All this operation affect the tensor on which the method is called (self). No additional memory is created.
[self] add(value)
[self] add(tensor)
[self] add(value, tensor)
[self] mul(value)
[self] cmul(tensor)
[self] addcmul(value, tensor1, tensor2)
[self] div(value)
[self] cdiv(tensor)
[self] addcdiv(value, tensor1, tensor2)
[number] dot(tensor)
addT2dotT1(value, mat, vec)
addT1outT1(value, vec1, vec2)
addT2dotT2(value, mat1, mat2)
addT4dotT2(value, t4, t2)
addT2outT2(value, mat1, mat2)