+NLopt 2.4.2 (20 May 2014)
+
+* Fix CRS for empty dimensions (where lower == upper bound) (issue #13).
+
+* Improvements to CMake (thanks to @xantares) and Windows builds (issue #12).
+
+* Fix guile2 compatibility (issue #21).
+
+NLopt 2.4.1 (19 November 2013)
+
+* Use cdecl calling convention instead of stdcall on Win32, to
+ simplify shared-library usage and for consistency with Win64.
+
+NLopt 2.4 (2 November 2013)
+
+* New genetic algorithm ESCH, thanks to Carlos Henrique da Silva Santos.
+
+* Fix swig dir for VPATH builds, thanks to Sandro Vitenti for the bug report.
+
+* Use python-config program in the configure script to find the include
+ directories for Python, if possible (may be overridden by PYTHON_CONFIG
+ environment variable).
+
+* Bugfix in copy constructor for C++ and Python interfaces.
+
+* Bugfix in return value for setting min/max objective.
+
+* Handle negative rescalings in COBYLA and BOBYQA.
+
+* Plugin installation honors the configure --prefix, if any
+ (thanks to xantares@github).
+
+* Various compilation fixes for compatibility with recent software.
+
+NLopt 2.3.1 (16 September 2012)
+
+* Bug fix: COBLYA should return ROUNDOFF_LIMITED rather than XTOL_REACHED
+ if the trust-region radius underflows to zero; thanks to David Liu.
+
+* Bug fix: incorrect return value from set min/max objective.
+
+* Handle case of negative rescalings (from negative dx) in COBYLA and BOBYQA;
+ thanks to Alexander Law for the bug report
+
+NLopt 2.3 (20 July 2012)
+
+* In Matlab/Octave interface, make returning NaN from the objective
+ equivalent to nlopt_force_stop; thanks to Norman Violet for the suggestion.
+
+* Added CCSA-quadratic (NLOPT_LD_CCSAQ), similar to MMA.
+
+* Added interface for supplying a preconditioner (approximate Hessian);
+ currently only supported in CCSAQ.
+
+* When adding mconstraints, allow tol==NULL as synonym for zero tolerances.
+
+* Added missing NLOPT_LD_SLSQP constant in Matlab/Octave.
+
+* Lower tolerance for dual optimization in MMA/CCSAQ; thanks to
+ Christophe Leruste for the problem report
+
+* Fixed bug in timer, thanks to William Vaughn for the patch.
+
+* Bug fix to convergence test in sbplx; thanks to Douglas Bates.
+
+NLopt 2.2.4 (9 June 2011)
+
+* Fixed linking problem for C++ and Python shared libraries on MacOS X;
+ thanks to Volker Lorrmann for the bug report.
+
+NLopt 2.2.3 (8 June 2011)
+
+* Fixed additional re-entrancy problem in Luksan algorithms missed
+ in the previous version; thanks to Gert Wollny for the bug report.
+
+* Fixed set/get vector_storage in Fortran interface for Fortran
+ compilers with all-uppercase linkage.
+
+NLopt 2.2.2 (26 May 2011)
+
+* Added set_vector_storage API to modify the memory usage and the
+ subspace dimension for low-storage quasi-Newton methods.
+
+* Fixed missing support for maxtime stopping criteria in Luksan and
+ ORIG_DIRECT algorithms; thanks to Jurgen Werner for the bug report.
+
+* Fixed algorithms to support the case where the lower and upper bounds
+ are equal for some variables (which effectively eliminates those
+ variables from optimization).
+
+* Added missing xtol check to SLSQP, which caused erroneous ROUNDOFF_LIMITED
+ error codes to be returned; thanks to Alexander Riess for the bug report.
+
+* Fixed slight overcounting of function evaluations for maxeval check
+ in SLSQP.
+
+* Fixed deprecated API to support xtol_abs == NULL for backward
+ compatibility (thanks to Francesco Biscani for the bug report).
+
+* Made Luksan algorithms (e.g. LBFGS and other quasi-Newton methods)
+ re-entrant; thanks to Gert Wollny for the bug report. (Fixes
+ Gentoo bug #368685.)
