2010-04-07 Brian Gough <bjg@network-theory.co.uk>
* test.c (test_fdf): handle case of GSL_ENOPROG from early returns
* test_funcs.c (simpleabs_f): new test function with non-zero
first derivatives around minimum
* steepest_descent.c (steepest_descent_iterate): return early if
trial point does not move within machine precision
* directional_minimize.c (intermediate_point): return early with
step=0 if trial point does not move from initial point within
machine precision
2009-08-05 Brian Gough <bjg@network-theory.co.uk>
* simplex2.c (contract_by_best): update the size and
center to avoid old values being used
* test.c (main): added a testcase for the spring function
2009-07-11 Brian Gough <bjg@network-theory.co.uk>
* simplex2.c (nmsimplex_set_rand): provide alternative
initialisation which randomizes the initial simplex
2009-07-09 Brian Gough <bjg@network-theory.co.uk>
* fminimizer.c (gsl_multimin_fminimizer_free): handle NULL
argument in free
* fdfminimizer.c (gsl_multimin_fdfminimizer_free): handle NULL
argument in free
2008-11-29 Brian Gough <bjg@network-theory.co.uk>
* test.c (test_f): extended fminimizer test to allow type to be
passed in as a parameter
* simplex2.c: use BLAS, keep track of center in state to avoid
unnecessary computation, compute size as RMS value to allow linear
update.
2008-07-03 Brian Gough <bjg@hp2.network-theory.co.uk>
* Makefile.am (INCLUDES): use top_srcdir instead of top_builddir
2008-05-06 Brian Gough <bjg@network-theory.co.uk>
* simplex.c (nmsimplex_iterate): fix search for second highest
point
2007-07-30 Brian Gough <bjg@network-theory.co.uk>
* history.c: removed (unused file)
2007-02-20 Brian Gough <bjg@network-theory.co.uk>
* vector_bfgs2.c (vector_bfgs2_iterate): use positive step size
2007-02-17 Brian Gough <bjg@network-theory.co.uk>
* linear_minimize.c (minimize): return GSL_ENOPROG for roundoff
2007-02-14 Brian Gough <bjg@network-theory.co.uk>
* linear_minimize.c: made all functions static
* linear_wrapper.c: made all functions static
2007-02-08 Brian Gough <bjg@network-theory.co.uk>
* linear_wrapper.c: convert multidimensional function to
one-dimensional for line minimisation
* linear_minimize.c: one-dimensional minimisation from Fletcher
* vector_bfgs2.c: added Fletcher's implementation
2006-02-18 Brian Gough <bjg@network-theory.co.uk>
* vector_bfgs.c (vector_bfgs_iterate): avoid division by zero if
dxdg == 0
2003-07-24 Brian Gough <bjg@network-theory.co.uk>
* simplex.c (nmsimplex_set): changed index variable i from int to
size_t
2003-04-17 Brian Gough <bjg@network-theory.co.uk>
* simplex.c (nmsimplex_iterate): bug fix to find the second
highest point correctly
* vector_bfgs.c (vector_bfgs_iterate): no need to update g0norm on
each downhill step, since g0norm is the norm for the initial
gradient.
* conjugate_pr.c (conjugate_pr_iterate): no need to update g0norm
on each downhill step, since g0norm is the norm for the initial
gradient.
* conjugate_fr.c (conjugate_fr_iterate): no need to update g0norm
on each downhill step, since g0norm is the norm for the initial
gradient.
Sun Sep 30 20:50:00 2002 Tuomo Keskitalo <tkeskita@pt.hut.fi>
* Added Nelder-Mead Simplex optimization algorithm and
fminimizer structure.
Sun Feb 10 21:57:36 2002 Brian Gough <bjg@network-theory.co.uk>
* conjugate_pr.c (conjugate_pr_iterate): return error ENOPROG if
cannot find downward step
* conjugate_fr.c (conjugate_fr_iterate): return error ENOPROG if
cannot find downward step
* vector_bfgs.c (vector_bfgs_iterate): return error ENOPROG if
cannot find downward step
Thu Oct 25 11:56:06 2001 Brian Gough <bjg@network-theory.co.uk>
* added a tolerance parameter for the line minimizations
Wed Oct 24 23:18:46 2001 Brian Gough <bjg@network-theory.co.uk>
* modified all routines to use a single minimiztion iteration,
instead of nested iterations for line and gradient search.
Thu Oct 18 22:56:52 2001 Brian Gough <bjg@network-theory.co.uk>
* renamed gsl_multimin_f_minimizer to gsl_multimin_fminimizer for
consistency with rest of the library
* renamed gsl_multimin_fdf_minimizer to gsl_multimin_fdfminimizer
for consistency with rest of the library
Mon Oct 8 21:41:51 2001 Brian Gough <bjg@network-theory.co.uk>
* diff.c (gsl_multimin_diff): pass params argument using
GSL_MULTIMIN_FN_EVAL (3 occurrences)
Sun Jul 15 17:54:15 2001 Brian Gough <bjg@network-theory.co.uk>
* fdfminimizer.c (gsl_multimin_fdf_minimizer_alloc): eliminated
use of interval type
Sat Apr 28 11:29:08 2001 Brian Gough <bjg@network-theory.co.uk>
* diff.c (gsl_multimin_diff): made indices unsigned
Mon Apr 23 13:22:31 2001 Brian Gough <bjg@network-theory.co.uk>
* gsl_multimin.h diff.c: made starting_point const throughout to
avoid compiler warnings
* made internal functions static
* gsl_multimin.h: added missing prototype for gsl_multimin_diff
Tue Apr 17 22:15:37 2001 Brian Gough <bjg@network-theory.co.uk>
* gsl_multimin.h: added missing prototype for gsl_multimin_compute_ep
Sun Feb 18 16:35:21 2001 Brian Gough <bjg@network-theory.co.uk>
* fdfminimizer.c (gsl_multimin_fdf_minimizer_alloc): modified to
account for change in calling convection of
gsl_min_fminimizer_alloc
Fri May 5 16:08:34 2000 Brian Gough <bjg@network-theory.co.uk>
* test.c (test_fdf): fixed warning about "control reaches end of
non-void function" by changing test_fdf to return type void
Tue May 2 19:20:46 2000 Brian Gough <bjg@network-theory.co.uk>
* test.c (main): added return gsl_test_summary() to main, so that
test results are returned through the exit status.
Mon Feb 14 13:12:16 2000 Brian Gough <bjg@network-theory.co.uk>
* made all internal functions static