$Id$
/* --------------------------------------------------------------------------
CppAD: C++ Algorithmic Differentiation: Copyright (C) 2003-15 Bradley M. Bell

CppAD is distributed under multiple licenses. This distribution is under
the terms of the
                    Eclipse Public License Version 1.0.

A copy of this license is included in the COPYING file of this distribution.
Please visit http://www.coin-or.org/CppAD/ for information on other licenses.
-------------------------------------------------------------------------- */

$begin whats_new_15$$
$dollar @$$
$spell
	dimensioned
	cskip_op
	adolc
	zdouble
	expm1
	eigen
	expm1
	atanh
	acosh
	asinh
	src
	op
	onetape
	CppAD
	colpack
	hes
	cppad
	Jacobian
	gettimeofday
	cmake
	hpp
	dir
	includedirs
	libdirs
	datadir
	docdir
	std
	cxx
	erf
	var
	NumRes
	chrono
	aeps
	CondExpGt
	hasnan
	Adolc
$$

$section CppAD Changes and Additions During 2015$$

$head Introduction$$
This section contains a list of the changes to CppAD during 2015
(in reverse order by date).
The purpose of this section is to
assist you in learning about changes between various versions of CppAD.

$head 06-16$$
$list number$$
There were some
$cref/unknown/cppad_assert/Unknown/$$ asserts
when the sparsity pattern $icode p$$ in
$cref/sparse_jacobian/sparse_jacobian/p/$$ and
$cref/sparse_hessian/sparse_hessian/p/$$ was not properly dimensioned.
These have been changed to
$cref/known/cppad_assert/Known/$$ asserts to give better
error reporting.
$lnext
In the special case where sparse Hessian $cref/work/sparse_hessian/work/$$ or
sparse Jacobian $cref/work/sparse_hessian/work/$$ was specified and the
set of elements to be computed was empty, the work vector is empty after
the call (and it appears to need to be calculated on subsequent calls).
This resulted in a bug when the sparsity pattern was not provided
on subsequent calls (and has been fixed).
$lend

$head 06-11$$
$list number$$
Some C++11 features were not being taken advantage of after the change on
$cref/05-10/whats_new_15/05-10/$$. To be specific, move semantics,
the high resolution clock, and null pointers.
This has been fixed.
$lnext
In the example $cref zdouble.cpp$$, the vector $code a1z$$ was not
properly dimensioned.
This has been fixed and the dimensions of all the variables have been
clarified.
$lend


$head 06-09$$
Add $cref/abort_op_index/WishList/abort_op_index/$$ to the wish list.

$head 06-07$$
Add $cref/absolute zero/WishList/Absolute Zero/$$
and $cref/numeric_limits/WishList/Numeric Limits/$$ items to the wish list.
Remove the multiple directions with list item.

$head 05-26$$

$subhead cond_exp_1$$
There was a problem using
$cref/conditional expressions/CondExp/$$
with $cref/multiple levels of AD/mul_level/$$ where
the result of the conditional expression might not be determined
during forward mode.
This would generate an assert of the form:
$codei%
	Error detected by false result for
		IdenticalPar(%side%)
    at line %number% in the file
		%...%/cppad/local/cskip_op.hpp
%$$
where $icode side$$ was $code left$$ or $code right$$
and $icode number$$ was the line number of an assert in $code cskip_op.hpp$$.
This has been fixed.

$subhead cond_exp_2$$
There was a problem with using
$cref/conditional expressions/CondExp/$$ and $cref/reverse mode/reverse/$$
with $cref/multiple levels of AD/mul_level/$$.
This was problem was represented by the file $code bug/cond_exp_2.sh$$.

