Element concatenation operations in numpy and pytorch: stack, concatenat, cat

Splicing of ndarray in numpy

numpy.concatenate

Concatenate multiple matrices along an existing dimension

Requirement: The number of dimensions of the two tensors needs to be the same, and the shapes of other dimensions need to be the same except for the spliced ​​dimension.

np.concatenate((,),axis=)'''The first parameter is a tuple of matrices to be concatenatedThe second parameter is the dimension to be spliced'''a = np.arange(3*3).reshape((3,3))b = np.arange(3*4).reshape((3,4))a,b(array([[0, 1, 2],[3, 4, 5],[6, 7, 8]]),array([[ 0, 1, 2, 3],[ 4, 5, 6, 7],[ 8, 9, 10, 11]]))np.concatenate([a,b],axis=1)array([[ 0, 1, 2, 0, 1, 2, 3],[ 3, 4, 5, 4, 5, 6, 7],[ 6, 7, 8, 8, 9, 10, 11]])

numpy.stack

Stack matrices of the same dimension and shape on a new dimension

Requirement: The matrices have exactly the same dimensions

np.stack((,), axis=)'''The parameters are the same as the function above'''x1 = np.arange(9).reshape((3,3))x2 = np.arange(10,19,1).reshape((3,3))y2 = np.stack((x1,x2),axis=0)output:[[[ 0 1 2][ 3 4 5 ][ 6 7 8 ]][[10 11 12][13 14 15][16 17 18]]]'y2.shape': (2,3,3)

PS:

np.hstack(tup) = np.concatenate(tup, axis=1)

np.vstack(tup) = np.concatenate(tup, axis=0)

Splicing of ndarrays in pytorch

There are also two functions in pytorch that are the same as the two operations in numpy, namely:

torch.cat((,), dim=) -> numpy.concatenate((,), axis=)

torch.stack((,), dim=) -> numpy.stack((,), axis=)

Difference:

torch concatenates tensor tensors, the second parameter of the function is dim=

numpy concatenates ndarray, the second parameter of the function is axis=