Find UNIque float array rows.
numpy.unique is an awesome function that alleviates headaches, fast.
Haven't you wished that it'd be applicable for 2D float arrays?
funi is here to help!
There are two available methods: axis and lexicographic.
axis will first project each array to an axis to sort, where as
lexicographic sorts given array in lexicographical manner.
pip install funiimport funi
import numpy as np
# create a random array with duplicating entries
arr = np.random.random((10000, 3))
arr = np.vstack((arr, arr, arr))
np.random.shuffle(arr)
# specify tolerance and if you want your unique ids to be stable sorted.
unique_data, unique_ids, inverse = funi.unique_rows(
arr,
tolerance=1e-11,
sorted_index=True,
method="axis",
)
# use ids to extract unique_data from the original array
assert np.allclose(unique_data, arr[unique_ids])
# use inverse to map unique_data back to the original array
assert np.allclose(arr, unique_data[inverse])
# sorted_index=True gives you sorted unique_ids and corresponding inverse
assert np.alltrue(np.sort(unique_ids) == unique_ids)