diag_embed¶
- paddle.nn.functional. diag_embed ( input, offset=0, dim1=- 2, dim2=- 1 ) [source]
-
This OP creates a tensor whose diagonals of certain 2D planes (specified by dim1 and dim2) are filled by
input. By default, a 2D plane formed by the last two dimensions of the returned tensor will be selected.The argument
offsetdetermines which diagonal is generated:If offset = 0, it is the main diagonal.
If offset > 0, it is above the main diagonal.
If offset < 0, it is below the main diagonal.
- Parameters
-
input (Tensor|numpy.ndarray) – The input tensor. Must be at least 1-dimensional. The input data type should be float32, float64, int32, int64.
offset (int, optional) – Which diagonal to consider. Default: 0 (main diagonal).
dim1 (int, optional) – The first dimension with respect to which to take diagonal. Default: -2.
dim2 (int, optional) – The second dimension with respect to which to take diagonal. Default: -1.
- Returns
-
Tensor, the output data type is the same as input data type.
Examples
import paddle.nn.functional as F import numpy as np diag_embed = np.random.randn(2, 3).astype('float32') # [[ 0.7545889 , -0.25074545, 0.5929117 ], # [-0.6097662 , -0.01753256, 0.619769 ]] data1 = F.diag_embed(diag_embed) data1.numpy() # [[[ 0.7545889 , 0. , 0. ], # [ 0. , -0.25074545, 0. ], # [ 0. , 0. , 0.5929117 ]], # [[-0.6097662 , 0. , 0. ], # [ 0. , -0.01753256, 0. ], # [ 0. , 0. , 0.619769 ]]] data2 = F.diag_embed(diag_embed, offset=-1, dim1=0, dim2=2) data2.numpy() # [[[ 0. , 0. , 0. , 0. ], # [ 0.7545889 , 0. , 0. , 0. ], # [ 0. , -0.25074545, 0. , 0. ], # [ 0. , 0. , 0.5929117 , 0. ]], # # [[ 0. , 0. , 0. , 0. ], # [-0.6097662 , 0. , 0. , 0. ], # [ 0. , -0.01753256, 0. , 0. ], # [ 0. , 0. , 0.619769 , 0. ]]] data3 = F.diag_embed(diag_embed, offset=1, dim1=0, dim2=2) data3.numpy() # [[[ 0. , 0.7545889 , 0. , 0. ], # [ 0. , -0.6097662 , 0. , 0. ]], # # [[ 0. , 0. , -0.25074545, 0. ], # [ 0. , 0. , -0.01753256, 0. ]], # # [[ 0. , 0. , 0. , 0.5929117 ], # [ 0. , 0. , 0. , 0.619769 ]], # # [[ 0. , 0. , 0. , 0. ], # [ 0. , 0. , 0. , 0. ]]]