Numpy Axis

In numpy, dimensions of a tensor is called the axis.

• Tensors are equivalent to lists of lists.
  • The higher the axis, the more “inner” list we are talking about.
  • Each axis’s value determines the number or elements that array has
• A tensor of shape (=dimension) (a,b,c) has a rows (axis 0), b columns (axis 1), and c of axis 2.
  • For example, a tensor with shape (10,2,20)

Tensor Multiplication §

When matrix A with shape (a1,a2) is multiplied by tensor B of shape (b1,b2):

• a2 must be equal to b1
• resulting matrix will have shape (a1,b2). (in Linear Algebra terms, it is a composite linear transformation of both $A$ and $B$). When you have multiple matrix multiplications you can combine that into one.