+
+* Fixed bug in DIRECT (not ORIG_DIRECT), where a typo caused suboptimal
+ convergence in some cases; thanks to Sinisa Hristov for the bug report.
+
+NLopt 2.2.1 (6 September 2010)
+
+* If you compile nlopt.h with the NLOPT_DLL_EXPORT symbol #defined,
+ it now uses the dllexport directive (under Windows), useful for
+ compiling an NLopt DLL under Microsoft compilers; thanks to Benoit
+ Scherrer for the suggestion.
+
+* Handle case where copysign function is missing, e.g. on Windows;
+ thanks to Benoit Scherrer for the bug report.
+
+* Remove C99-style mixed declaration and code in a couple files, so
+ that code compiles in C89; thanks to Benoit Scherrer for the bug report.
+
+* Removed a few compiler warnings under Microsoft compilers; thanks
+ to Benoit Scherrer for the bug report.
+
+* Export nlopt_get_algorithm_name function on Windows; thanks to Ofek
+ Shilon for the bug report.
+
+* Don't use dllimport directive with lcc on Windows (which doesn't
+ support it); thanks to Laurent Vanbeylen for the bug report.
+
+* Update Nodedal README directory to indicate that Nocedal's LBFGS code
+ is now available under the GPL, and therefore may be distributed with
+ a future NLopt version (although Luksan's LBFGS code already works well).
+
+* Bug fix in set/get_xtol_abs; thanks to David Rivest-Henault for the report.
+
+NLopt 2.2 (15 July 2010)
+
+* Added SLSQP algorithm for gradient-based local optimization with
+ nonlinear constraints via sequential quadratic programming, based
+ on the implementation by Dieter Kraft that was adapted for SciPy.
+
+* Modified BOBYQA and COBYLA algorithms to support unequal initial
+ step sizes in different directions; thanks to Tom Fiddaman for pointing
+ out the need for this in the case where different directions have
+ very different scales.
+
+* Added Python module docstring; thanks to Sebastian Walter for the suggestion.
+
+* Added GUILE_INSTALL_DIR variable to allow the user to change the
+ Guile installation directory.
+
+* Added Fortran interface for vector-valued constraints.
+
+* Throw correct exceptions in Python for the add_*constraint functions;
+ thanks to Dmitrey Kroshko for the bug report.
+
+* Support forced stop and exceptions in ORIG_DIRECT algorithm.
+
+* Remove arbitrary 1e20 upper bound on function values from ORIG_DIRECT
+ code.
+
+* Bugfix in C++ interface (and some other language front-ends) when
+ deallocating the nlopt_opt object in cases like MLSL where
+ local_optimizer is used; thanks to Jurgen Werner for the bug report.
+
+NLopt 2.1.2 (8 July 2010)
+
+* The Python mconstraint (vector-valued constraint) functions
+ now pass a 2-dimensional array for the gradient argument, rather
+ than a flattened 1d array.
+
+* Improved handling of exceptions and forced stops for constrained
+ optimization, making sure that no constraints are evaluated after
+ the stop.
+
+* Return an NLOPT_INVALID_ARGS error if more than n equality constraints
+ are added in an n-dimensional problem.
+
+* Fix bug that could cause spurious NLOPT_INVALID_ARGS errors when
+ adding constraints under rare circumstances.
+
+* Eliminate a few small memory leaks that could occur under error conditions.
+
+NLopt 2.1.1 (7 July 2010)
+
+* More robust configure check for Python include directories, via
+ distutils.sysconfig.get_python_inc() and numpy.get_include().
+ Thanks to Nathaniel Smith for the tip.
+
+* Bug fix in Guile interface: added missing prefix to nlopt-version-major
+ etcetera.
+
+NLopt 2.1 (6 July 2010)
+
+* New vector-valued constraint feature; thanks to Dmitrey Kroshko of OpenOpt
+ for the suggestion.
+
+* COBYLA now accepts equality constraints.
+
+* Guard against multiple inclusion in nlopt.hpp; thanks to Saul
+ Thurrowgood for the suggestion.
+
NLopt 2.0.2 (17 June 2010)
* Fixed compilation failure in Microsoft Visual Studio, due to
~ianmdlvl / nlopt.git / blobdiff