$list number$$
The problem above has been fixed by adding the base type $code zdouble$$, see
$cref/CppAD motivation/zdouble/Motivation/CppAD/$$ for this new type.
$lnext
The sections
$cref mul_level$$,
$cref change_param.cpp$$,
$cref mul_level.cpp$$,
and $cref mul_level_ode.cpp$$ were changed to use $cref zdouble$$.
$lnext
The $cref/adolc/adolc_prefix/$$ multi-level examples
$cref mul_level_adolc.cpp$$ and $cref mul_level_adolc_ode.cpp$$
were changed to mention the limitations because Adolc does not have an
$cref/absolute zero/zdouble/Absolute Zero/$$.
$lnext
The example above were also changed so that AD variable names that
indicated the level of AD for the variable.
$lnext
The base type requirements were modified to include mention of
$cref/absolute zero/base_require/Absolute Zero/$$.
In addition, the base type requirements
$cref/warning/base_require/Warning/$$ is now more informative.
$lend

$head 05-11$$
Reorganize the $cref unary_standard_math$$ documentation.

$head 05-10$$
$list number$$
Add the exponential minus one function $cref log1p$$.
If you are defining your own base type,
note that $cref/log1p/base_std_math/erf, asinh, acosh, atanh, expm1, log1p/$$
was added to the base type requirements.
$lnext
Use the single preprocessor flag
$code CPPAD_USE_CPLUSPLUS_2011$$ to signal that the functions
$cref/erf, asinh, acosh, atanh, expm1, log1p
	/base_std_math
	/erf, asinh, acosh, atanh, expm1, log1p
/$$
are part of the base type requirements.
$lend

$head 05-09$$
$list number$$
Add the exponential minus one function $cref expm1$$.
If you are defining your own base type,
note that $cref/expm1/base_std_math/erf, asinh, acosh, atanh, expm1, log1p/$$
was added to the base type requirements.
$lnext
Fix some warnings about comparing signed and unsigned integers
when using $cref/eigen/cppad_testvector/eigen/$$ for the CppAD test vector.
(The eigen vector $code size()$$ function returns an $code int$$ instead of a
$code size_t$$.)
$lend

$head 05-08$$
$list number$$
Add the inverse hyperbolic sine function $cref atanh$$.
If you are defining your own base type,
note that $cref/atanh/base_std_math/erf, asinh, acosh, atanh, expm1, log1p/$$
was added to the base type requirements.
$lnext
Fix a bug in the implementation of the $code acosh$$
multiple direction forward mode $cref forward_dir$$
(when compiler has $cref/acosh/acosh/CPPAD_USE_CPLUSPLUS_2011/$$).
$lend

$head 05-07$$
Add the inverse hyperbolic sine function $cref acosh$$.
If you are defining your own base type,
note that $cref/acosh/base_std_math/erf, asinh, acosh, atanh, expm1, log1p/$$
was added to the base type requirements.

$head 05-05$$
Add the inverse hyperbolic sine function $cref asinh$$.
If you are defining your own base type,
note that $cref/asinh/base_std_math/erf, asinh, acosh, atanh, expm1, log1p/$$
was added to the base type requirements.

$head 04-18$$
In the sparse jacobian and sparse hessian calculations,
If $icode work$$ is present, and has already been computed,
the sparsity pattern $icode p$$ is not used.
This has been added to the documentation; see
$cref/sparse jacobian/sparse_jacobian/work/p/$$ and
$cref/sparse hessian/sparse_hessian/work/p/$$ documentation
for $icode work$$ and $icode p$$.

$head 03-13$$
Remove the syntax
$codei%
	AD<%Base%> %y% = %x%
%$$
for the $cref/AD constructor/ad_ctor/$$ documentation because it does not
work when the constructor is $cref/explicit/ad_ctor/x/explicit/$$.
Also document the restriction that the constructor in the
$cref/assignment/ad_assign/$$ must be implicit.

$head 03-06$$
The developers of the
$href%https://github.com/kaskr/adcomp%TMB%$$ package reported that
for large $cref ADFun$$ tapes, the $cref optimize$$ routine uses
a large amount of memory because it allocates a standard set for
each variable on the tape. These sets are only necessary for variables in
$cref/conditional expressions/condexp/$$ that can be skipped once
the independent variables have a set value.
The problem has been reduced by using a NULL pointer for the empty set
and similar changes. It still needs more work.

