Coverage for /home/runner/work/torchcvnn/torchcvnn/src/torchcvnn/nn/modules/conv.py: 100%

1# MIT License 

2 

3# Copyright (c) 2023 Jérémie Levi, Victor Dhédin, Jeremy Fix 

4 

5# Permission is hereby granted, free of charge, to any person obtaining a copy 

6# of this software and associated documentation files (the "Software"), to deal 

7# in the Software without restriction, including without limitation the rights 

8# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell 

9# copies of the Software, and to permit persons to whom the Software is 

10# furnished to do so, subject to the following conditions: 

11 

12# The above copyright notice and this permission notice shall be included in 

13# all copies or substantial portions of the Software. 

14 

15# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR 

16# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, 

17# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE 

18# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER 

19# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, 

20# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE 

21# SOFTWARE. 

22 

23# External imports 

24import torch 

25import torch.nn as nn 

26from torch.nn.common_types import _size_2_t 

27 

28 

29class ConvTranspose2d(nn.Module): 

30 """ 

31 Implementation of torch.nn.Conv2dTranspose for complex numbers. 

32 Apply Conv2dTranspose on real and imaginary part of the complex number. 

33 

34 The parameters are the same than from the upstream pytorch layer: 

35 

36 Arguments: 

37 in_channels (int): Number of channels in the input image 

38 out_channels (int): Number of channels produced by the convolution 

39 kernel_size (int or tuple): Size of the convolving kernel 

40 stride (int or tuple, optional): Stride of the convolution. Default: 1 

41 padding (int or tuple, optional): ``dilation * (kernel_size - 1) - padding`` zero-padding will be added to both sides of each dimension in the input. Default: 0 

42 output_padding (int or tuple, optional): Additional size added to one side of each dimension in the output shape. Default: 0 

43 groups (int, optional): Number of blocked connections from input channels to output channels. Default: 1 

44 bias (bool, optional): If ``True``, adds a learnable bias to the output. Default: ``True`` 

45 dilation (int or tuple, optional): Spacing between kernel elements. Default: 1 

46 padding_mode (str, optional): ``'zeros'``, ``'reflect'``, ``'replicate'`` or ``'circular'``. Default: ``'zeros'`` 

47 """ 

48 

49 def __init__( 

50 self, 

51 in_channels: int, 

52 out_channels: int, 

53 kernel_size: _size_2_t, 

54 stride: _size_2_t = 1, 

55 padding: _size_2_t = 0, 

56 output_padding: _size_2_t = 0, 

57 groups: int = 1, 

58 bias: bool = True, 

59 dilation: _size_2_t = 1, 

60 padding_mode: str = "zeros", 

61 device=None, 

62 dtype=None, 

63 ) -> None: 

64 super().__init__() 

65 

66 self.m_real = torch.nn.ConvTranspose2d( 

67 in_channels, 

68 out_channels, 

69 kernel_size, 

70 stride, 

71 padding, 

72 output_padding, 

73 groups, 

74 bias, 

75 dilation, 

76 padding_mode, 

77 device, 

78 dtype, 

79 ) 

80 

81 self.m_imag = torch.nn.ConvTranspose2d( 

82 in_channels, 

83 out_channels, 

84 kernel_size, 

85 stride, 

86 padding, 

87 output_padding, 

88 groups, 

89 bias, 

90 dilation, 

91 padding_mode, 

92 device, 

93 dtype, 

94 ) 

95 

96 def forward(self, z: torch.Tensor) -> torch.Tensor: 

97 """ 

98 Performs the forward pass, applying real valued ConvTranspose2d 

99 independently on both the real and imaginary parts of the input. 

100 """ 

101 return torch.view_as_complex( 

102 torch.cat( 

103 ( 

104 torch.unsqueeze(self.m_real(z.real) - self.m_imag(z.imag), -1), 

105 torch.unsqueeze(self.m_real(z.imag) + self.m_imag(z.real), -1), 

106 ), 

107 axis=-1, 

108 ) 

109 )