Boost.Math 1.86 Standalone

What's Changed

• Don't use obsolete predef/checks/predef.jam B2 module. by @grafikrobot in #1122
• Improve reliability of codecov upload by @mborland in #1124
• Update geometric.hpp by @kenarab in #1127
• Improve Heuman Lambda precision: by @jzmaddock in #1125
• Change skew normal quantile to use bracket_and_solve_root. by @jzmaddock in #1123
• Correct ellint_1 logic for types with no infinity. by @jzmaddock in #1130
• Correct spurious underflow issues in non-central beta and t (also eff… by @jzmaddock in #1134
• Correct float_next(+INF) and float_prior(-INF) by @jzmaddock in #1135
• Add guards to user configurable macro definitions by @mborland in #1138
• Update CI by @mborland in #1136
• Fix rational conditional by @mborland in #1142
• Prevent passing denormals in calculation. by @jzmaddock in #1141
• Make boost::math::ccmath::fminf and boost::math::ccmath::fminl constexpr by @quxflux in #1146
• Update Bessel functions at infinity. by @jzmaddock in #1144
• Remove overload no longer required by multiprecision. by @jzmaddock in #1147
• Add randomize_va_space to drone by @mborland in #1149

New Contributors

• @grafikrobot made their first contribution in #1122
• @kenarab made their first contribution in #1127
• @quxflux made their first contribution in #1146

Full Changelog: boost-1.85.0...boost-1.86.0

Boost.Math 1.84 Standalone

What's Changed

• Use Dijkstra identities for 0(0th) Fibonacci convention by @ryanelandt in #1001
• update beta error messages and checks by @ryanelandt in #1003
• Update kolmogorov quantile newton ub to 1 by @ryanelandt in #1002
• only solve problems if have enough precision by @ryanelandt in #1004
• Update sanitizer images by @mborland in #1011
• fix unused abterm1 compile warnings by @ryanelandt in #1014
• MSVC 14_0 stdndards and has-include corrections by @ckormanyos in #1017
• Eliminate unused parameter by @ckormanyos in #1019
• test fix for multiprecision 562 by @ryanelandt in #1015
• Fix inverse_discrete_quantile for large guess by @Yuhta in #1007
• Fix issue #1023 by @mborland in #1024
• Improve ibeta power term calculation in the Sterling case by @jzmaddock in #1025
• Avoid spurious overflow and divide by zero in ibeta. by @jzmaddock in #1026
• Add clang-win to CI runs by @mborland in #1030
• k should be size_t by @AtariDreams in #1031
• Adjust recursion when using the Bessel function approximation with la… by @jzmaddock in #1036
• Correct non-central-t series convergence bug. by @jzmaddock in #1038
• ten should be a constexpr as a float by @AtariDreams in #1033
• Incorporate cast simplifications from PR880 by @jzmaddock in #1039
• has_denorm_now by @ryanelandt in #1029
• Fix clang-14 run in CI by @mborland in #1041
• Qualify log1p calls. by @jzmaddock in #1043
• Fix detection of ccmath usage in trunc by @mborland in #1044
• Update/improve ccmath configuration. by @jzmaddock in #1045
• Merge for 1.84 by @mborland in #1042
• Improve ccmath support detection. by @jzmaddock in #1046
• Merge ccmath updates to master by @mborland in #1047
• Update version history on master by @mborland in #1054

New Contributors

• @ryanelandt made their first contribution in #1001
• @Yuhta made their first contribution in #1007

Full Changelog: boost-1.83.0...boost-1.84.0

Boost.Math 1.83.0 Standalone

What's Changed

• Fix -Wmaybe-uninitialized in bessel jy series by @mborland in #967
• Fix rounding for non-representable numbers by @mborland in #968
• Add logcdf to distributions by @mborland in #946
• Fix fisher_f mode by @maitbayev in #976
• Fix for scipy issue 18302 by @mborland in #977
• Fix case where b is a negative integer and z is also negative. by @jzmaddock in #983
• Minor PCHIP cleanup by @NAThompson in #984
• Fix for scipy issues 18506 and 18511 by @mborland in #987
• Disable numeric_limits specialization for GCC-14 by @mborland in #993
• Numerical evaluation of Fourier transform of Daubechies scaling funct… by @NAThompson in #921
• Small github actions updates by @sdarwin in #995
• median_absolute_deviation bug fix for non-zero center by @rasmushenningsson in #997
• C++23 Floats by @mborland in #978

