1 NLopt 1.2 (18 November 2009)
3 * Added Powell's BOBYQA algorithm for box-constrained optimization
4 without derivatives, an improvement on NEWUOA.
6 * Added ISRES genetic algorithm, supporting nonlinearly constrained
9 * New functions nlopt_{set/get}_stochastic_population to provide
10 optional greater control over the random "population" sizes in
11 stochastic algorithms (although it still has a sensible default).
13 * Bug fix: remove extraneous text accidentally included in nlopt.f Fortran
16 * Bug fix: configure script now correctly handles Matlab installation
17 when MEX_INSTALL_DIR is specified manually by the user.
19 NLopt 1.1 (12 November 2009)
21 * configure script detects whether --enable-shared is required
22 in order to compile Matlab and Octave plugins (as is the case
23 on x86_64), and disables compilation of those plugins if
24 --enable-shared is not used.
26 * Added --without-octave and --without-matlab configure options to
27 disable Octave and Matlab plugins, respectively.
29 * Modified COBYLA algorithm to have better support for bound
32 * Added new NLOPT_ROUNDOFF_LIMITED failure code to indicate
33 cases in which optimization breaks down due to roundoff errors,
34 in which case it is possible that useful results were obtained.
36 * Experimental support for nonlinear equality constraints via
37 augmented-Lagrangian method.
39 * Support for compiling under Windows (ideally with MinGW) as a
40 DLL, although you have to manually add #define NLOPT_DLL
41 to nlopt.h *after* installing (after compiling NLopt).
43 * Added several checks for roundoff-related breakdown to NEWUOA code.
45 * When only a relative error tolerance is specified, no longer
46 fails to halt when exact convergence to zero is obtained.
48 * Workaround for incompatible qsort_r functions in BSD and GNU libc
49 by always using my own version; thanks to Wendy Vandoolaeghe
50 and Philippe Preux for the bug report and explanation.
52 * Workaround for gcc 3.4.x conflict with HUGE_VAL definition in Solaris
55 * Better identification of Matlab-plugin installation directory.
57 * Fixed identification of Octave-plugin installation directory for
58 recent Octave versions.
60 NLopt 1.0.1 (13 Nov. 2008)
62 * Allow user to override Matlab-plugin installation directory with
65 * Bug fix in my DIRECT code that prevented convergence (DIRECT-L unaffected).
67 * MLSL needs a nonzero default ftol_rel and/or xtol_rel to ensure that
68 its local searches terminate; use roughly machine precision as defaults.
70 NLopt 1.0 (11 Nov. 2008)
72 * Initial public release.