Inverse mapping of numpy.sort

Question

I want to get an index for the inverse mapping of np.sort

In other words, if I have a code

x=[[4,3],[2,3],[4,2]]
ind1=np.lexsort((x[:,1],x[:,0]))
y=x[ind1] #y=[[2,3],[4,2],[4,3]]

then I want an index such that

y[ind]

returns the original function x. Is there any way to get this index?

Answer 1

Use np.argsort on ind1:

idx = np.argsort(ind1)
y[idx]
# array([[4, 3],
#        [2, 3],
#        [4, 2]])