1 Summary of important user-visible changes for optim 1.2.0:
2 -------------------------------------------------------------------
4 ** Together with the new backend "lm_feasible" there is now a
5 complete suite of backends for optimization with linear and
6 general equality and inequality constraints, for scalar valued
7 objective functions and for array valued model function, which
8 features, a.o., honouring of constraints throughout optimization
9 and handling of structure-based parameters. The respective user
10 functions (frontends) are
12 nonlin_min nonlin_residmin nonlin_curvefit
14 together with a user function for statistics
18 ** The requirement of nonlin_min, nonlin_residmin, and
19 nonlin_curvefit for the general constraint functions being able to
20 honour an index of constraints has been removed, the respective
21 feature is still available by setting some options.
23 ** Makefile fixed to work with non-standard linker options e.g on
27 Summary of important user-visible changes for optim 1.1.0:
28 -------------------------------------------------------------------
30 ** The following functions are new optim 1.1.0:
32 powell cauchy nonlin_min
34 ** The following functions have been deprecated since they have been
35 implemented in Octave core:
37 fminunc_compat optimset_compat
39 ** The function `fmin' has been deprecated in favour of using `fminbnd'
40 directly. If one really wishes to use the short version, one can
41 easily create an alias on an octaverc file (see `doc startup') with
44 function out=fmin(varargin) out=fminbnd(varargin{:}); endfunction
46 ** The package Makefile has been adapted for compatibility with Octave 3.6.0.
48 ** Bugfixes on the functions:
52 ** The function `line_min' has a configurable setpesize and max evals.
54 ** Added possibility to restrict a parameter to samin.
56 ** Package is no longer automatically loaded.
59 Some important changes of the last versions of optim before 1.1.0:
60 ------------------------------------------------------------------
64 jacobs: complex step derivative approximation
66 nonlin_residmin, nonlin_curvefit: Frontends with a general
67 interface for constrained residual-based optimization. They
68 allow a.o. optimization of structure-based named parameters or
69 parameter-arrays. A backend is included, which is derived from
70 the algorithm of leasqr, but enables feasible-path optimization
71 with linear and general constraints.