PlotBasics: Difference between revisions

From CASA Guides
Jump to navigationJump to search
Line 11: Line 11:
3.  You can define plot objects, and then use the pyplot methods on those objects.
3.  You can define plot objects, and then use the pyplot methods on those objects.

The last way gives you most control, but the other two are somewhat easier.  The Supernova Cosmology Example below uses the second way.
The last way gives you most control, but the other two are somewhat easier.  The examples below uses the second way.

=== Supernova Cosmology Example ===
=== Supernova Cosmology Example ===

Revision as of 12:01, 8 November 2011

Back to the PythonOverview.

The Three Ways to plot

There are three ways to go about plotting in matplotlib.

1. You can use the pylab environment

2. You can use the matplotlib.pyplot environment, with plotting commands and functions.

3. You can define plot objects, and then use the pyplot methods on those objects.

The last way gives you most control, but the other two are somewhat easier. The examples below uses the second way.

Supernova Cosmology Example

  1. We will import the pyplot and numpy packages
import numpy as np
import matplotlib.pyplot as plt

Reading Data from the Web

We are going to download data from the internet so we will import the urllib package

import urllib

Begin by reading the Union2 SN cosmology data from LBL, because they are fun.

SN_list = []          
z_array = np.array([])
mod_array = np.array([])
moderr_array = np.array([])
f = urllib.urlopen('')
for line in f:
    if line[0] == '#': continue    # Ignore anything that starts with a #
    SN, z, mod, moderr = line.split()
    z_array = np.append(z_array,np.float64(z))
    mod_array = np.append(mod_array,np.float64(mod))
    moderr_array = np.append(moderr_array,np.float64(moderr))   

Simple Plots

Now to the plotting. First we close whatever windows we might have:


Now let us plot some points:

plt.plot(z_array, mod_array)

Notice it is a mess; by default it connects the lines. We will close it and start over.


We clearly need a title and axes labels

plt.title("Union2 SN Cosmology Data")
plt.xlabel('z', fontsize=20)

But we want a Greek Letter, so we can put some LaTeX syle code with the r command:

plt.ylabel(r'$\mu=m-M$', fontsize=20)

Now we add a format string to the plot command:

plt.plot(z_array, mod_array,'ro')

Now this looks better, we have red circles. The ro string, is what told it to give us red circles. A b. would give us blue dots, or ys would give yellow squares.

Now we start a new plot:


Plotting with Error Bars

Now let's plot with the error bars

plt.errorbar(z_array, mod_array, yerr=moderr_array, fmt='.')

And putting labels and colors we can do:

plt.xlabel(r'$z$', fontsize=20)
plt.ylabel(r'$\mu=m-M$', fontsize=20)
plt.title("Union2 SN Cosmology Data")
plt.errorbar(z_array, mod_array, yerr=moderr_array, fmt='.', capsize=0,
    elinewidth=1.0, ecolor=(0.6,0.0,1.0), color='green' )

Notice that colors can be specified with a fmt or a color keyword. With the color keyword they can be given via an RGB tuple, an RGB hex code, a name, or a single number between '0.0' and '1.0' for gray scale.


Plotting Histograms

Now there are a lot of points, so let's figure our what our distribution is in z:

plt.hist(z_array, 25)
# And slap a label on it
plt.xlabel(r'$z$', fontsize=20)

Plotting With Asymmetrical Error Bars

Now, let's find the "real" distance from the distance modulus. To do this we will define a function and a constant.

def distance_Mpc(m,z):
    return 0.00001 * (10**(m/5)) / (1.0 + z)
c = 299792.458    # km/s

So the distance can be found with:

d_array = distance_Mpc(mod_array,z_array)

Now we will calculate the error bars in the distance, both ways:

d_error_plus  = ( distance_Mpc((mod_array+moderr_array),z_array) - d_array )
d_error_minus = ( d_array - distance_Mpc((mod_array-moderr_array),z_array) )

And plot the graph with asymetrical horizontal error bars, and labels Notice the different color formats that can be used.

plt.errorbar(z_array, d_array, yerr=(d_error_minus,d_error_plus), fmt='s',
    capsize=5, elinewidth=1.0, color=(0.4,0.0,1.0), ecolor='aqua', barsabove=True)
plt.xlabel('z', fontsize=15, color='0.0')
plt.ylabel('Distance (Mpc)', fontsize=15, color='g')
plt.title("Union2 SN Cosmology Data", color=(0.4,0.0,1.0))

Plotting With a Second Axis

Now we will make a plot with both Mpc and Gly on the same plot.

First we read the existing plot limits into an array

axes1_range = np.array( plt.axis() ) 

We want to use the same x-axis so we use the function.