$head 02-28$$
It used to be the case that the
$cref/Reverse mode/reverse/$$ would propagate $cref nan$$
through the $cref/conditional expression/condexp/$$
case that is not used.
For example, if
$codep
	Independent(ax);
	AD<double> aeps = 1e-10;
	ay[0] = CondExpGt( ax[0], aeps, 1.0/ax[0], 1.0/aeps );
	ADFun<double> f(ax, ay);
$$
The corresponding reverse mode calculation,
at $code x[0] = 0.0$$, would result in
$codep
	Error detected by false result for
    	! ( hasnan(value) && check_for_nan_ )
$$
This has been fixed so that only the conditional expression case that is used
affects the reverse mode results.
The example $cref cond_exp.cpp$$ was changed to reflect this
(a check for $code nan$$ was changed to a check for zero).
Note that this fix only works when
$cref/IdenticalPar/base_identical/Identical/IdenticalPar/$$ is true for the
base type of the result in the conditional expression; e.g.,
one can still get a $code nan$$ effect from the case that is not selected
when using $code AD< AD<double> >$$ conditional expressions.

$head 02-18$$
If the compiler supports the c++11 feature
$code std::chrono:high_resolution_clock$$ then use it for
the $cref elapsed_seconds$$ function.

$head 02-16$$
The new example $cref sub_sparse_hes.cpp$$ shows one way to
compute a Hessian for a subset of variables without having to compute
the sparsity pattern for the entire functions.

$head 02-14$$
Fix another bug in the derivative calculations for the
c++11 version of the error function; see
$cref/CPPAD_USE_CPLUSPLUS_2011/erf/CPPAD_USE_CPLUSPLUS_2011/$$.

$head 02-11$$
Fix a bug in the optimization of conditional expressions. To be specific,
if $cref/NDEBUG/faq/Speed/NDEBUG/$$ is not defined, one could get
an assert with the message:
$codep
	Error detected by false result for
		var_index_ >= NumRes(op_)
$$

$head 02-10$$
The change on $cref/2014-12-23/whats_new_14/12-23/$$ introduced a
bug when the c++11 version of the error function was used with
an $cref/optimized/optimize/$$ function. see
$cref/CPPAD_USE_CPLUSPLUS_2011/erf/CPPAD_USE_CPLUSPLUS_2011/$$.
There was also a bug in the sparsity calculations for when
this erf function was included.
These bugs have been fixed.

$head 02-09$$
The test $code test_more/optimize.cpp$$ was failing on some systems
because an exactly equality check should have been a near equal check.
This has been fixed.

$head 02-07$$
On some systems, the library
corresponding to $code speed/src$$ could not be found.
This library is only used for testing and so has been changed to always be
static (hence does not need to be found at run time).

$head 02-06$$
There was a bug in the coloring method change on
$cref/2015-01-07/whats_new_15/01-07/$$.
To be specific, $icode%work%.color_method%$$ was not being set
to $code "cppad.symmetric"$$ after $icode%work%.color_method.clear()%$$.
This has been fixed.

$head 02-04$$
$list number$$
Enable the same install of CppAD to be used both with and without C++11
features; e.g., with both $code g++ --std=c++11$$ and with
$code g++ --std=c++98$$. Previously if the
$cref/cppad_cxx_flags/cmake/cppad_cxx_flags/$$ specified C++11,
then it could only be used in that way.
$lnext
The $cref/cmake command/cmake/CMake Command/$$ now requires the version
of $code cmake$$ to be greater than or equal 2.8
(due a bug in $code cmake$$ version 2.6).
$lend

$head 02-03$$
Improved the searching for the boost multi-threading library
which is used for by the $cref team_bthread.cpp$$ case of the
$cref thread_test.cpp$$ example and test.

$head 02-02$$
Improve the documentation for the
$cref/cmake command/cmake/CMake Command/$$ line options
$codei%
	cmake_install_%dir%
%$$
for $icode dir$$ equal to
$code prefix$$, $code postfix$$, $code includedirs$$, $code libdirs$$,
$code datadir$$, and $code docdir$$.

$head 01-30$$
Fix bug in $cref link_sparse_hessian$$ speed test introduced on
$cref/01-09/whats_new_15/01-09/$$ below.

$head 01-29$$
Fix some warnings generated by $code g++ 4.9.2$$.

$head 01-26$$
The change of global variables to local in
$code cppad/local/op_code.hpp$$ on $cref/2014-50-14/whats_new_14/05-14/$$
created a bug in $cref parallel_ad$$ (some local statics needed to
be initialized). This has been fixed.

