4.1. Plotting outputs in 2d: plots and histograms

GNA defines a set of convenience methods to plot the transformation outputs with matplotlib. The methods are wrappers for regular matplotlib commands. For the complete matplotlib documentation please refer the official site.

4.1.1. Plotting arrays

A plot(...) method is defined implementing plot(y, …) call passing output contents as y. The method works the same way for both arrays and histograms.

01_points_plot.py

import gna.constructors as C
import numpy as np
from gna.bindings import common

# Create numpy arrays for 1d and 2d cases
narray1 = np.arange(5, 10)
narray2 = np.arange(5, 20).reshape(5,3)
# Create Points instances
parray1 = C.Points(narray1)
parray2 = C.Points(narray2)

print('Data 1d:',   parray1.points.points.data())
print('Data 2d:\n', parray2.points.points.data())

from matplotlib import pyplot as plt
fig = plt.figure()
ax = plt.subplot( 111 )
ax.minorticks_on()
ax.grid()
ax.set_xlabel( 'x label' )
ax.set_ylabel( 'y label' )
ax.set_title( 'Plot title' )

parray1.points.points.plot('-o', label='plot 1d')
parray2.points.points.plot('-s', label='columns of 2d')

ax.legend(loc='upper left')

plt.show()

When 1d array is passed (line 24) it is plotted as is while for 2d array (line 25) each column is plotted in separate. The latter produces the blue line on the following figure while the former produces orange, green and red lines.

An example output.plot() method for outputs with 1d and 2d arrays.

Keyword options

transpose=True

transpose array before plotting

4.1.2. Plotting arrays vs other arrays

If X vs Y is desired output_y.plot_vs(output_x, ...) syntax may be used. Matplotlib plot(x, y, …) function is used.

The twin method output_x.vs_plot(output_y, ...) may be used in case reversed order is desired.

02_points_plot_vs.py

import gna.constructors as C
import numpy as np
from gna.bindings import common

# Create numpy arrays for 1d and 2d cases
narray_x = np.arange(5, 10)
narray_y = np.arange(0.13, 0.08, -0.01)
# Create Points instances
parray_x = C.Points(narray_x)
parray_y = C.Points(narray_y)

from matplotlib import pyplot as plt
fig = plt.figure()
ax = plt.subplot( 111 )
ax.minorticks_on()
ax.grid()
ax.set_xlabel( 'x label' )
ax.set_ylabel( 'y label' )
ax.set_title( 'Plot title' )

parray_y.points.points.plot_vs(parray_x.points.points, '-o', label='plot Y vs X')

ax.legend(loc='upper right')

plt.show()

An example output_y.plot_vs(output_x) method for outputs.

Keyword options

transpose=True

transpose arrays before plotting

Note

The argument of plot_vs() and vs_plot() methods may be numpy array as well.

4.1.3. Plotting histograms

There are two options to plot 1d histograms provided. First one plot_hist() is producing regular line plot via pyplot.plot(), the second one plot_bar() is passing data to pyplot.bar(). See the example below.

The plot_hist() has an extra option zero_value, which is by default 0. The options set the ground value for the histogram that affects how the edges of first and last bins are plotted: they are drawn till the zero_value.

04_hist_plot.py

import gna.constructors as C
import numpy as np
from gna.bindings import common
from matplotlib import pyplot as plt
# Create numpy array for data points
nbins = 100
narray = np.arange(nbins)**2 * np.arange(nbins)[::-1]**2
# Create numpy array for bin edges
edges  = np.linspace(1.0, 40.0, nbins+1)

# Create a histogram instance with data, stored in `narray`
# and edges, stored in `edges`
hist = C.Histogram(edges, narray)

fig = plt.figure(figsize=(12, 5))
ax = plt.subplot( 121 )
ax.set_title( 'Plot title (left)' )
ax.minorticks_on()
ax.grid()
ax.set_xlabel( 'x label' )
ax.set_ylabel( 'y label' )
plt.ticklabel_format(style='sci', axis='y', scilimits=(-2,2), useMathText=True)

hist.hist.hist.plot_hist(label='histogram 1')

ax.legend()

ax = plt.subplot( 122 )
ax.set_title( 'Plot title (right)' )
ax.minorticks_on()
ax.grid()
ax.set_xlabel( 'x label' )
ax.set_ylabel( 'y label' )
plt.ticklabel_format(style='sci', axis='y', scilimits=(-2,2), useMathText=True)

hist.hist.hist.plot_bar(label='histogram 1 (bar)', alpha=0.6)

ax.legend()

plt.show()

An example output.plot_hist() and output.plot_bar() methods for outputs.

