John Maddock
b652dc390c
Fix bug that effects the behaviour when the result is negative.
...
[SVN r43400]
2008-02-23 09:46:43 +00:00
John Maddock
5e22dad52e
Added non-central T to distributions.hpp.
...
Updated generic_mode to use additive stepping where appropriate.
Improved NC-Beta PDF calculation and termination conditions.
More or less got non-central T finished off: quantiles now work, PDF and mode almost correct.
Fixed typo in toms748_solve.hpp
Updated NC-T tests to include mode and quantile.
[SVN r43333]
2008-02-20 18:11:43 +00:00
John Maddock
3e6b00970c
Added some fixes to check for iterations >= the limit rather than ==.
...
Added extra common error handler.
Factored out some generic quantile code.
Killed some warnings in NC Beta.
Added initial non central T.
Fixed a few doc typos.
Added initial NC T tests.
[SVN r43324]
2008-02-19 16:52:02 +00:00
Paul A. Bristow
6dd0f0d42c
Corrected range to permit df == 2
...
and Added comment(-ed code) about mode with df < 2.
No reply to question about this on Wikipedia 16 Feb 08.
[SVN r43275]
2008-02-16 10:36:02 +00:00
John Maddock
d77728bffb
Ooops, fix comment.
...
[SVN r43223]
2008-02-11 17:30:28 +00:00
John Maddock
12df5b2419
Updated non central beta cdf to accept both x and 1-x on input: allows greater accuracy when using these routines in the non central F and T distributions.
...
Updated non central F to take advantage of this.
Added some non central distributions to the graph generator.
[SVN r43222]
2008-02-11 17:21:14 +00:00
John Maddock
8e6ce7d00c
Made the non-central chi-squared PDF more robust.
...
Set generic mode calculation to throw if the original guess yields a zero PDF.
Updated tests to match.
[SVN r43098]
2008-02-04 18:57:37 +00:00
John Maddock
698f280920
Added non central F distribution.
...
Tidied up non-central beta and Chi squared distribution by factoring out common generic-mode code.
Updated RR bindings.
[SVN r43075]
2008-02-03 17:17:32 +00:00
John Maddock
38e3301e09
More or less finished off the non central beta.
...
[SVN r43020]
2008-01-30 12:42:26 +00:00
John Maddock
b45ba0d428
Added initial commit of non-central beta distribution: CDF and quantile work OK, but nothing else!
...
[SVN r42999]
2008-01-28 18:29:37 +00:00
John Maddock
04a99ac565
Optimise the sums when we're going to be subtracting the result from 1.
...
[SVN r42998]
2008-01-28 18:28:16 +00:00
John Maddock
928b8d8a63
Added error handling to the rounding functions.
...
Added better error handling to the non-central chi squared, and updated the tests.
[SVN r42920]
2008-01-23 10:41:06 +00:00
John Maddock
bd2518ee9e
Add needed <iostream> include.
...
[SVN r42851]
2008-01-18 16:56:57 +00:00
John Maddock
c73de6a963
Finish off Noncentral Chi Squared docs: except for a graph.
...
Update tests to cope with other platforms.
Disabled a few tests that can never pass.
Added nccs to concept checks.
Added warning fixes and special case handling to implementation.
[SVN r42849]
2008-01-18 16:35:56 +00:00
John Maddock
8081dc49df
Commit of more or less complete non-central chi-square distribution: docs still to come.
...
[SVN r42820]
2008-01-16 19:02:45 +00:00
Andrew Sutton
ccb9b8c656
Added a modification to the extreme value distribution. This allows the specification
...
of a sign to determine the minimum or maximum case. Note that the original extreme value
distribution refers to the maximum case. This class will need some work. Discussion
leaned towards the use of an enumerated parameter rather than a real-typed sign
for the determination of min/max case.
[SVN r42762]
2008-01-14 13:54:49 +00:00
Andrew Sutton
8e24be2ffe
Added stubs for new distributions. These are neither fully tested nor documented and,
...
in some cases, incomplete. However, migrating them into the sandbox will hopefully
help motivate their completion.
[SVN r42760]
2008-01-14 13:36:23 +00:00
John Maddock
fe3db87cc4
Change concepts classes so they can be used when there is no long double support.
...
Added first cut of the non-central chi squared distribution.
Removed almost all occurrences of real_cast: replaced with calls to the truncation/rounding functions instead.
[SVN r42744]
2008-01-14 09:27:44 +00:00
John Maddock
2f7fadc594
Change Poisson PDF to use gamma_p_derivative as it avoids overflows and other nasties.
...
[SVN r42371]
2007-12-30 13:09:12 +00:00
John Maddock
db2ffd21c2
Added trunc, round and fmod family of functions.
