1# Copyright 2017 The TensorFlow Authors. All Rights Reserved. 2# 3# Licensed under the Apache License, Version 2.0 (the "License"); 4# you may not use this file except in compliance with the License. 5# You may obtain a copy of the License at 6# 7# http://www.apache.org/licenses/LICENSE-2.0 8# 9# Unless required by applicable law or agreed to in writing, software 10# distributed under the License is distributed on an "AS IS" BASIS, 11# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 12# See the License for the specific language governing permissions and 13# limitations under the License. 14# ============================================================================== 15"""Test cases for segment reduction ops.""" 16 17from __future__ import absolute_import 18from __future__ import division 19from __future__ import print_function 20 21import functools 22import numpy as np 23 24from tensorflow.compiler.tests.xla_test import XLATestCase 25from tensorflow.python.ops import array_ops 26from tensorflow.python.ops import math_ops 27from tensorflow.python.platform import googletest 28 29 30class SegmentReductionOpsTest(XLATestCase): 31 """Test cases for segment reduction ops.""" 32 33 def UnsortedSegmentSum(self, data, indices, num_segments): 34 with self.test_session() as sess, self.test_scope(): 35 d = array_ops.placeholder(data.dtype, shape=data.shape) 36 if isinstance(indices, int): 37 i = array_ops.placeholder(np.int32, shape=[]) 38 else: 39 i = array_ops.placeholder(indices.dtype, shape=indices.shape) 40 return sess.run( 41 math_ops.unsorted_segment_sum(d, i, num_segments), 42 {d: data, 43 i: indices}) 44 45 def testUnsortedSegmentSum0DIndices1DData(self): 46 for dtype in self.numeric_types: 47 self.assertAllClose( 48 np.array( 49 [[0, 0, 0, 0, 0, 0], [0, 0, 0, 0, 0, 0], [0, 1, 2, 3, 4, 5], 50 [0, 0, 0, 0, 0, 0]], 51 dtype=dtype), 52 self.UnsortedSegmentSum( 53 np.array([0, 1, 2, 3, 4, 5], dtype=dtype), 2, 4)) 54 55 def testUnsortedSegmentSum1DIndices1DData(self): 56 for dtype in self.numeric_types: 57 self.assertAllClose( 58 np.array([1, 3, 2, 9], dtype=dtype), 59 self.UnsortedSegmentSum( 60 np.array([0, 1, 2, 3, 4, 5], dtype=dtype), 61 np.array([3, 0, 2, 1, 3, 3], dtype=np.int32), 4)) 62 63 def testUnsortedSegmentSum1DIndices1DDataNegativeIndices(self): 64 for dtype in self.numeric_types: 65 self.assertAllClose( 66 np.array([6, 3, 0, 6], dtype=dtype), 67 self.UnsortedSegmentSum( 68 np.array([0, 1, 2, 3, 4, 5, 6], dtype=dtype), 69 np.array([3, -1, 0, 1, 0, -1, 3], dtype=np.int32), 4)) 70 71 def testUnsortedSegmentSum1DIndices2DDataDisjoint(self): 72 for dtype in self.numeric_types: 73 data = np.array( 74 [[0, 1, 2, 3], [20, 21, 22, 23], [30, 31, 32, 33], [40, 41, 42, 43], 75 [50, 51, 52, 53]], 76 dtype=dtype) 77 indices = np.array([8, 1, 0, 3, 7], dtype=np.int32) 78 num_segments = 10 79 y = self.UnsortedSegmentSum(data, indices, num_segments) 80 self.assertAllClose( 81 np.array( 82 [[30, 31, 32, 33], [20, 21, 22, 23], [0, 0, 0, 0], 83 [40, 41, 42, 43], [0, 0, 0, 0], [0, 0, 0, 0], [0, 0, 0, 0], 84 [50, 51, 52, 53], [0, 1, 2, 3], [0, 0, 0, 0]], 85 dtype=dtype), y) 86 87 def testUnsortedSegmentSum1DIndices2DDataNonDisjoint(self): 88 for dtype in self.numeric_types: 89 data = np.array( 90 [[0, 1, 2, 3], [20, 21, 22, 23], [30, 31, 32, 33], [40, 41, 42, 43], 91 [50, 51, 52, 53]], 92 dtype=dtype) 93 indices = np.array([0, 1, 2, 0, 1], dtype=np.int32) 94 num_segments = 4 95 y = self.UnsortedSegmentSum(data, indices, num_segments) 96 self.assertAllClose( 97 np.array( 98 [[40, 42, 44, 46], [70, 72, 74, 76], [30, 31, 32, 33], 99 [0, 0, 0, 0]], 100 dtype=dtype), y) 101 102 def testUnsortedSegmentSum2DIndices3DData(self): 103 for dtype in self.numeric_types: 104 data = np.array( 105 [[[0, 1, 2], [10, 11, 12]], [[100, 101, 102], [110, 111, 112]], 106 [[200, 201, 202], [210, 211, 212]], [[300, 301, 302], 107 [310, 311, 312]]], 108 dtype=dtype) 109 indices = np.array([[3, 5], [3, 1], [5, 0], [6, 2]], dtype=np.int32) 110 num_segments = 8 111 y = self.UnsortedSegmentSum(data, indices, num_segments) 112 self.assertAllClose( 113 np.array( 114 [[210, 211, 212], [110, 111, 112], [310, 311, 312], 115 [100, 102, 104], [0, 0, 0.], [210, 212, 214], [300, 301, 116 302], [0, 0, 0]], 117 dtype=dtype), y) 118 119 def testUnsortedSegmentSum1DIndices3DData(self): 120 for dtype in self.numeric_types: 121 data = np.array( 122 [[[0, 1, 2], [10, 11, 12]], [[100, 101, 102], [110, 111, 112]], 123 [[200, 201, 202], [210, 211, 212]], [[300, 301, 302], 124 [310, 311, 312]]], 125 dtype=dtype) 126 indices = np.array([3, 0, 2, 5], dtype=np.int32) 127 num_segments = 6 128 y = self.UnsortedSegmentSum(data, indices, num_segments) 129 self.assertAllClose( 130 np.array( 131 [[[100, 101, 102.], [110, 111, 112]], [[0, 0, 0], [0, 0, 0]], 132 [[200, 201, 202], [210, 211, 212]], [[0, 1, 2.], [10, 11, 12]], 133 [[0, 0, 0], [0, 0, 0]], [[300, 301, 302], [310, 311, 312]]], 134 dtype=dtype), y) 135 136 def testUnsortedSegmentSumShapeError(self): 137 for dtype in self.numeric_types: 138 data = np.ones((4, 8, 7), dtype=dtype) 139 indices = np.ones((3, 2), dtype=np.int32) 140 num_segments = 4 141 self.assertRaises(ValueError, 142 functools.partial(self.UnsortedSegmentSum, data, 143 indices, num_segments)) 144 145 146if __name__ == '__main__': 147 googletest.main() 148