New Contributors

• @maitbayev made their first contribution in #976
• @rasmushenningsson made their first contribution in #997

Full Changelog: boost-1.82.0...boost-1.83.0

Boost.Math 1.82.0 Standalone

What's Changed

• Disable C6326: Potential comparison of a constant with another constant by @mborland in #857
• Improve the mention of Octonion in readme by @Bobini1 in #859
• Use package manager for cygwin runs by @mborland in #862
• Fix warning 4146 and 4244 in ccmath by @mborland in #858
• Use 64-bit RNG for types that are 64 bits by @AtariDreams in #865
• Assert when features which require C++17 are compiled for an earlier … by @jzmaddock in #856
• Junk removal by @AtariDreams in #863
• Remove thread limitation from GHA runner by @mborland in #874
• Fix typos by @bwignall in #876
• Fix for issue #871 by @mborland in #872
• Correct root finders in extreme cases. by @jzmaddock in #875
• Fix -Wimplicit-const-int-float-conversion by @mborland in #879
• Better error handling by @jzmaddock in #883
• More and better handling of ibeta small args. by @jzmaddock in #884
• Fix warnings in special functions by @mborland in #864
• Correct tests to make them Multiprecision safe. by @jzmaddock in #887
• Fix for binomial distribution quantile edge case by @mborland in #890
• Fix for Non-Central T Distribution FE_INVALID by @mborland in #892
• Remove using std::cbrt to fix build on platforms which don't support it. by @NAThompson in #888
• Consolidate windows testing by @mborland in #895
• Use cfloat instead of float.h by @AtariDreams in #899
• Fix tanh_sinh integration for edge case, by @jzmaddock in #901
• Make integrators const correct. by @jzmaddock in #903
• Use size_t in size_to_precision by @Flamefire in #907
• Require C++14 support by @mborland in #905
• Improve NAN handling in ccmath by @mborland in #912
• S390x testing by @jzmaddock in #908
• libraries.json: fix typo by @drewrisinger in #918
• Add deprecation notice to complex functions by @mborland in #917
• Remove most CI and just test arm64/drone. by @jzmaddock in #916
• Update tanh_sinh handling of boundaries. by @jzmaddock in #894
• Add additional architecture testing to drone by @mborland in #868
• Fix for Scipy Issue 16079. by @jzmaddock in #930
• Ignore warnings in distribution tests suite by @mborland in #933
• Downgrade condition numbers language requirement to C++14 by @mborland in #934
• Fix for scipy issue 17916 by @mborland in #939
• Enable GHA to kill in progress builds on new push by @mborland in #940
• Add Estrin's method for polynomial evaluation by @NAThompson in #932
• Fix hypergeometric for use with -fno-exceptions. by @jzmaddock in #950
• Fix for scipy issue 15101 by @mborland in #949
• Turn debug symbols off for msvc-14.0. by @jzmaddock in #951
• Add cmake install by @mborland in #954
• Replace int with long long in nct pdf and cdf by @mborland in #955
• Update cxxstd field by @mborland in #958
• Fix for issue #959 by @mborland in #960
• Fix ibeta_inv for very small p. by @jzmaddock in #962
• Improve error handling in non-central distributions. by @jzmaddock in #963

New Contributors

• @Bobini1 made their first contribution in #859
• @Flamefire made their first contribution in #907
• @drewrisinger made their first contribution in #918