<source lang="Python">
plt.twinx()   #  This swaps the Y axis
plt.ylabel('Distance in Billions of Light Years', fontsize=15, color='purple')

Now we will make the second axes the right scale.

axes2_range = axes1_range.copy()  # Don't forget, we need to make a copy of it.
axes2_range[2:4] = 3.26*axes1_range[2:4]/1000.0  # set second y-axis to Gly  

Plotting With a Multiple Axes and Annotation

Now imagine we want to make a plot for the public, so we need to find the distance in light years. Also, we want to plot cz.

plt.errorbar(3.26*d_array/1000.0, c*z_array,
    xerr=(3.26*d_error_minus/1000.0, 3.26*d_error_plus/1000.0),
    fmt='.', capsize=0, elinewidth=1.0, color=(0.4,0.0,1.0), ecolor='aqua',
    barsabove=False, zorder=2)

Now we make room for each axis

plt.subplots_adjust(right=0.875, top=0.8)

And plot the labels.

plt.ylabel('cz (km/s)', fontsize=15, color='aqua')
plt.xlabel('Distance (Billions of Light Years)', fontsize=15, color='aqua')

Now, you want to make the plot also include z and Mpc, so we need to convert our axes:

axes1_range = np.array( plt.axis() ) # get the existing axes and convert to an array
axes2_range = axes1_range.copy()  # Don't forget, we need to make a copy of it.
axes2_range[0:2] = 1000*axes1_range[0:2]/3.26  # set second x-axis to Mpc  
axes2_range[2:4] = axes1_range[2:4]/c # set second y-axis to z

Now we switch to the second axis

plt.twinx()   #  This swaps the Y axis
plt.ylabel('z', fontsize=15, color='r')  #  I am not sure why this has to be before plt.twiny()
plt.twiny()   #  This swaps the X axis 
plt.xlabel('Distance (Mpc)', fontsize=15, color='purple')
plt.title("Union2 SN Cosmology Data", color=(0.4,0.0,1.0), x=0.5, y=1.15)

And we will put Hubble's law on it just for fun. Using the zorder keyword, we can make sure the line is plotted on top on the points.

H = 70.8  # km/s/Mpc
z_limits = np.array([0.0,1.0])  #  Plot is relative to the new axes
d_limits = z_limits*c/H
plt.plot(d_limits,z_limits, 'y-', zorder=1)

Notice that the second plot rescaled the axes, which is a problem for this plot, because both x axes and both y axes represent the same physical quantity. So we reissue the axis function:

plt.axis(axes2_range)  # this needs to be after the plot command

Now we will put an arrow on the graph and a label.

plt.annotate("Hubble's law", xy=(d_limits[1],z_limits[1]), color='y', zorder=3, 
    xytext=(1.2*d_limits[1],1.2*z_limits[1]), arrowprops=dict(facecolor='y', shrink=0.05))

Saving as a PDF

We can save a file as a PDF or most any other common format this way:

plt.savefig('Union2_plot.pdf', format="pdf", transparent=True, bbox_inches='tight')

It is important to note that some file formats are raster graphics and others are scalable vector graphics.

Go back to the PythonOverview page.

Magnetic Needle Example

Here we show another plotting example, where we calculate the magnetic field surrounding a thin magnetic needle.

First we import matplolib and numpy:

import matplotlib
import numpy as np
import matplotlib.pyplot as plt

Next we define the grid size, length or needle, and the size of circular masks around the poles.

N = 20      # Size of NxN array.
R = 0.25    # Size of the two circular masks.
l = 1.0     # Length of the rod.
l2 = l/2.0  # Saves on typing l/2.0 all the time.

Now we will define min and max in the cylindrically radial coordinate (s), and the axial coordinate (z).

# Define s and z
s_delt = 1.0/(N/2)
s_max =  1*(1.0 + 0.5*s_delt)
s_min =  -1*(1.0 - 0.5*s_delt)
# Square Axes
z_delt = 1.0/(N/2)
z_min = -1*(1.0 - 0.5*z_delt)
z_max =  1*(1.0 + 0.5*z_delt)

Now we will find the 1-D space. Notice that it would have been slightly easier to use linspace here instead.

s_space_1D = np.arange(s_min,s_max,s_delt)
z_space_1D = np.arange(z_min,z_max,z_delt)

Now we define a 2-D grid of s and z values:

s, z = np.meshgrid(s_space_1D,z_space_1D)

And of course we find the magnetic field:

# Find each term in turn
Bs = 1.0/np.sqrt(s**2 + (z - l2)**2) - 1.0/np.sqrt(s**2 + (z + l2)**2)
Bs = Bs + (z + l2)**2 / (s**2 + (z + l2)**2)**1.5 - (z - l2)**2 / (s**2 + (z - l2)**2)**1.5
Bs = (1.0/(l*s))*Bs
Bz = (z + l2) / (s**2 + (z + l2)**2)**1.5 - (z - l2) / (s**2 + (z - l2)**2)**1.5
Bz = (-1.0/l)*Bz

Now we make and apply the mask to cover both poles:

mask = np.logical_or(np.sqrt(s**2 + (z-l2)**2) < R,np.sqrt(s**2 + (z+l2)**2) < R)
Bs =, mask)
Bz =, mask)

Now we can make an arrow plot of the magnetic field:

plt.axis((-1.1, 1.1, -1.1, 1.1))
plt.title('Magnetic Field Surrounding a Thin Magnetic Needle')
plt.xlabel(r'$\bf s$', fontsize=20)
plt.ylabel(r'$\bf z$', fontsize=20)
plt.savefig("Magnetic_field_of_thin_magnet.pdf", format="pdf", transparent=True, 

Go back to the PythonOverview page.