boost/math
2
0
Fork 0
mirror of https://github.com/boostorg/math.git synced 2026-01-27 19:12:08 +00:00
Code Issues Projects Releases Wiki Activity
6,645 Commits 37 Branches 125 Tags
c4e15ddda0e8b79e9bd46e7fac8dcbf2aca3e131
Commit Graph

3180 Commits

Author SHA1 Message Date
Matt Borland
c4e15ddda0 Fix index for MSVC 14.0 2023-12-22 06:06:57 -05:00
Matt Borland
8d575067e9 Rename bessel components 2023-12-22 05:17:55 -05:00
Matt Borland
a8ee8eb5d8 Protect dependencies with macros 2023-12-22 05:15:38 -05:00
Matt Borland
de67ad3207 Rename our components 2023-12-22 05:09:32 -05:00
Matt Borland
bd49c8311a Improve logic and shield other boost libraries in non-standalone mode 2023-12-21 08:33:23 -05:00
Matt Borland
e24e884562 Undef I after pushing onto the stack 2023-12-21 07:48:13 -05:00
Matt Borland
23d54b385b Fix for msvc defining _Complex_I in complex.h 2023-12-21 07:20:24 -05:00
Matt Borland
b3a21bcbcb Fix cstdfloat numeric limits detection macro for libstdc++14 2023-12-13 08:43:43 +01:00
Nick
e0c17f6856 Improve accuracy of quartic_roots (#1056) 2023-11-27 17:52:56 +01:00
Matt Borland
2dd018b8c7 Revert "Update version history." 
This reverts commit d938d3e836.
 2023-11-18 08:30:17 +01:00
jzmaddock
d938d3e836 Update version history. 
[CI SKIP]
 2023-11-17 18:25:02 +00:00
Matt Borland
0c5b957300 Merge pull request #1051 from boostorg/1048 
Fix for issue 1048
 2023-11-17 10:59:10 +01:00
Matt Borland
fe377373e4 Use identity function in floor if val >= 1/epsilon 2023-11-16 11:26:41 +01:00
Matt Borland
d18964d5aa Check update of xterm for denom of 0 2023-11-16 09:38:11 +01:00
jzmaddock
90c509398b Improve error detection further. 2023-11-06 09:40:26 +00:00
jzmaddock
d7048ecfa7 Improve ccmath support detection. 
See discussion in https://github.com/boostorg/math/pull/1045.
 2023-11-05 19:32:33 +00:00
jzmaddock
21230d698e Correct header include order for standalone mode. 2023-11-03 18:46:23 +00:00
jzmaddock
6d6b0a8c82 Correct new headers include guards. 2023-11-03 18:37:33 +00:00
jzmaddock
7f39320b8d Update/improve ccmath configuration. 2023-11-03 18:32:21 +00:00
Matt Borland
d298a15b1d Reorder macro and add missing include 2023-11-03 08:27:30 +01:00
jzmaddock
ae6571af58 Qualify log1p calls. 
This fixes ambiguous overloads in the multiprecision case between boost::multiprecision and boost::math versions of log1p.
See https://github.com/boostorg/multiprecision/pull/568#issuecomment-1790195318
 2023-11-02 12:14:06 +00:00
ryanelandt
2c74b149b0 has_denorm_now (#1029) 
* initial commit

