example:
------------
#!/usr/bin/python
#
from numpy import *
import matplotlib.pyplot as plt
lx = 100.
ly = 10.
pi = 3.14159
kx = pi/lx
ky = pi/ly
nx = 21
ny = 21
x = linspace(0,lx,nx)
y = linspace(0,ly,ny)
(X,Y) = meshgrid(x,y)
# this is a streamfunction
psi = sin(kx*X)*sin(ky*Y)
# these are the velocity components derived from the streamfunction
vx = ky*sin(kx*X)*cos(ky*Y)
vy = -kx*cos(kx*X)*sin(ky*Y)
# plot the x velocity
cx = plt.contourf(x,y,vx)
b = plt.colorbar(cx, orientation='vertical')
tl = plt.title("vx")
xlab = plt.xlabel("x")
ylab = plt.ylabel("y")
plt.show()
# plot the y velocity
cy = plt.contourf(x,y,vy)
b = plt.colorbar(cy, orientation='vertical')
tl = plt.title("vy")
xlab = plt.xlabel("x")
ylab = plt.ylabel("y")
plt.show()
# plot the streamfunction and the velocity vectors using the angles=xy option
cp = plt.contour(x,y,psi)
q = plt.quiver(X,Y,vx,vy,angles='xy')
tl = plt.title("psi contours, (vx,vy) vectors")
xlab = plt.xlabel("x")
ylab = plt.ylabel("y")
plt.show()
------------------------------------
The contour plot of vy, the y component of the vector, clearly shows
that vy is non-zero at y = 5. However, the vector plot has it zero
along this line. Interestingly, vx appears to be represented correctly
on the x = 50 line. The magnitude of vectors should be inversely
proportional to the spacing of streamfunction contours, and this is
manifestly untrue in the vector plot.
Please tell me if I am doing something stupid.
Dave
PS: I am using matplotlib 1.1.1 and numpy 1.6.2 on Arch linux. (Yes,
I have to set the #!/bin/python line in the example to #!/bin/python2
on Arch, since Arch has python3 as default!)
-- 
David J. Raymond
Prof. of Physics
New Mexico Tech
http://www.physics.nmt.edu/~raymond/index.html

thanks, Keith, looks like Michael fixed the apostrophe. But not the
misspelling of Eliot's last name. I've update it now.
On Tue, Aug 28, 2012 at 8:50 AM, <kei...@bt...> wrote:
> "TS Elliots’ maxim" should be "T. S. Eliot’s maxim" J
>
>
>
> Keith
>
>
>
>
>
>
> ------------------------------------------------------------------------------
> Live Security Virtual Conference
> Exclusive live event will cover all the ways today's security and
> threat landscape has changed and how IT managers can respond. Discussions
> will include endpoint security, mobile security and the latest in malware
> threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
> _______________________________________________
> Matplotlib-devel mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-devel
>
-- 
Paul Ivanov
314 address only used for lists, off-list direct email at:
http://pirsquared.org | GPG/PGP key id: 0x0F3E28F7

On Wed, Aug 22, 2012 at 6:40 AM, Eric Firing <ef...@ha...> wrote:
> Correction: now I can't reproduce what I thought I was seeing; plt.gca()
> is returning an AxesSubplot as it should. Maybe the problem is in the
> axes_grid1 toolkit. It is appearing in the last figure of the
> tight_layout tutorial in the docs.
For axes which uses axes_locator, tight_layout works if the
axes_locator have associated subplotspec.
And, only allowing instance of SubplotBase is too strict.
The PR below addresses this issue.
https://github.com/matplotlib/matplotlib/pull/1170
And it will work again for the axes_grid1 cases.
Regards,
-JJ