torch.cdist#

torch.cdist(x1,x2,p=2.0,compute_mode='use_mm_for_euclid_dist_if_necessary')[source]#

Computes batched the p-norm distance between each pair of the two collections of row vectors.

Parameters:

• x1 (Tensor) – input tensor where the last two dimensions represent the points and the feature dimension respectively.The shape can be$D_1\times D_2\times \cdots \times D_n\times P\times M$D1​×D2​×⋯×Dn​×P×M,where$P$P is the number of points and$M$M is the feature dimension.

• x2 (Tensor) – input tensor where the last two dimensions also represent the points and the feature dimension respectively.The shape can be$D_1^{\mathrm{\prime }}\times D_2^{\mathrm{\prime }}\times \cdots \times D_m^{\mathrm{\prime }}\times R\times M$D1′​×D2′​×⋯×Dm′​×R×M,where$R$R is the number of points and$M$M is the feature dimension,which should match the feature dimension ofx1.

• p (float) – p value for the p-norm distance to calculate between each vector pair$\in [0,\mathrm{\infty }]$∈[0,∞].

• compute_mode (str) – ‘use_mm_for_euclid_dist_if_necessary’ - will use matrix multiplication approach to calculateeuclidean distance (p = 2) if P > 25 or R > 25‘use_mm_for_euclid_dist’ - will always use matrix multiplication approach to calculateeuclidean distance (p = 2)‘donot_use_mm_for_euclid_dist’ - will never use matrix multiplication approach to calculateeuclidean distance (p = 2)Default: use_mm_for_euclid_dist_if_necessary.

Return type:

Tensor

If x1 has shape$B\times P\times M$B×P×M and x2 has shape$B\times R\times M$B×R×M then theoutput will have shape$B\times P\times R$B×P×R.

This function is equivalent toscipy.spatial.distance.cdist(input,’minkowski’, p=p)if$p\in (0,\mathrm{\infty })$p∈(0,∞). When$p=0$p=0 it is equivalent toscipy.spatial.distance.cdist(input, ‘hamming’) * M. When$p=\mathrm{\infty }$p=∞, the closestscipy function isscipy.spatial.distance.cdist(xn, lambda x, y: np.abs(x - y).max()).

Example

>>>a=torch.tensor([[0.9041,0.0196],[-0.3108,-2.4423],[-0.4821,1.059]])>>>atensor([[ 0.9041,  0.0196],        [-0.3108, -2.4423],        [-0.4821,  1.0590]])>>>b=torch.tensor([[-2.1763,-0.4713],[-0.6986,1.3702]])>>>btensor([[-2.1763, -0.4713],        [-0.6986,  1.3702]])>>>torch.cdist(a,b,p=2)tensor([[3.1193, 2.0959],        [2.7138, 3.8322],        [2.2830, 0.3791]])