+- Python Forum (https://python-forum.io)
+-- Forum: Python Coding (https://python-forum.io/forum-7.html)
+--- Forum: General Coding Help (https://python-forum.io/forum-8.html)
+--- Thread: How to avoid slow cursor operations? (/thread-21726.html)
How to avoid slow cursor operations? - sevensixtwo - Oct-11-2019
Hi. I'm using numpy as np. I have a 2D array called CC of all zeros and ones. For every one in CC, I want to change all adjacent array values to one. I want to do this recursively, and then I want to count how many steps it takes for the last zero to disappear. That is the point of the code: how far away from a one is the farthest zero. I initialized CC with
CC = np.zeros((N,N),dtype="int")and then I wrote some ones into it. My first idea was to make 8 copies of CC, each with the values shifted by one, (eight copies for up/down,left/right, and 4 diag) and then add those copies to CC, and then iterate. However, all the shift commands I found like roll do not shift the entire array but they send the end of each line to the beginning of the next. As it is, if CC is 1000 x 1000, then I have the below cursor doing ~eight operations for one million times. There must be a better way!
MAX_RADIUS = 0
nn = 0
CCtmp = np.zeros((N,N),dtype="int")
while nn < 1:
for i in range(len(CC)-1):
for j in range(len(CC)-1):
if i**2+ j**2 > N**2: ## Condition for the part of CC I don't care about
CCtmp[i,j] = 1
if (CC[i,j] == 1 ):
CCtmp[i+1,j] = 1 ## For every black pixel, color adjacent black
CCtmp[i,j+1] = 1
CCtmp[i+1,j+1] = 1
CCtmp[i-1,j] = 1
CCtmp[i,j-1] = 1
CCtmp[i-1,j-1] = 1
CCtmp[i-1,j+1] = 1
CCtmp[i+1,j-1] = 1
CC = CC + CCtmp
nn = np.amin(CC)
MAX_RADIUS += 1
print('Max radius within ' + str(N) + ' is less than ' + str(MAX_RADIUS) + '.\n\n') Keeping mind the point of the code: how far away from a one is the farthest zero, any other suggestions are welcome.