在循环中创建 matplotlib.plots 时 Memory 泄漏

Question

I know it was discussed not only once. Unfortunately, I do not get further.

first the basic: I am trying to loop through OHLC data and create charts. Unfortunately the memory keeps running full.

Apparently it has something to do with the size of the chart.

plt.ioff()
fig = plt.figure(figsize=(50, 25))

for line in ohlc_data:

   # Make magic with the data
   # ...

   df = x.get_ohlc_data()
   ohlc_area = plt.subplot2grid((6, 3), (0, 0), colspan=3, rowspan=4)
   ohlc_area.plot(df[['Date']], df[['close']], 'k-', markevery=markers_on, marker='D',
                                                       label='close')
   plt.grid(axis='x', color='0.95')
   plt.grid(axis='y', color='0.95')
   plt.savefig("Diagramme/" + str(x.id) + "_" + x.state + ".png")

   # Try to clean up everything
   fig.clf()
   plt.cla()
   plt.clf()
   plt.close('all')
   gc.collect()

Up to this point it works. Unfortunately fig.clf() also resets the size of the diagram. If I try to resize the diagram within the loop, there is a memory leak and I don't understand why.

plt.ioff()
fig = plt.figure(figsize=(50, 25))

for line in ohlc_data:

   # Make magic with the data
   # ...

   #plt.figure(figsize=(50, 25))
   plt.rcParams["figure.figsize"] = (50, 25)
   #fig = plt.figure(figsize=(50, 25))

   df = x.get_ohlc_data()
   ohlc_area = plt.subplot2grid((6, 3), (0, 0), colspan=3, rowspan=4)
   ohlc_area.plot(df[['Date']], df[['close']], 'k-', markevery=markers_on, marker='D',
                                                       label='close')
   plt.grid(axis='x', color='0.95')
   plt.grid(axis='y', color='0.95')
   plt.savefig("Diagramme/" + str(x.id) + "_" + x.state + ".png")

   # Try to clean up everything
   fig.clf()
   plt.cla()
   plt.clf()
   plt.close('all')
   gc.collect()

I tried at the beginning of the loop with:

   #plt.figure(figsize=(50, 25))
   plt.rcParams["figure.figsize"] = (50, 25)
   #fig = plt.figure(figsize=(50, 25))

Unfortunately it does not work, can someone tell me how I can prevent the memory increase or how this problem comes about?

Thanks in advance

Answer 1

Not sure if this helps, but perhaps moves you in the right direction.

You are creating a new plot axis during each for loop iteration with the call to subplot2grid. Move this outside of the loop completely:

plt.ioff()
fig = plt.figure(figsize=(50, 25))
ohlc_area = plt.subplot2grid((6, 3), (0, 0), colspan=3, rowspan=4)

for line in ohlc_data:

   # Make magic with the data
   # ...


   df = x.get_ohlc_data() 
   
   ohlc_area.clear()
   ohlc_area.plot(df[['Date']], df[['close']], 'k-', markevery=markers_on, marker='D',
                                                       label='close')
   ohlc_area.grid(axis='x', color='0.95')
   ohlc_area.grid(axis='y', color='0.95')
   plt.savefig("Diagramme/" + str(x.id) + "_" + x.state + ".png")

   # remove the other lines where you were attempting garbage cleanup... the ohlc_area.clear() should clean the axis during each iteration so you can make and save a new plot ... 

Also, you can remove all the garbage cleanup stuff/

Note, there are a lot of unknowns for someone reading your code, for example df = x.get_ohlc_data() , I have no idea what thats doing, and it could also be creating some kind of memory leak. Best to reduce your problem to something others reading your post can run on their end and observe the memory leak...