CODEWITHSUNDEEP
pythonnumpyscipy
FooBar
FooBar

Reputation: 16488

Re-Meshing ("use output of meshgrid in order to create new meshgrid")

say I have three grids,

a = arange(0,5)
b = arange(0,3)
c = arange(10,12)

And for some reason, in my code, I first need to mesh

A, B = meshgrid(a,b,indexing='ij')

Is there a short way in which I could do

A, B, C = remeshgrid(A, B, c, indexing='ij)

such that A, B, C all correspond to meshgrid(a,b,c, indexing='ij')?

The scenario is the following. - I first have a, b and mesh A, B - Later on, I generate c - At this point, I need to remesh everything containing c. But the code does not have a,b at disposal anymore.

So now it's a tradeoff of adjusting code in order to pass over a, b - if remeshing A,B is not possible or too inefficient.

Do you guys have any thoughts on this?

Upvotes: 0

Views: 130

Answers (2)

unutbu
unutbu

Reputation: 879501

Building on Davidmh's idea, you could use A[:,0] and B[0,:]. This will work even if A or B contains duplicate values; and taking a slice is faster than calling np.unique.

In [71]: A[:,0]
Out[71]: array([0, 1, 2, 3, 4])

In [72]: B[0,:]
Out[72]: array([0, 1, 2])

In [73]: A, B, C = np.meshgrid(A[:,0], B[0,:], c, indexing='ij')

Upvotes: 1

Davidmh
Davidmh

Reputation: 3865

You can reconstruct a and b taking the unique elements:

print A
array([[0, 0, 0],
       [1, 1, 1],
       [2, 2, 2],
       [3, 3, 3],
       [4, 4, 4]])

print B
array([[0, 1, 2],
       [0, 1, 2],
       [0, 1, 2],
       [0, 1, 2],
       [0, 1, 2]])

np.unique(A)
array([0, 1, 2, 3, 4])

np.unique(B)
array([0, 1, 2])

Upvotes: 0

Related Questions