Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > comp.lang.python > #100200 > unrolled thread

np.searchSorted over 2D array

Back to article view | Back to comp.lang.python

Contents

  np.searchSorted over 2D array Heli <hemla21@gmail.com> - 2015-12-09 08:06 -0800
    Re: np.searchSorted over 2D array Peter Otten <__peter__@web.de> - 2015-12-09 17:36 +0100
      Re: np.searchSorted over 2D array Heli <hemla21@gmail.com> - 2015-12-10 06:40 -0800
        Re: np.searchSorted over 2D array Oscar Benjamin <oscar.j.benjamin@gmail.com> - 2015-12-10 19:55 +0000

#100200 — np.searchSorted over 2D array

Dear all, 

I need to check whether two 2d numpy arrays have intersections and if so I will need to have the cell indices of the intersection.

By intersection, I exactly mean the intersection definition used in set theory.

I will give an example of what I need to do:

a=[[0,1,2],[3,4,5],[6,7,8]]
b=[[0,1,2],[3,4,5]]

I would like to check whether b is subset of a and then get the indices in a where b matches. 

cellindices=[[True,True,True],[True,True,True],[False,False,False]]

What is the best way to do this in numpy? 

 
Thanks in Advance for your help, 

[toc] | [next] | [standalone]

#100201

Heli wrote:

[Please don't open a new thread for the same problem] 

> I need to check whether two 2d numpy arrays have intersections and if so I
> will need to have the cell indices of the intersection.
> 
> By intersection, I exactly mean the intersection definition used in set
> theory.
> 
> I will give an example of what I need to do:
> 
> a=[[0,1,2],[3,4,5],[6,7,8]]
> b=[[0,1,2],[3,4,5]]
> 
> I would like to check whether b is subset of a and then get the indices in
> a where b matches.
> 
> cellindices=[[True,True,True],[True,True,True],[False,False,False]]
> 
> What is the best way to do this in numpy?

Providing an example is an improvement over your previous post, but to me 
it's still not clear what you want.

>>> functools.reduce(lambda x, y: x | y, (a == i for i in b.flatten()))
array([[ True,  True,  True],
       [ True,  True,  True],
       [False, False, False]], dtype=bool)

produces the desired result for the example input, but how do you want to 
handle repeating numbers as in 

>>> a = numpy.array([[0,1,2],[3,4,5],[3, 2, 1]])
>>> functools.reduce(lambda x, y: x | y, (a == i for i in b.flatten()))
array([[ True,  True,  True],
       [ True,  True,  True],
       [ True,  True,  True]], dtype=bool)

?

Try to be clear about your requirement, describe it in english and provide a 
bit of context. You might even write a solution that doesn't use numpy and 
ask for help in translating it.

At the very least we need more/better examples.

[toc] | [prev] | [next] | [standalone]

#100235

Thanks Peter, 

I will try to explain what I really need. 

I have a 3D numpy array of 100*100*100 (1M elements). Then I have another numpy array of for example 10*2*10 (200 elements). I want to know if in the bigger dataset of 100*100*100, there is anywhere, where the second numpy array of 200 elements with shape 10*2*10 appears. If it does, then I want the indices of the bigger dataset where this happens.

I hope I´m making myself more clear :)
Thanks for your comments,


On Wednesday, December 9, 2015 at 5:37:07 PM UTC+1, Peter Otten wrote:
> Heli wrote:
> 
> [Please don't open a new thread for the same problem] 
> 
> > I need to check whether two 2d numpy arrays have intersections and if so I
> > will need to have the cell indices of the intersection.
> > 
> > By intersection, I exactly mean the intersection definition used in set
> > theory.
> > 
> > I will give an example of what I need to do:
> > 
> > a=[[0,1,2],[3,4,5],[6,7,8]]
> > b=[[0,1,2],[3,4,5]]
> > 
> > I would like to check whether b is subset of a and then get the indices in
> > a where b matches.
> > 
> > cellindices=[[True,True,True],[True,True,True],[False,False,False]]
> > 
> > What is the best way to do this in numpy?
> 
> Providing an example is an improvement over your previous post, but to me 
> it's still not clear what you want.
> 
> >>> functools.reduce(lambda x, y: x | y, (a == i for i in b.flatten()))
> array([[ True,  True,  True],
>        [ True,  True,  True],
>        [False, False, False]], dtype=bool)
> 
> produces the desired result for the example input, but how do you want to 
> handle repeating numbers as in 
> 
> >>> a = numpy.array([[0,1,2],[3,4,5],[3, 2, 1]])
> >>> functools.reduce(lambda x, y: x | y, (a == i for i in b.flatten()))
> array([[ True,  True,  True],
>        [ True,  True,  True],
>        [ True,  True,  True]], dtype=bool)
> 
> ?
> 
> Try to be clear about your requirement, describe it in english and provide a 
> bit of context. You might even write a solution that doesn't use numpy and 
> ask for help in translating it.
> 
> At the very least we need more/better examples.

[toc] | [prev] | [next] | [standalone]

#100244

On 10 Dec 2015 14:46, "Heli" <hemla21@gmail.com> wrote:
>
> Thanks Peter,
>
> I will try to explain what I really need.
>
> I have a 3D numpy array of 100*100*100 (1M elements). Then I have another
numpy array of for example 10*2*10 (200 elements). I want to know if in the
bigger dataset of 100*100*100, there is anywhere, where the second numpy
array of 200 elements with shape 10*2*10 appears. If it does, then I want
the indices of the bigger dataset where this happens.
>

So you want to find N in M. First find all occurrences of N[0][0][0] in
M[:90][:98][:90]. Then for each of those extract the same size subview from
M and check if (Ms == N).all().

--
Oscar

[toc] | [prev] | [standalone]

Back to top | Article view | comp.lang.python

csiph-web