FuncAnimation run twice of frames in the tk backends in the Jupyter notebook

Question

I have some plot-related questions regarding FuncAnimation used in Jupyter notebook.

• The main issue is the FuncAnimation runs twice of frames number while it saves the right number of frames.
• The minor issue is in the %matplotlib tk magic, the GUI window does not show the immediate plot.
• The even minor issue is that tqdm progress bar only shows in the first line and frozens as 0%.

Here is a sample code (I guess there are many unnecessary global variables in the following codes, as I am not clear about FuncAnimation on passing variables. Any suggestions will be greatly welcome):

from tqdm.notebook import tqdm
import matplotlib.animation as animation
from IPython import display
import sys

def ini_plot(fig, axs, iter_num, y, A):

    line, = axs[0,0].plot(np.arange(iter_num), y)
    im0 = axs[0,1].imshow(A, origin='lower', extent=(0,1,0,1))
    cbar = fig.colorbar(im0, ax=axs[0,:].ravel().tolist(), shrink=0.8)

    return line, cbar

def con_plot(fig, axs, frame, y, A, z):
    
    # each time update a point on the line
    y[frame] = z[-1]
    line.set_ydata(np.log10(y))
    im0 = axs[0,1].imshow(A, origin='lower', extent=(0,1,0,1))
    cbar.update_normal(im0)
    
    fig.canvas.draw()

    return line, cbar

def update(frame):
    global y, A, z, fig, axs

    A = A_update()
    z = z_update()

    line, cbar = con_plot(fig, axs, frame, y, A, z)
    return line, cbar

## Main function
plt.close('all')

%matplotlib tk

iter_num = 100
y = y_ini()
A = A_ini()

fig, axs = plt.subplots(1,2, figsize=(16,8))
plt.ioff()
line, cbar = ini_plot(fig, axs, iter_num, y, A)
global line, cbar

ani = FuncAnimation(fig, update, frames=tqdm(range(iter_num), file=sys.stdout), 
                   interval=10, blit=False, init_func=None, repeat= False)

ani.save('animation.mp4', writer='ffmpeg')
     
 

• In the above code, I expected update to run for 100 frames, the GUI window to show those 100 frames immediately and 100 frames to be saved. The result is: animation.mp4 saves the first 100 frames while the GUI window shows from 101st frame to 200th frame.
• Nno matter if I tried any combination of plt.ion(), plt.ioff(), plt.show(). The GUI window is popped after initialization and stays blank. It shows the plot after first 100 frames are computed, and I didn't expect it to continue the runs after 100 frames...