* remove == std::denorm_present

* remove extra ) in ccmath/next
 2023-10-30 09:37:29 +01:00
jzmaddock
c6f4a88809 Incorporate cast simplifications from https://github.com/boostorg/math/pull/880 
Add tests and make sure everything is covered, and fix resulting errors.
 2023-10-17 17:09:44 +01:00
Rose
0e3b29f389 ten should be a constexpr as a float 
Its int counterpart is a constexpr, and the ten variable never changes, so for consistency, we should make float_type ten constexpr as welll.
 2023-10-15 18:54:03 -04:00
jzmaddock
eaf876c81e Correct non-central-t series convergence bug. 
Fixes https://github.com/boostorg/math/issues/1035.
See also https://github.com/scipy/scipy/issues/19348.
Accuracy in left tail is still poor, and the reflection formula appears to be to blame as it's use causes the series to cancel out the first term, but it appears we have no real choice in the matter here.  At least we do now get a few digits correct.
 2023-10-13 18:45:03 +01:00
jzmaddock
57b09f4919 More test cases and corrections. 2023-10-12 17:06:37 +01:00
jzmaddock
2dbd7698cc Adjust recursion when using the Bessel function approximation with large z. 
Avoids hitting tgamma on a negative integer, also improves accuracy when b-a-0.5 is close to a negative integer.
Fixes https://github.com/boostorg/math/issues/1034.
 2023-10-11 11:45:33 +01:00
AtariDreams
1722ef9261 k should be size_t (#1031) 2023-09-30 18:51:27 -07:00
jzmaddock
2d9d202ff1 Fix conceptual failures. 2023-09-01 17:21:25 +01:00
jzmaddock
8ff6bc114e Avoid spurious overflow and divide by zero in ibeta. 
Add tests extracted from SciPy bug report.
Fixes https://github.com/boostorg/math/issues/1006.
See also https://github.com/boostorg/math/pull/1009.
 2023-09-01 15:48:43 +01:00
jzmaddock
469bff7e17 Improve method selection logic in ibeta_power_terms 
(Sterling version)
 2023-08-31 12:36:19 +01:00
jzmaddock
0c4fc70036 Merge branch 'develop' into pr1007 
Fixes Conflicts:
	test/test_binomial.cpp
 2023-08-30 11:53:29 +01:00
jzmaddock
d967bf73d0 Improve accuracy in ibeta power terms usig Sterling's approximation (the non Lanczos case). 
Completes work started here: https://github.com/boostorg/math/pull/1007
 2023-08-30 11:50:28 +01:00
Matt Borland
0d941f0d07 Fix issue #1023 (#1024) 2023-08-28 13:10:58 -04:00
Jimmy Lu
74bb4bccf2 Fix inverse_discrete_quantile for large guess (#1007) 
If `guess` passed to `inverse_discrete_quantile` cannot be represented
as floating point number, it is possible that `guess + 1` or `guess - 1`
does not change the value at all and we are stuck in infinite loop
inside `round_to_floor` or `round_to_ceil`.  Fix this by
increase/decrease more than 1 in these cases.

Example code to reproduce this:
```c++
boost::math::binomial_distribution<> dist(9079765771874083840, 0.561815);
boost::math::quantile(dist, 0.0365346);
```
 2023-08-24 08:31:37 -04:00
Christopher Kormanyos
7b37e2378b Eliminate unused parameter 2023-08-22 19:35:53 +02:00
Christopher Kormanyos
4c964bb2ec Agree on PP syntax for <stdfloat> query 2023-08-22 15:07:22 +02:00
Christopher Kormanyos
320d6f8ab7 Use elementary yet more-obscure PP checks 2023-08-22 11:28:28 +02:00
Christopher Kormanyos
56b71a77c0 MSVC 14_0 stdndards and has-include corrections 2023-08-22 11:09:43 +02:00
John Maddock
f7e94ac22c Merge branch 'fix-inverse_discrete_quantile' of https://github.com/Yuhta/boost-math into pr1007 2023-08-19 12:37:40 +01:00
Ryan
4508717de6 fix unused abterm1 compile warnings 2023-08-18 16:20:15 -04:00
Jimmy Lu
ee80892282 Fix inverse_discrete_quantile for large guess 
If `guess` passed to `inverse_discrete_quantile` cannot be represented
as floating point number, it is possible that `guess + 1` or `guess - 1`
does not change the value at all and we are stuck in infinite loop
inside `round_to_floor` or `round_to_ceil`.  Fix this by
increase/decrease more than 1 in these cases.

Example code to reproduce this:
```c++
boost::math::binomial_distribution<> dist(9079765771874083840, 0.561815);
boost::math::quantile(dist, 0.0365346);
```
 2023-08-05 12:53:30 -04:00
ryanelandt
50ef83a47b only solve problems if have enough precision (#1004) 2023-07-31 10:36:29 -04:00
ryanelandt
c62682b1a6 Update kolmogorov quantile newton ub to 1 (#1002) 2023-07-31 10:19:42 -04:00
ryanelandt
fb0af63f2a update beta error messages and checks (#1003) 2023-07-31 09:54:33 -04:00
ryanelandt
281f491bd0 use identities for 0(0th) fib convention (#1001) 2023-07-25 09:55:24 -07:00
Matt Borland
8bb0d164f2 Merge pull request #978 from boostorg/cpp23-float 
C++23 Floats
 2023-06-28 08:02:25 +02:00
Matt Borland
851b357615 Fix pessimization on unaffected platforms 2023-06-27 16:03:43 +02:00
jzmaddock
0a014fd5ca Fix for expression template use in chebyshev.hpp. 2023-06-21 08:56:02 +01:00
Matt Borland
b57749d9c3 Fix casting errors 2023-06-20 12:01:47 +02:00