matplotlib pcolormesh plot from x,y,z data

Question

I have data in a textfile in tableform with three columns. I use np.genfromtxt to read all the columns into matplotlib as x, y, z.

I want to create a color meshplot

Answer 1

It seems you are plotting X and Y as 2D arrays while Z is still a 1D array. Try something like:

Znew=np.reshape(z,(len(xmesh[:,0]),len(xmesh[0,:])))
diagram1.pcolormesh(xmesh,ymesh,Znew) 

Update: Tou have a X/Y grid of size 4x4:

x = np.genfromtxt('mesh.txt', dtype=float, delimiter=' ', usecols = (0))
y = np.genfromtxt('mesh.txt', dtype=float, delimiter=' ', usecols = (1))
z = np.genfromtxt('mesh.txt', dtype=float, delimiter=' ', usecols = (2))

Reshape the arrays as suggestet by @Gustav Larsson and myself like this:

Xnew=np.reshape(x,(4,4)) 
Xnew=np.reshape(y,(4,4))
Znew=np.reshape(z,(4,4))

Which gives you three 4x4 arrays to plot using pcolormesh:

diagram1.pcolormesh(Xnew,Ynew,Znew) 

Answer 2

In the example data provided above, x, y, and z can be easily reshaped to get 2D array. The answer below is for someone who is looking for more generalized answer with random x,y, and z arrays.

import matplotlib.pyplot as plt
from matplotlib.mlab import griddata
import numpy

# use your x,y and z arrays here
x = numpy.random.randint(1,30, 50)
y = numpy.random.randint(1,30, 50)
z = numpy.random.randint(1,30, 50)

yy, xx = numpy.meshgrid(y,x)
zz = griddata(x,y,z,xx,yy, interp='linear')
plt.pcolor(zz)
#plt.contourf(xx,yy,zz) # if you want contour plot
#plt.imshow(zz)
plt.pcolorbar()
plt.show()

Answer 3

I had the same problem and agree with Gustav Larsson's suggestion to use

scatter(x, y, c=z)

In my particular case, I set the linewidths of the scatter points to zero:

scatter(x, y, c=z, linewidths=0)

of course, you can play around with other decorations, color schemes etc., the reference of matplotlib.pyplot.scatter will help you further.

Answer 4

My guess is that x, y and z will be read as one-dimensional vectors of the same length, let's say N. The problem is that when you create your xmesh and ymesh, they are N x N, which your z values should be as well. It's only N, which is why you are getting an error.

What is the layout of your file? I'm guessing each row is a (x,y,z) that you want to create a mesh from. In order to do this, you need to know how the points are ordered as a mesh (either as row-major or column-major). Once you know this, instead of creating xmesh and ymesh, you can do something like this:

N = np.sqrt(len(x)) # Only if squared, adjust accordingly
x = x.reshape((N, N))
y = y.reshape((N, N))
z = z.reshape((N, N))
pcolormesh(x, y, z)

Before doing this, I would start by doing this:

scatter(x, y, c=z)

which will give you the points of the mesh, which is a good starting point.