-
Notifications
You must be signed in to change notification settings - Fork 6
/
Copy pathChangeLog
114 lines (78 loc) · 3.74 KB
/
ChangeLog
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
2023-07-26 Nakata Maho [email protected]
* mplapack 2.0.1 integration
* catch up to recent compilers and autotools.
2015-03-20 Makoto Yamashita <[email protected]>
* Modification for the latest gcc compilers.
2009-10-05 Nakata Maho <[email protected]>
* mpack 0.5.1 integration.
2009-09-28 Nakata Maho <[email protected]>
* read sufficiently large precision (reported by Henry Cohn)
2009-02-12 Nakata Maho <[email protected]>
* Allow link against external mpack.
./configure --with-system-mpack
* Now sdpa_gmp is statically linked which is faster except for MacOSX [1].
http://developer.apple.com/qa/qa2001/qa1118.html [1]
2009-02-02 Nakata Maho <[email protected]>
* migrate to mpack for linear algebra part. Just take routines.
http://mplapack.sourceforge.net/
2008-08-04 Nakata Maho <[email protected]>
* Fix a bug in gmp_dsterf [1]
* remove mpf_abs_
Reported by "Hayato Waki" <[email protected]> [1]
2008-08-01 Nakata Maho <[email protected]>
* peep hole speedup for gmp_sign
* add mBSD header to sdpa_gmpalgebra.cpp
2008-07-31 Nakata Maho <[email protected]>
* better implementation of gmp_dlamchE; once it has calculated,
then saves the value.
* better implementation of gmp_dlamchS by using mp_limb_t.
2^31-1 for exponent for i386 machines. still it does not seem
to be correct. 2^122 seems to be the right number, but not
consistent with the manual [manual says:
2^-68719476768 to 2^68719476736; which are about 2^36-1 to 2^36-1]
* peep hole speedup for gmp_sign
Reported by "Hayato Waki" <[email protected]>
2008-06-17 Nakata Maho <[email protected]>
* Catch up to SDPA 7.1.1, and updated accordingly.
* Fix floating point error during the calculation.
This was due to passing 0.0 to double log10().
It happens when the "digits" of sdpa_io.cpp is very small
so that it becomes zero when converted to double from mpf_class.
* Correct implementation for quicksort of eigenvalues.
Reported by "Hayato Waki" <[email protected]>
2008-06-16 Nakata Maho <[email protected]>
* Fix DIMACS_ERROS -> DIMACS_ERRORS.
Reported by "Hayato Waki" <[email protected]>
2008-06-13 Nakata Maho <[email protected]>
* Use P_FORMAT instead of GMP_P_FORMAT. Accordance with the manual.
* Fix DIMACS print build and correctly output DIMACS error
to the result file [2].
* remove param.sdpa~
Reported by "Hayato Waki" <[email protected]> [1], [2]
2008-04-01 Nakata Maho
* Catch up to sdpa-7.1.0.
2008-02-09 Nakata Maho <[email protected]>
* 7.0.5 RC2
* Many clean ups.
2008-02-09 Fujisawa Katsuki
* 7.0.5 RC1
* sdpa_main.cpp KAPPA value has been changed to 1.2
-static double KAPPA = 1.0;
+static double KAPPA = 1.2;
* remove some warnings
2007-12-11 Nakata Maho <[email protected]>
* Catch up to 7.0.5 of double version.
2007-09-16 Kazuhide Nakata
* Modify stopping criteria and step length.
* Add SolveInfo::check (print various information of current point)
2007-06-29 Nakata Maho <[email protected]>
* Remove GPL.txt. COPYING is newer and better.
* Update to 7.0.2. Fixes of memory leaks from Kazuhide Nakata.
* better configure.in, Makefile.am.
2007-06-26 Nakata Maho <[email protected]>
* Remove sdpa_dpotrf.h, sdpa_dpotrf.cpp, sdpa_linear.h, sdpa_linear.cpp.
We do not need them since we use GNU Multi Precision.
2007-06-25 Nakata Maho <[email protected]>
* Rewrite for gmp version.
2007-06-01 Nakata Maho <[email protected]>
* Initial ./configure support.