Multiplication in pytorch: mul (), multiply (), matmul (), mm (), MV (), dot ()

torch.mul()

The functionality ： Pair by pair input and other Multiply the corresponding elements in .

This operation supports broadcasting , therefore input and other Both can be tensors or numbers

import torch
a = torch.randn((1,2))
b = torch.randn((2,1))
print(a,b)
torch.mul(a,b)

torch.multiply()

torch.mul() The nickname

torch.matmul()

matmul You can do tensor multiplication , The input can be high-dimensional .

torch.dot()

The functionality ： Calculation input and output Dot product of , This function requires input and output Must be a one-dimensional tensor （ Its shape There is only one value in the property ）！ And the number of elements of both is required to be the same ！

import torch 

x = torch.Tensor([1,2])
y = torch.Tensor([3,4])
z = torch.dot(x,y)
z

torch.mm()

The functionality ： Realize matrix multiplication in Linear Algebra （matrix multiplication）：(n×m) × (m×p) = (n×p) .

This function does not allow broadcasting ！

import torch
x = torch.randn((3,4))
y = torch.randn((4,5))
z = torch.mm(x,y)
z

torch.mv()

The functionality ： Realize matrix and vector （matrix × vector） Multiplication of , requirement input The shape of is n×m,output by torch.Size([m]) One dimension of tensor.

import torch
x = torch.randn((3,4))
y = torch.randn(4)
z = torch.mv(x,y)
z