def test_nduplicates(self):
# some tuples
tuples = torch.tensor([[[5, 5], [1, 1], [2, 3], [1, 1]],
[[3, 2], [3, 2], [5, 5], [5, 5]]])
# what they should look like after masking out the duplicates
dedup = torch.tensor([[[5, 5], [1, 1], [2, 3], [0, 0]],
[[3, 2], [0, 0], [5, 5], [0, 0]]])
# add a load of dimensions
tuples = tuples[None, None, None, :, :, :].expand(3, 5, 7, 2, 4,
2).contiguous()
dedup = dedup[None, None, None, :, :, :].expand(3, 5, 7, 2, 4,
2).contiguous()
# find the duplicates
dup = util.nduplicates(tuples)
# mask them out
tuples[dup, :] = tuples[dup, :] * 0
self.assertEqual((tuples != dedup).sum(),
0) # assert equal to expected
# second test: explicitly test the bitmask returned by nduplicates
tuples = torch.tensor([[[3, 1], [3, 2], [3, 1], [0, 3], [0, 2], [3, 0],
[0, 3], [0, 0]]])
tuples = tuples[None, None, None, :, :, :].expand(8, 1, 7, 1, 8,
2).contiguous()
self.assertEqual([0, 0, 1, 0, 0, 0, 1, 0],
list(
util.nduplicates(tuples)[0, 0, 0, :, :].view(-1)))
# third test: single element tuples
tuples = torch.tensor([[[5], [1], [2], [1]], [[3], [3], [5], [5]]])
dedup = torch.tensor([[[5], [1], [2], [0]], [[3], [0], [5], [0]]])
tuples = tuples[None, None, None, :, :, :].expand(3, 5, 7, 2, 4,
2).contiguous()
dedup = dedup[None, None, None, :, :, :].expand(3, 5, 7, 2, 4,
2).contiguous()
dup = util.nduplicates(tuples)
tuples[dup, :] = tuples[dup, :] * 0
self.assertEqual((tuples != dedup).sum(), 0)
def test_nduplicates_recursion(self):
"""
Reproducing observed recursion error
:return:
"""
# tensor of 6 1-tuples
tuples = torch.tensor([[[[74], [75], [175], [246], [72], [72]]]])
dedup = torch.tensor([[[[74], [75], [175], [246], [72], [0]]]])
dup = util.nduplicates(tuples)
tuples[dup, :] = tuples[dup, :] * 0
self.assertEqual((tuples != dedup).sum(), 0)