Keyword options for plot_hist()

zero_value=0.0

set the ground value of the histogram

Histograms may be plotted with errorbar() method. See the file 04_hist_plot_errorbar.py for example.

An example output.plot_errorbar() method for outputs.

4.1.4. Overlapping histograms

Both plotting methods may be used for plotting multiple histograms.

4.1.4.1. Line plots

Multiple plot_hist() are plotted as regular plots.

05_hists_plot.py

import gna.constructors as C
import numpy as np
from gna.bindings import common
from matplotlib import pyplot as plt
# Create numpy array for data points
nbins = 40
# Create numpy array for bin edges
edges  = np.linspace(1.0, 10.0, nbins+1)
narray1 = np.exp(edges[:-1])
narray2 = np.flip(narray1)
narray3 = narray1[-1]*np.exp(-0.5*(5.0-edges[:-1])**2/1.0)/(2*np.pi)**0.5

# Create a histogram instance with data, stored in `narray`
# and edges, stored in `edges`
hist1 = C.Histogram(edges, narray1)
hist2 = C.Histogram(edges, narray2)
hist3 = C.Histogram(edges, narray3)

fig = plt.figure()
ax = plt.subplot( 111 )
ax.set_title( 'Plot title (left)' )
ax.minorticks_on()
ax.grid()
ax.set_xlabel( 'x label' )
ax.set_ylabel( 'y label' )
plt.ticklabel_format(style='sci', axis='y', scilimits=(-2,2), useMathText=True)

hist1.hist.hist.plot_hist(label='exp(+)')
hist2.hist.hist.plot_hist(label='exp(-)')
hist3.hist.hist.plot_hist(label='gauss')

ax.legend()

plt.show()

Several histograms superimposed in plot_hist() version.

For the bar version there are two ways to plot overlapping histograms.

4.1.4.2. Bars’ transparency

First one is to modify the histograms’ transparency by setting alpha option below 1.

05_hists_plot.py

fig = plt.figure()
ax = plt.subplot( 111 )
ax.set_title( 'Plot title (left)' )
ax.minorticks_on()
ax.grid()
ax.set_xlabel( 'x label' )
ax.set_ylabel( 'y label' )
plt.ticklabel_format(style='sci', axis='y', scilimits=(-2,2), useMathText=True)

hist1.hist.hist.plot_bar(label='exp(+)', alpha=0.4)
hist2.hist.hist.plot_bar(label='exp(-)', alpha=0.4)
hist3.hist.hist.plot_bar(label='gauss' , alpha=0.4)

ax.legend()

plt.show()

Several histograms superimposed in plot_bar() version with transparency.

4.1.4.3. Bars’ width

The second option is controlled by divide and shift options. divide is an integer factor dividing the bin width. Setting divide=3 will shrink the bin width three times. The option shift defines where to plot the shrunk bin within it’s old width: shift=0 shifts it to the left side, shift=1 to the center and shift=2 to the right side. It is possible to plot overlapping histograms without bins actually overlapping.

05_hists_plot.py

fig = plt.figure()
ax = plt.subplot( 111 )
ax.set_title( 'Plot title (left)' )
ax.minorticks_on()
ax.grid()
ax.set_xlabel( 'x label' )
ax.set_ylabel( 'y label' )
plt.ticklabel_format(style='sci', axis='y', scilimits=(-2,2), useMathText=True)

hist1.hist.hist.plot_bar(label='exp(+)', divide=3, shift=0)
hist2.hist.hist.plot_bar(label='exp(-)', divide=3, shift=1)
hist3.hist.hist.plot_bar(label='gauss' , divide=3, shift=2)

ax.legend()

plt.show()

Several histograms superimposed in plot_bar() version with shrunk bins.

Keyword options for plot_bar()

divide=N	Divide each bin width by N
shift=M	Shift each bin by N widths (M<N)