lines 5-194 of file: xrst/reverse/reverse_two.xrst

{xrst_begin reverse_two}
{xrst_spell
   dw
}

Second Order Reverse Mode
#########################

Syntax
******
*dw* = *f* . ``Reverse`` (2, *w* )

Purpose
*******
We use :math:`F : \B{R}^n \rightarrow \B{R}^m` to denote the
:ref:`glossary@AD Function` corresponding to *f* .
Reverse mode computes the derivative of the :ref:`Forward-name` mode
:ref:`Taylor coefficients<glossary@Taylor Coefficient>`
with respect to the domain variable :math:`x`.

x^(k)
*****
For :math:`k = 0, 1`,
the vector :math:`x^{(k)} \in \B{R}^n` is defined as the value of
*x_k* in the previous call (counting this call) of the form

   *f* . ``Forward`` ( *k* , *x_k* )

If there is no previous call with :math:`k = 0`,
:math:`x^{(0)}` is the value of the independent variables when the
corresponding
AD of *Base*
:ref:`operation sequence<glossary@Operation@Sequence>` was recorded.

Capital W
*********
The functions
:math:`W_0 : \B{R}^n \rightarrow \B{R}` and
:math:`W_1 : \B{R}^n \rightarrow \B{R}`
are defined by

.. math::
   :nowrap:

   \begin{eqnarray}
   W_0 ( u ) & = & w_0 * F_0 ( u ) + \cdots + w_{m-1} * F_{m-1} (u)
   \\
   W_1 ( u ) & = &
   w_0 * F_0^{(1)} ( u ) * x^{(1)}
      + \cdots + w_{m-1} * F_{m-1}^{(1)} (u) * x^{(1)}
   \end{eqnarray}

This operation computes the derivatives

.. math::
   :nowrap:

   \begin{eqnarray}
   W_0^{(1)} (u) & = &
      w_0 * F_0^{(1)} ( u ) + \cdots + w_{m-1} * F_{m-1}^{(1)} (u)
   \\
   W_1^{(1)} (u) & = &
      w_0 * \left( x^{(1)} \right)^\R{T} * F_0^{(2)} ( u )
      + \cdots +
      w_{m-1} * \left( x^{(1)} \right)^\R{T} F_{m-1}^{(2)} (u)
   \end{eqnarray}

at :math:`u = x^{(0)}`.

f
*
The object *f* has prototype

   ``const ADFun`` < *Base* > *f*

Before this call to ``Reverse`` , the value returned by

   *f* . ``size_order`` ()

must be greater than or equal two (see :ref:`size_order-name` ).

Lower w
*******
The argument *w* has prototype

   ``const`` *Vector* & *w*

(see :ref:`reverse_two@Vector` below)
and its size
must be equal to *m* , the dimension of the
:ref:`fun_property@Range` space for *f* .

dw
**
The result *dw* has prototype

   *Vector* *dw*

(see :ref:`reverse_two@Vector` below).
It contains both the derivative
:math:`W^{(1)} (x)` and the derivative :math:`U^{(1)} (x)`.
The size of *dw*
is equal to :math:`n \times 2`,
where :math:`n` is the dimension of the
:ref:`fun_property@Domain` space for *f* .

First Order Partials
====================
For :math:`j = 0 , \ldots , n - 1`,

.. math::

   dw [ j * 2 + 0 ]
   =
   \D{ W_0 }{ u_j } \left( x^{(0)} \right)
   =
   w_0 * \D{ F_0 }{ u_j } \left( x^{(0)} \right)
   + \cdots +
   w_{m-1} * \D{ F_{m-1} }{ u_j } \left( x^{(0)} \right)

This part of *dw* contains the same values as are returned
by :ref:`reverse_one-name` .

Second Order Partials
=====================
For :math:`j = 0 , \ldots , n - 1`,

.. math::

   dw [ j * 2 + 1 ]
   =
   \D{ W_1 }{ u_j } \left( x^{(0)} \right)
   =
   \sum_{\ell=0}^{n-1} x_\ell^{(1)} \left[
   w_0 * \DD{ F_0 }{ u_\ell }{ u_j } \left( x^{(0)} \right)
   + \cdots +
   w_{m-1} * \DD{ F_{m-1} }{ u_\ell }{ u_j } \left( x^{(0)} \right)
   \right]

Vector
******
The type *Vector* must be a :ref:`SimpleVector-name` class with
:ref:`elements of type<SimpleVector@Elements of Specified Type>`
*Base* .
The routine :ref:`CheckSimpleVector-name` will generate an error message
if this is not the case.

Hessian Times Direction
***********************
Suppose that :math:`w` is the *i*-th elementary vector.
It follows that for :math:`j = 0, \ldots, n-1`

.. math::
   :nowrap:

   \begin{eqnarray}
   dw[ j * 2 + 1 ]
   & = &
   w_i \sum_{\ell=0}^{n-1}
   \DD{F_i}{ u_j }{ u_\ell } \left( x^{(0)} \right) x_\ell^{(1)}
   \\
   & = &
   \left[ F_i^{(2)} \left( x^{(0)} \right) * x^{(1)} \right]_j
   \end{eqnarray}

Thus the vector :math:`( dw[1], dw[3], \ldots , dw[ n * q - 1 ] )`
is equal to the Hessian of :math:`F_i (x)` times the direction
:math:`x^{(1)}`.
In the special case where
:math:`x^{(1)}` is the *l*-th
:ref:`glossary@Elementary Vector` ,

.. math::

   dw[ j * 2 + 1 ] = \DD{ F_i }{ x_j }{ x_\ell } \left( x^{(0)} \right)

Example
*******
{xrst_toc_hidden
   example/general/reverse_two.cpp
   example/general/hes_times_dir.cpp
}
The files
:ref:`reverse_two.cpp-name`
and
:ref:`hes_times_dir.cpp-name`
contain a examples and tests of reverse mode calculations.
They return true if they succeed and false otherwise.

{xrst_end reverse_two}
