12009-07-25 Brian Gough <bjg@network-theory.co.uk> 2 3 * vegas.c (gsl_monte_vegas_params_get) 4 (gsl_monte_vegas_params_set): added get/set functions for params 5 (gsl_monte_vegas_chisq): added chisq accessor 6 7 * gsl_monte_vegas.h: added separate params struct 8 9 * miser.c (gsl_monte_miser_params_set) 10 (gsl_monte_miser_params_get): added get/set functions for params 11 12 * gsl_monte_miser.h: added separate params struct 13 142009-07-09 Brian Gough <bjg@network-theory.co.uk> 15 16 * vegas.c (gsl_monte_vegas_free): handle NULL argument in free 17 18 * plain.c (gsl_monte_plain_free): handle NULL argument in free 19 20 * miser.c (gsl_monte_miser_free): handle NULL argument in free 21 222009-02-10 Brian Gough <bjg@network-theory.co.uk> 23 24 * vegas.c (gsl_monte_vegas_integrate): use gsl_pow_int to compute 25 tot_boxes, avoids potentially inaccurate pow functions (MinGW). 26 272008-11-20 Brian Gough <bjg@network-theory.co.uk> 28 29 * vegas.c (gsl_monte_vegas_integrate): improve the chisq 30 calculation to avoid cancellation errors in the original 31 formula (fixes bug #24510) 32 33 * test.c (MONTE_ERROR_TEST): added test cases for negative chisq 34 in vegas 35 362008-07-03 Brian Gough <bjg@hp2.network-theory.co.uk> 37 38 * Makefile.am (INCLUDES): use top_srcdir instead of top_builddir 39 402008-04-26 Brian Gough <bjg@network-theory.co.uk> 41 42 * vegas.c (gsl_monte_vegas_integrate): compute running totals as 43 volatile double to prevent problems with excess precision, use 44 meaningful variable names for sum of squares, variance and sigma. 45 46 * test_main.c: compute the ensemble mean more accurately for the tests 47 48 * test.c: added a test for warm-start vegas, as well as cold-start 49 50 * miser.c (gsl_monte_miser_integrate): catch zero weights to avoid 51 division by zero 52 53 * test.c (main): added test for step-type function 54 552004-06-02 Brian Gough <bjg@network-theory.co.uk> 56 57 * test_main.c: handle the case where sd==0 && error[i] !=0 more 58 carefully 59 60Mon Apr 23 13:23:47 2001 Brian Gough <bjg@network-theory.co.uk> 61 62 * test_main.c: removed unused status variable 63 64Sat Jan 6 19:56:49 2001 Brian Gough <bjg@network-theory.co.uk> 65 66 * miser.c (gsl_monte_miser_free): fixed memory leak for s->hits_{r,l} 67 68 * vegas.c (gsl_monte_vegas_free): fixed memory leak for s->x 69 70Fri Dec 22 21:43:04 2000 Brian Gough <bjg@network-theory.co.uk> 71 72 * miser.c: removed max-min estimation method for subvolumes. We 73 will use the standard variance method and try to use a sufficient 74 number of points to estimate the variance reliably. 75 76Wed Dec 20 21:32:40 2000 Brian Gough <bjg@network-theory.co.uk> 77 78 * vegas.c: tidied up the algorithm, deal with cases of sigma = 0 79 explicitly. 80 81Sat Dec 9 14:19:53 2000 Brian Gough <bjg@network-theory.co.uk> 82 83 * reorganization and clean up 84 85Thu Nov 16 19:50:27 2000 Brian Gough <bjg@network-theory.co.uk> 86 87 * miser.c: rewrite to fix overflows and make calling conventions 88 consistent 89 90 * plain.c: rewrite to fix overflows and make calling conventions 91 consistent 92 93Thu Oct 26 20:06:36 2000 Brian Gough <bjg@network-theory.co.uk> 94 95 * plain.c (gsl_monte_plain_integrate): integer factor 96 calls*(calls-1) used in numerical expression can easily overflow, 97 changed to calls*(calls-1.0). 