Full Changelog: boost-1.81.0...boost-1.82.0

Boost.Math 1.81.0 Standalone

What's Changed

• Constexpr next by @mborland in #789
• constexpr signbit by @mborland in #793
• Cleanup Warnings by @mborland in #795
• Fix unsafe bool usage by @mborland in #794
• More warning fixes by @mborland in #796
• Fix right endpoint derivative estimate error in cardinal cubic b spline by @sjrowlinson in #797
• Fixes for -Wzero-as-null-pointer-constant by @mborland in #798
• Corrected paths for cubic and quartic roots headers in doc by @RL-S in #801
• Make sure ibeta_derivative triggers an underflow error when required. by @jzmaddock in #802
• Avoid overflow in intermediate bessel_ik computation by @mckib2 in #805
• Add BOOST_NOINLINE to standalone config and apply to erf_inv_initiali… by @jzmaddock in #809
• Fix for issue #810 by @mborland in #811
• Add #include tests for the hypergeometric functions. by @jzmaddock in #812
• Fix for issue #800 by @mborland in #813
• Update Readme by @mborland in #817
• Improve performance of the Complete Elliptic Integrals. by @jzmaddock in #815
• Improve ccmath signbit and copysign by @mborland in #819
• remove extra semicolon by @HDembinski in #824
• Modernize exception specification by @apolukhin in #822
• Correct 1F1[-n, -n, x] by @jzmaddock in #832
• Fix Issue 833 by @jzmaddock in #835
• Fix quartic roots when depressed cubic only has single real root by @NAThompson in #838
• BREAKING CHANGE: Correct definition of user_rounding_error, by @jzmaddock in #836
• Fix Clang-15 warnings by @AtariDreams in #831
• Specify seed in random tests. by @NAThompson in #839
• Fix bytes behavior by @AtariDreams in #828
• short-circuit overflow protection for M1 Mac clang by @mckib2 in #827
• Change looping over array to be unsigned by @AtariDreams in #830
• Update deprecated workflow by @mborland in #840
• Fix spelling of "Ubuntu" by @AtariDreams in #843
• Fix warnings in quadrature tests by @mborland in #844
• Fix for issue 845 by @mborland in #846
• Fix for scipy issue 17146 by @mborland in #847
• Drone minor config update by @sdarwin in #850
• DOC: Fix a typo in the 'Standalone Usage' section. by @WarrenWeckesser in #852
• Fix for issue #871 (#872) by @mborland in #897

New Contributors

• @sjrowlinson made their first contribution in #797
• @RL-S made their first contribution in #801
• @mckib2 made their first contribution in #805
• @HDembinski made their first contribution in #824
• @apolukhin made their first contribution in #822
• @sdarwin made their first contribution in #850

Full Changelog: boost-1.80.0...boost-1.81.0

Boost 1.80.0 Standalone

What's Changed

• Fix logic error in roots.hpp by @jzmaddock in #777
• Minor 1F1 bug fixes. by @jzmaddock in #778
• Add further color map example by @jzmaddock in #774
• Fix repeated name in example/policy_eg_1.cpp by @WarrenWeckesser in #780
• Fix for issue 773 by @mborland in #782
• Logpdf support by @mborland in #762
• Redeclaration of max_factorial value by @ckormanyos in #784
• Chatterjee Correlation Coefficient by @mborland in #770
• Constexpr fma by @mborland in #734
• Set BOOST_NO_EXCEPTIONS and BOOST_NO_RTTI for GCC and math in standal… by @jzmaddock in #786
• Bump minimum language standard to C++14 by @mborland in #788
• Move all GCC-5/6 testing to drone. by @jzmaddock in #790
• Try and consolidate more Github tests. by @jzmaddock in #791
• Replace non-UTF-8 copyright character by @jwakely in #792

New Contributors

• @WarrenWeckesser made their first contribution in #780
• @jwakely made their first contribution in #792

