# Copyright (c) 2026 PaddlePaddle Authors. All Rights Reserved. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. """ 进阶数学操作单元测试 / Advanced Math Operations Unit Tests 测试目标 / Test Target: paddle.tensor.math 进阶函数 (python/paddle/tensor/math.py, 覆盖率约78.8%) 覆盖的模块 / Covered Modules: - paddle.cumsum, cumprod: 累积求和/积 - paddle.diff: 差分 - paddle.digamma, lgamma, erf, erfc: 特殊函数 - paddle.frexp, ldexp: 浮点分解 - paddle.hypot: 斜边长度 - paddle.i0, i0e, i1, i1e: 贝塞尔函数 作用 / Purpose: 覆盖特殊数学函数的代码路径，补充进阶数学计算的测试。 """ import unittest import numpy as np import paddle paddle.disable_static() class TestCumsumCumprod(unittest.TestCase): """测试累积求和和累积积 / Test cumsum and cumprod""" def test_cumsum_1d(self): """测试1D累积求和 / Test 1D cumsum""" x = paddle.to_tensor([1.0, 2.0, 3.0, 4.0]) result = paddle.cumsum(x) expected = np.array([1.0, 3.0, 6.0, 10.0]) np.testing.assert_allclose(result.numpy(), expected, rtol=1e-5) def test_cumsum_2d_axis0(self): """测试2D沿axis=0的累积求和 / Test 2D cumsum along axis=0""" x = paddle.to_tensor([[1.0, 2.0], [3.0, 4.0]]) result = paddle.cumsum(x, axis=0) expected = np.array([[1.0, 2.0], [4.0, 6.0]]) np.testing.assert_allclose(result.numpy(), expected, rtol=1e-5) def test_cumsum_2d_axis1(self): """测试2D沿axis=1的累积求和 / Test 2D cumsum along axis=1""" x = paddle.to_tensor([[1.0, 2.0, 3.0]]) result = paddle.cumsum(x, axis=1) expected = np.array([[1.0, 3.0, 6.0]]) np.testing.assert_allclose(result.numpy(), expected, rtol=1e-5) def test_cumprod_1d(self): """测试1D累积积 / Test 1D cumprod""" x = paddle.to_tensor([1.0, 2.0, 3.0, 4.0]) result = paddle.cumprod(x) expected = np.array([1.0, 2.0, 6.0, 24.0]) np.testing.assert_allclose(result.numpy(), expected, rtol=1e-5) def test_cumsum_dtype(self): """测试cumsum类型转换 / Test cumsum type conversion""" x = paddle.to_tensor([1, 2, 3, 4]) result = paddle.cumsum(x, dtype='float32') self.assertEqual(result.dtype, paddle.float32) class TestDiffOps(unittest.TestCase): """测试差分操作 / Test diff operations""" def test_diff_basic(self): """测试基本差分 / Test basic diff""" x = paddle.to_tensor([1.0, 3.0, 6.0, 10.0]) result = paddle.diff(x) expected = np.array([2.0, 3.0, 4.0]) np.testing.assert_allclose(result.numpy(), expected, rtol=1e-5) def test_diff_n2(self): """测试二阶差分 / Test second-order diff""" x = paddle.to_tensor([1.0, 3.0, 6.0, 10.0]) result = paddle.diff(x, n=2) expected = np.array([1.0, 1.0]) np.testing.assert_allclose(result.numpy(), expected, rtol=1e-5) def test_diff_2d(self): """测试2D差分 / Test 2D diff""" x = paddle.to_tensor([[1.0, 2.0, 4.0], [1.0, 3.0, 6.0]]) result = paddle.diff(x, axis=1) expected = np.array([[1.0, 2.0], [2.0, 3.0]]) np.testing.assert_allclose(result.numpy(), expected, rtol=1e-5) class TestSpecialFunctions(unittest.TestCase): """测试特殊数学函数 / Test special mathematical functions""" def test_erf(self): """测试误差函数 / Test error function""" x = paddle.to_tensor([0.0, 1.0, -1.0]) result = paddle.erf(x) # erf(0) = 0 self.assertAlmostEqual(float(result[0].numpy()), 0.0, places=5) # erf is odd: erf(-x) = -erf(x) self.assertAlmostEqual( float(result[2].numpy()), -float(result[1].numpy()), places=5 ) def test_erfc(self): """测试余误差函数 / Test complementary error function""" # erfc = 1 - erf, erfc(0) = 1 x = paddle.to_tensor([0.0]) erf_result = paddle.erf(x) erfc_approx = 1.0 - float(erf_result.numpy()[0]) self.assertAlmostEqual(erfc_approx, 1.0, places=5) def test_erfinv(self): """测试逆误差函数 / Test inverse error function""" x = paddle.to_tensor([0.0, 0.5, -0.5]) result = paddle.erfinv(x) self.assertEqual(result.shape, [3]) self.assertAlmostEqual(float(result[0].numpy()), 0.0, places=5) def test_digamma(self): """测试Digamma函数 / Test digamma function""" x = paddle.to_tensor([1.0, 2.0, 3.0]) result = paddle.digamma(x) self.assertEqual(result.shape, [3]) def test_lgamma(self): """测试Log-Gamma函数 / Test log-gamma function""" x = paddle.to_tensor([1.0, 2.0, 3.0]) result = paddle.lgamma(x) # lgamma(1) = 0, lgamma(2) = 0, lgamma(3) = log(2) self.assertAlmostEqual(float(result[0].numpy()), 0.0, places=5) self.assertAlmostEqual(float(result[1].numpy()), 0.0, places=5) def test_polygamma(self): """测试Polygamma函数 / Test polygamma function""" x = paddle.to_tensor([1.0, 2.0]) # polygamma(x, n): x is tensor, n is int result = paddle.polygamma(x, 1) self.assertEqual(result.shape, [2]) def test_i0(self): """测试第一类零阶修正贝塞尔函数 / Test 0th-order modified Bessel function""" x = paddle.to_tensor([0.0, 1.0]) result = paddle.i0(x) # i0(0) = 1 self.assertAlmostEqual(float(result[0].numpy()), 1.0, places=4) class TestHyperbolicFunctions(unittest.TestCase): """测试双曲函数 / Test hyperbolic functions""" def test_sinh(self): """测试sinh / Test sinh""" x = paddle.to_tensor([0.0, 1.0, -1.0]) result = paddle.sinh(x) expected = np.sinh(np.array([0.0, 1.0, -1.0])) np.testing.assert_allclose( result.numpy(), expected.astype('float32'), rtol=1e-5 ) def test_cosh(self): """测试cosh / Test cosh""" x = paddle.to_tensor([0.0, 1.0]) result = paddle.cosh(x) # cosh(0) = 1 self.assertAlmostEqual(float(result[0].numpy()), 1.0, places=5) def test_tanh(self): """测试tanh / Test tanh""" x = paddle.to_tensor([0.0, 1.0, -1.0]) result = paddle.tanh(x) self.assertAlmostEqual(float(result[0].numpy()), 0.0, places=5) def test_asinh(self): """测试asinh / Test asinh""" x = paddle.to_tensor([0.0, 1.0]) result = paddle.asinh(x) self.assertAlmostEqual(float(result[0].numpy()), 0.0, places=5) def test_acosh(self): """测试acosh / Test acosh""" x = paddle.to_tensor([1.0, 2.0]) result = paddle.acosh(x) self.assertAlmostEqual(float(result[0].numpy()), 0.0, places=5) def test_atanh(self): """测试atanh / Test atanh""" x = paddle.to_tensor([0.0, 0.5]) result = paddle.atanh(x) self.assertAlmostEqual(float(result[0].numpy()), 0.0, places=5) class TestArithmeticOps(unittest.TestCase): """测试算术操作 / Test arithmetic operations""" def test_hypot(self): """测试斜边计算 / Test hypotenuse calculation""" x = paddle.to_tensor([3.0, 5.0]) y = paddle.to_tensor([4.0, 12.0]) result = paddle.hypot(x, y) expected = np.array([5.0, 13.0]) np.testing.assert_allclose(result.numpy(), expected, rtol=1e-5) def test_frexp(self): """测试浮点分解 / Test floating point decomposition""" x = paddle.to_tensor([6.0, -3.0, 1.5]) mantissa, exponent = paddle.frexp(x) self.assertEqual(mantissa.shape, [3]) self.assertEqual(exponent.shape, [3]) # 6.0 = 0.75 * 2^3 self.assertAlmostEqual(float(mantissa[0].numpy()), 0.75, places=5) self.assertEqual(int(exponent[0].numpy()), 3) def test_ldexp(self): """测试浮点合成 / Test floating point composition""" x = paddle.to_tensor([0.75, 0.5]) exponent = paddle.to_tensor([3, 2]) result = paddle.ldexp(x, exponent) # 0.75 * 2^3 = 6.0, 0.5 * 2^2 = 2.0 np.testing.assert_allclose( result.numpy(), np.array([6.0, 2.0]), rtol=1e-5 ) def test_logit(self): """测试logit函数 / Test logit function""" x = paddle.to_tensor([0.1, 0.5, 0.9]) result = paddle.logit(x) # logit(0.5) = 0 self.assertAlmostEqual(float(result[1].numpy()), 0.0, places=5) if __name__ == '__main__': unittest.main()