Obtaining indexes of changed data between two numpy arrays

Question

I have two 3d numpy arrays a1 and a2 where len(a1) == len(a2) and a1[x, y, z] = id

I'm using this code to find out if there is changed data in any of the z layers

eq = a1 == a2
if eq.any():
    #find the indexes of the changed data

as the comment and title says I need to find the indexes of the changed data. basically I have a list of objects that correspond the the positions in the array and I need to update those objects based on the id pulled form the array. I want to do this as fast as possible as this list can get REALLY large possibly more than 120,000 entries. but only a hundred or so of these entries are likely to change at any one time. as such I'd like to obtain a list of the changed indexes so I can call the object at that index and update it.

I do need to maintain the three components of the index

is there a way to do this without looping through the list? perhaps with numpy.nonzero()

Paul · Accepted Answer · 2011-02-22 06:17:49Z

4

Two more options:

np.argwhere(a1!=a2)
np.where(a1!=a2)

They both do the same thing but produce results in different formats (one suitable for indexing the arrays, the other more readable)

answered Feb 22, 2011 at 6:17

Paul

43.9k17 gold badges112 silver badges126 bronze badges

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

1 Comment

Ryex Over a year ago

np.argwhere() appears to be what I'm looking for. thank you. it returns an array of arrays with the indexes that match the condition which is exactly what I needed.

doug · Accepted Answer · 2011-02-22 06:25:40Z

1

I didn't understand the specifics of your question (e.g., the size/shape of your array); this solution should be general enough though. In any event, it's loop-less:

# two 2d arrays
x = NP.random.randint(0, 10, 20).reshape(5, 4)
y = NP.random.randint(0, 10, 20).reshape(5, 4)

ndx1, ndx2 = NP.nonzero((x - y))

# all indices in which the values of x & y differ, each index is a tuple
ndx = zip(ndx1, ndx2)  

# you can also use argwhere, returns each index as a 1d array vs. tuple:
NP.argwhere(dxy)

# finally, just generating an array the same size as the diff array with e.g.,
# '1' for each diff element and '0' for each non-diff:
dxy = x - y
NP.where(dxy=0, 0, 1)

edited Feb 22, 2011 at 6:25

answered Feb 22, 2011 at 5:09

doug

70.3k26 gold badges171 silver badges201 bronze badges

Collectives™ on Stack Overflow

Obtaining indexes of changed data between two numpy arrays

2 Answers 2

1 Comment

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

1 Comment

Comments

Your Answer

Sign up or log in

Post as a guest

Related