98 99Sat Oct 21 20:36:06 2000 Brian Gough <bjg@network-theory.co.uk> 100 101 * miser.c (gsl_monte_miser_integrate): fixed bug where hits_l was 102 used in place of hits_r 103 104Tue Sep 19 19:16:37 2000 Brian Gough <bjg@network-theory.co.uk> 105 106 * plain.c (gsl_monte_plain_alloc): initialise check_done to avoid 107 warning about use of unitialised memory 108 109 * vegas.c (gsl_monte_vegas_alloc): as above 110 111 * miser.c (gsl_monte_miser_alloc): as above 112 113 * plain.c miser.c: removed use of sprintf for error messages 114 115Wed May 31 19:50:19 2000 Brian Gough <bjg@network-theory.co.uk> 116 117 * miser_test.c (main): increased tolerances to allow tests to pass 118 with different compilers 119 120Mon May 15 15:26:22 2000 Brian Gough <bjg@network-theory.co.uk> 121 122 * vegas_test.c (main): added gsl_ieee_env_setup() to allow change 123 of precision in tests 124 125 * miser_test.c (main): ditto 126 127 * plain_test.c (main): ditto 128 129Fri Feb 26 14:49:56 1999 Brian Gough <bjg@netsci.freeserve.co.uk> 130 131 * Makefile.am: removed ..._LDFLAGS = -static since this is 132 specific to gcc 133 134 135 136Wed Nov 18 10:59:56 1998 Brian Gough <bjg@vvv.lanl.gov> 137 138 * use standard headers templates_on.h and templates_off.h instead 139 of source.h 140 141Tue Nov 17 16:49:12 1998 Brian Gough <bjg@vvv.lanl.gov> 142 143 * added #include <config.h> to all top-level source files 144 145 * plain.c (gsl_monte_plain_integrate): replaced myMAX by GSL_MAX 146 147 * utils.c: move macros around to avoid double definition 148 149Fri Aug 14 10:12:06 1998 Brian Gough <bjg@vvv.lanl.gov> 150 151 * Makefile.am: I needed to add utils.h to libgslmonte_a_SOURCES to 152 get it to work with my automake 153 154Thu Jul 30 17:31:29 1998 booth <booth@planck.pha.jhu.edu> 155 156 * gsl_monte_miser.h, miser.c: 157 Turn off the annoying warning in miser unless the user requests it. 158 159Wed Jul 29 20:24:54 1998 bjg <bjg@vvv.lanl.gov> 160 161 * Makefile.am, Makefile.in: some fixes to pass make distcheck 162 163 * Makefile.am, Makefile.in: 164 experimenting with new top level makefile.am 165 166Tue Jul 28 17:05:20 1998 booth <booth@planck.pha.jhu.edu> 167 168 * plain.c, vegas.c, miser.c: 169 make all the _free functions check their argument. Also, the init functions 170 now throw EFAULT if given a null pointer. 171 172 * gsl_monte_vegas.h, vegas.c: 173 vegas1, vegas2 and vegas3 all go away. Get them by setting the "stage" 174 variabe appropriately. Also, make _free check its argument. 175 176 * vegas.c, gsl_monte_vegas.h: minor cleanup prior to be change. 177 178 * vegas.c: minor changes, commiting by accident. 179 180Mon Jul 27 22:52:49 1998 bjg <bjg@vvv.lanl.gov> 181 182 * Makefile.in, Makefile.am: 183 fixed some of the include requirements for make dist 184 185Mon Jul 27 15:19:54 1998 booth <booth@planck.pha.jhu.edu> 186 187 * vegas_print.h, vegas_test.c, miser_test.c, vegas-print.c, vegas.c, Attic/gsl_vegas.h, Attic/gsl_vegas_print.h, gsl_monte_vegas.h, miser.c, Attic/gsl_miser.h, Makefile.am, Makefile.in, gsl_monte_miser.h: 188 Renamed public header files to follow convention correctly. 189 190 * vegas.c, vegas_test.c, miser.c, miser_test.c, plain.c, plain_test.c, Attic/gsl_miser.h, Attic/gsl_vegas.h, gsl_monte_plain.h: 191 All the integration functions now end with _integrate (except for 192 vegas1, vegas2 and vegas3 which are going away RSN). 193 194Tue Jul 21 21:54:33 1998 booth <booth@planck.pha.jhu.edu> 195 196 * vegas.c, vegas-print.c, Attic/gsl_vegas_print.h, Attic/gsl_vegas.h: 197 trivial stuff: eliminate compiler warnings, eliminate some unneeded variables, 198 change some types to make consistent with plain and miser. 199 200 * gsl_monte_plain.h, plain.c, plain_test.c: 201 Make "plain" conform to same style as miser and vegas. 202 203Fri Jul 17 02:23:40 1998 jungman <jungman@nnn.lanl.gov> 204 205 * Makefile.in: we're gonna make it... 206 207Thu Jul 16 16:23:45 1998 booth <booth@planck.pha.jhu.edu> 208 209 * vegas-print.c, vegas.c, Attic/gsl_vegas_print.h: 210 Have now completely eliminated all static variables. 211 212 * vegas_test.c, vegas.c, Attic/gsl_vegas.h, Attic/gsl_vegas_print.h, vegas-print.c: 213 Continuing to remove all static variables from vegas. 214 215Wed Jul 15 19:10:24 1998 booth <booth@planck.pha.jhu.edu> 216 217 * vegas.c, vegas_test.c, Attic/gsl_vegas.h: 218 Do the state-structure thing for vegas. Not finished, so far the only 219 real content is the rng structure. 220 221