Forward, reverse and mixed mode automatic differentiation combinators with a common API.
Typelevel "branding" is used to both prevent the end user from confusing infinitesimals and to limit unsafe access to the implementation details of each Mode.
Each mode has a separate module full of combinators.
computing simple derivatives.
AD. It is good for computing gradients in one pass. It generates a Wengert list (linear tape) using 'Data.Reflection'.
It is good for computing gradients in one pass. It generates a treelike tape that needs to be topologically sorted in the end.
for higher derivatives or large numbers of outputs.
higher derivatives of single input functions.
to each individual combinator.
While not every mode can provide all operations, the following basic operations are supported, modified as appropriate by the suffixes below:
'grad' computes the gradient (partial derivatives) of a function at a point.
'jacobian' computes the Jacobian matrix of a function at a point.
'diff' computes the derivative of a function at a point.
'du' computes a directional derivative of a function at a point.
'hessian' computes the Hessian matrix (matrix of second partial derivatives)
of a function at a point.
The following suffixes alter the meanings of the functions above as follows:
'''  also return the answer
'With' lets the user supply a function to blend the input with the output
'F' is a version of the base function lifted to return a 'Traversable' (or
'Functor') result
fbranching 'Stream'
formulation.
'0' means that the resulting derivative list is padded with 0s at the end.
'NoEq' means that an infinite list of converging values is returned rather
than truncating the list when they become constant.
Package Version  Update ID  Released  Package Hub Version  Platforms  Subpackages 

4.3.3bp150.1.3 info  GA Release  20180730  15 

