I have a program which crunches data and displays the results in a multi-axis figure. I have many different sets of figures, which I'm trying to generate into a report format. To save memory, I'm making a single figure instance and clearing at the end of each loop. Below is an example of the form:
import matplotlib.pyplot as plt
import numpy as np
def the_figure():
#I want a figure that is persistent and accessible
#So I make the figure an attribute of a function
the_figure.fig = plt.figure()
the_figure.axes = dict(
t_ax = plt.subplot2grid((6,2),(1,0)),
t_fit_ax = plt.subplot2grid((6,2),(1,1)),
o_ax = plt.subplot2grid((6,2),(2,0)),
o_fit_ax = plt.subplot2grid((6,2),(2,1)),
table = plt.subplot2grid((6,2),(3,0),
rowspan = 3, colspan = 2)
)
#A function which makes figures using the single figure function
def Disp(i=5):
try:
the_figure.fig
except:
the_figure()
pi = 3.141592653589793
axes = the_figure.axes
xs = np.linspace(-pi/2,pi/2)
for n in range(i):
for name,ax in axes.items():
ax.plot(xs,np.sin(xs*n))
the_figure.fig.savefig('test_folder\\bad'+str(n),transparent=True)
the_figure.fig.savefig('test_folder\\good'+str(n),transparent=False)
#Clear the axes for reuse, supposedly
for name,ax in axes.items():
ax.cla()
When it's finished, the figures save with a transparent=True get an overlay of the curves from their loop AND curves from the previous loop. I have no idea what's going on.
import matplotlib.pyplot as plt
import numpy as np
fig = plt.figure(1) # This is as persistent as assigning to whatever function
def init_axes(fig):
fig.clear()
return dict(
t_ax = plt.subplot2grid((6,2),(1,0)),
t_fit_ax = plt.subplot2grid((6,2),(1,1)),
o_ax = plt.subplot2grid((6,2),(2,0)),
o_fit_ax = plt.subplot2grid((6,2),(2,1)),
table = plt.subplot2grid((6,2),(3,0),
rowspan = 3, colspan = 2)
)
#A function which makes figures using the single figure
def Disp(i=5):
pi = 3.141592653589793
xs = np.linspace(-pi/2,pi/2)
for n in range(i):
axes = init_axes(fig)
for name,ax in axes.items():
ax.plot(xs,np.sin(xs*n))
fig.savefig('bad'+str(n),transparent=True)
fig.savefig('good'+str(n),transparent=False)