Python: Use numpy flip with memmap

Question

I writting large array so I'm using numpy memmap

At the end of my operation I would like to use flip function but the problem is, the flip function does not apply after using the flush function on memap

import numpy as np
from matplotlib import pyplot as plt

#Create array on disk

fpc = np.memmap("test3.rgb", dtype=np.uint8, mode='w+', shape=(10000,20000,3))
fpc[0:5000] = [255,0,0]     #Top red
fpc[5000:10000] = [0,255,0] #Bottom green


plt.subplot(131)
plt.title("Before flip")
plt.imshow(fpc)


#Apply flip function, so green come on top
fpc = np.flip(fpc,0)

plt.subplot(132)
plt.imshow(fpc)
plt.title("After flip")

#Save changes
fpc.flush()

#Reload array
fpc = np.memmap("test3.rgb", dtype=np.uint8, mode='r', shape=(10000,20000,3))


plt.subplot(133)
plt.title("After reload")
plt.imshow(fpc)
plt.show()

Result:

Is it really possible to use memmap with a flip function?

Jérôme Richard · Accepted Answer · 2022-04-12 11:06:45Z

3

np.flip creates a new array in memory. You need to set the one mapped to the file. The current code only change the reference of fpc so to reference the new array. You can fix that with:

fpc[:,:] = np.flip(fpc,0)

UPDATE: note that if you use memmap to reduce the memory footprint, then doing a np.flip will defeat this optimization since a new array is created anyway. You can swap pairs of lines manually so to reduce the memory footprint. For better performance, you can flip&swap chunks of lines.

edited Apr 12, 2022 at 11:06

answered Apr 12, 2022 at 10:41

Jérôme Richard

53.3k6 gold badges48 silver badges77 bronze badges

Sign up to request clarification or add additional context in comments.

Collectives™ on Stack Overflow

Python: Use numpy flip with memmap

1 Answer 1

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

Comments

Your Answer

Sign up or log in

Post as a guest

Related