$head 01-23$$
There was a bug in the $cref cmake$$ install detection of compiler features.
One symptom of this bug was that on systems that had the $code gettimeofday$$
function, the cmake install would sometimes report
$codei%
	cppad_has_gettimeofday = 0
%$$
This has been fixed.

$head 01-21$$
The deprecated $cref auto_tools$$ procedure had a bug in the
detection of when the size of an $code unsigned int$$
was the same as the size of a $code size_t$$. This has been fixed.

$head 01-20$$
$list number$$
The new $cref compare_change$$ interface has been created
and the old $cref CompareChange$$ function has been deprecated;
see the $cref compare_change.cpp$$ example.
This enables one to determine the source code during taping
that corresponds to changes in the comparisons during
$cref/zero order forward/forward_zero/$$ operations; see
$cref/abort_op_index/Independent/abort_op_index/$$.

$lnext
This new $cref compare_change$$ interface can detect comparison changes
even if $cref/NDEBUG/Faq/Speed/NDEBUG/$$ is defined and
even if $cref/f.optimize()/optimize/$$ has been called.
The deprecated function $code CompareChange$$ used to always return zero after
$codei%
	%f%.optimize()
%$$
and was not even defined when $code NDEBUG$$ was defined.
There was a resulting speed effect for this; see the wish list
$cref/compare change/WishList/Optimization/Comparison Changes/$$ entry.

$lnext
The date when some features where deprecated has been added to the
documentation. For example, see
$cref/Deprecated 2006-12-17/include_deprecated/Deprecated 2006-12-17/$$.
$lend

$head 01-09$$
$list number$$
The change 01-07 below included (but did not mention) using
a sparse, instead of full, structure for the Hessian in the test.
This has also been done for the
$cref/sparse Jacobian/link_sparse_jacobian/$$ test.
$lnext
For both the
$cref/sparse_jacobian/link_sparse_jacobian/$$ and
$cref/sparse_hessian/link_sparse_hessian/$$ tests,
the sparse function is only chosen once
(it used to be different for every repeat).
This reduced the amount of computation not connected what is being tested.
It also make the $cref/onetape/speed_main/option_list/onetape/$$ a valid
option for these tests.
$lnext
There was a bug in the
$cref/multiple direction forward/forward_dir/$$ routine.
Results for function values that are
$cref/parameter/parvar/$$ were not being computed properly
(all the derivatives are zero in this case).
This has been fixed.
$lend


$head 01-07$$
The following changes were merged in from the $code color_hes$$ branch:
$list number$$
Specify the type of
$cref/coloring/sparse_hessian/work/color_method/$$
for the sparse hessian calculations.
To be specific, instead of $code "cppad"$$ and $code "colpack"$$,
the choices are $code "cppad.symmetric"$$, $code "cppad.general"$$,
and $code "colpack.star"$$.
This is not compatible with the change on
$cref/01-02/whats_new_15/01-02/$$, which was so recent
that this should not be a problem.
$lnext
The $cref/n_sweep/link_sparse_hessian/n_sweep/$$ values were
not being returned properly by
$cref cppad_sparse_hessian.cpp$$ and
$cref adolc_sparse_hessian.cpp$$.
The CppAD version has been fixed and the ADOL-C version
has been set to zero.
$lnext
The $cref link_sparse_hessian$$ example case was to sparse for good
testing (by mistake).
This has been fixed.
$lnext
Add $icode n_sweep$$ to
$cref/link_sparse_hessian/link_sparse_hessian/n_sweep/$$ and
$cref/speed_main/speed_main/Speed Results/n_sweep/$$.
$lnext
Change the $code cppad$$ sparse Hessian
$cref/color_method/sparse_hessian/work/color_method/$$
to take advantage of the symmetry of the Hessian
(in a similar fashion to the $code colpack$$ coloring method).
$lend

$head 01-02$$
Added to option to uses
$cref/colpack/colpack_prefix/$$ for the sparse Hessian
$cref/coloring method/sparse_hessian/work/color_method/$$;
see the example $cref colpack_hes.cpp$$.

$end