Full Changelog: boost-1.79.0...boost-1.80.0

Boost 1.79.0 Standalone

What's Changed

• Fix for beta link error by @mborland in #580
• Fix gini coefficient calculation. by @jzmaddock in #593
• Post-1.77 beta changes. by @jzmaddock in #661
• constexpr fpclassify by @mborland in #679
• I506 cstdfloat complex std by @ckormanyos in #507
• Provide graceful failure for Bernoulli code on platforms with no atomics. by @jzmaddock in #684
• constexpr frexp by @mborland in #686
• constexpr ldexp by @mborland in #688
• Add abs specializations to ccmath by @mborland in #690
• constexpr div by @mborland in #691
• Constexpr logb and ilogb by @mborland in #692
• constexpr scalbn and scalbln by @mborland in #693
• constexpr floor and ceil by @mborland in #694
• constexpr trunc by @mborland in #695
• Fix cstdfloat_math_incl_test on ppc64le with ieee128 by @mscastanho in #685
• constexpr modf by @mborland in #696
• constexpr round by @mborland in #697
• Expand the catmull-rom spline doc by @oleg-alexandrov in #698
• constexpr fmod by @mborland in #699
• constexpr remainder by @mborland in #700
• constexpr copysign by @mborland in #701
• Disallow integral return types. by @NAThompson in #704
• Correct test for cpp_dec_float division. by @jzmaddock in #712
• TanhSinh: Correct m_inital_row_length in double precision case. by @jzmaddock in #711
• Fix low hanging fruit test failures on s390x. by @jzmaddock in #687
• Cubic roots by @NAThompson in #703
• Update limits used in erf/erfc approximations. by @jzmaddock in #713
• Improve tanh_sinh. by @jzmaddock in #714
• Remove unicode from comments and loosen up error tolerance. by @NAThompson in #716
• Constexpr hypot by @mborland in #721
• Fix tests for Apple M1 by @mborland in #725
• Quartic roots. by @NAThompson in #718
• Fix and tests for issue #705 by @mborland in #731
• Constexpr fdim by @mborland in #730
• Constexpr fmin and fmax by @mborland in #732
• Fix for linker errors from multiprecision changes. by @jzmaddock in #735
• Constexpr comparison functions by @mborland in #736
• Fix for issue #737 by @mborland in #738
• Complete CI test with standalone mode by @mborland in #680
• Remove use of old-style casts from exp-sinh quadrature. by @NAThompson in #741
• Improving DE performance Part 2 by @jzmaddock in #722
• Remove old-style casts. by @NAThompson in #743
• Remove C-Style Casts by @mborland in #744
• Begin to systematically remove direct use of lexical_cast in favour o… by @jzmaddock in #742
• constexpr isunordered by @mborland in #745
• More C-Style Cast Removal by @mborland in #746
• Fix for #750 by @mborland in #751
• Rework thread-safety logic in bernoulli_details.hpp. by @jzmaddock in #753
• Add MSVC-2022 to CI Testing by @mborland in #720
• Color Maps by @mborland in #752
• Add GCC-5 to drone config by @mborland in #758
• Add class template argument deduction guides for distributions by @jamesfolberth in #756
• Do not use an unguarded Newton iterate to polish roots; it goes crazy… by @NAThompson in #759
• Fix for issue 757. by @NAThompson in #760
• Cygwin, FreeBSD and Long Doubles by @mborland in #729
• Fix for issue #764 by @mborland in #765
• Implement logaddexp by @mborland in #763
• small correction to F distribution docs by @grlee77 in #766

New Contributors

• @oleg-alexandrov made their first contribution in #698
• @jamesfolberth made their first contribution in #756
• @grlee77 made their first contribution in #766

Full Changelog: v1.77-standalone...boost-1.79.0

Boost 1.77 Standalone Release

After some hammering on the beta release by many community contributors, we are proud to release the Boost Math 1.77-standalone. This is nearly identical to the 1.77 release, with a few bugs fixed in standalone mode.

A few highlights:

• Improved mode of the non-central χ² algorithm from @vindex10
• Numerous fixes for the ARM platform from @StephanTLavavej
• constexpr versions of many cmath functions, like isnan, isfinite, from @mborland
• Bezier polynomials
• Bilinear uniform interpolation

Standalone Beta

This is the first standalone version of Boost.Math, corresponding roughly to Boost version 1.76. The main differences from the mainline version 1.76 are some diffs to make sure this works in isolation from the rest of Boost.