Differentation

Forward Mode Automatic Differentiation

uses dual numbers

implemented in tensor flow

see also Reverse Mode Automatic Differentiation

Fast,Flexible,Exact