...
[SVN r42244]
2007-12-22 19:01:37 +00:00
John Maddock
b27b39532b
Added expint docs.
...
Updated zeta and expint tests with hooks for other libraries.
Added #pragma once to header files.
[SVN r42140]
2007-12-18 18:32:50 +00:00
Paul A. Bristow
c327e0ac0f
Changed support range to avoid a discontinuity at zero.
...
[SVN r42124]
2007-12-17 15:58:23 +00:00
Paul A. Bristow
72e5db15f6
Changed support range to avoid a discontinuity at zero.
...
[SVN r42123]
2007-12-17 15:51:53 +00:00
Paul A. Bristow
99cf02892c
Changed support range to avoid a discontinuity at zero.
...
[SVN r42122]
2007-12-17 15:51:08 +00:00
John Maddock
6ae334402a
Fix error messages so they work with Boost.Format.
...
[SVN r42074]
2007-12-15 17:10:03 +00:00
John Maddock
fc5f875671
Fixes #1454 .
...
Previous fix was incomplete, so added new tests, confirmed problem report, and added fixes.
All the tests now compile cleanly with g++ -pedantic -Wall -Wno-long-long (checked with gcc-4.4.1 and 3.4.4).
[SVN r42012]
2007-12-13 18:45:58 +00:00
John Maddock
128f2adc14
Fixes #1454 .
...
[SVN r41966]
2007-12-11 16:37:22 +00:00
John Maddock
ab9017cf29
Added expint and zeta functions, plus tests.
...
Fixed ntl::RR iostream operators.
Updated minimax program with zeta and expint approximations.
[SVN r41958]
2007-12-10 18:35:56 +00:00
John Maddock
a9fa7b4924
Fix << iostream operator for the NTL bindings.
...
[SVN r41272]
2007-11-21 09:46:36 +00:00
John Maddock
e504edf582
aCC on PA-RISC emits hard errors if a numeric constant underflows: enable existing workaround for that compiler.
...
[SVN r41251]
2007-11-20 11:12:36 +00:00
John Maddock
38fc694025
Oops: added missing template specialisation argument.
...
[SVN r41218]
2007-11-19 10:02:16 +00:00
John Maddock
9a0b70f42b
Added missing template argument to specialisations.
...
[SVN r41199]
2007-11-18 12:23:37 +00:00
John Maddock
605c55e00e
Added workarounds for IBM xlc C++: the compiler can't resolve an overloaded function template to a function pointer unless the template arguments are explicitly provided.
...
Fixed some bugs uncovered by the above workaround: some forward declarations didn't match the actual definition!
Hopefully fixed remaining Sun compiler issues: mostly fixed by above fixes anyway.
[SVN r41142]
2007-11-16 11:30:43 +00:00
John Maddock
5c7b465759
Fix some compiler warnings when building with the Intel and aCC compilers.
...
[SVN r41103]
2007-11-15 09:52:26 +00:00
John Maddock
8c7cd40a78
Added WinCE workaround.
...
[SVN r41088]
2007-11-14 12:41:32 +00:00
John Maddock
2b1fac126f
Merged changes from the Sandbox to Trunk.
...
[SVN r41086]
2007-11-14 12:14:19 +00:00
John Maddock
e7db05cc1a
Merged changes from Trunk.
...
[SVN r41085]
2007-11-14 12:11:23 +00:00
John Maddock
850b746f1c
Ooops, disable long double overloads when there is no long double support (Borland fix).
...
[SVN r41062]
2007-11-13 16:15:27 +00:00
John Maddock
69031ebe9d
Added static assertion to catch use of long double function overloads when BOOST_MATH_NO_LONG_DOUBLE_MATH_FUNCTIONS is defined.
...
Made sure that if BOOST_MATH_NO_LONG_DOUBLE_MATH_FUNCTIONS is defined, then no long double overloads are actually instantiated by the tests.
[SVN r41007]
2007-11-11 14:05:50 +00:00
John Maddock
d37562d2ac
Added needed macro suppression to use of isfinite.
...
[SVN r40969]
2007-11-09 16:57:42 +00:00
John Maddock
af3f1e97f5
Fix calls to fpclassify so they don't trigger macro expansion when fpclassify is a macro.
...
[SVN r40879]
2007-11-07 10:09:50 +00:00
John Maddock
fdac2f83cd
Ooops, got the pp-logic wrong and disable long double support for all Unix platforms, instead of just Intel on Linux prior to version 10.
...
[SVN r40878]
2007-11-07 10:07:08 +00:00
John Maddock
8ff3199598
Disabled static assertions for compilers that don't support them.
...
[SVN r40843]
2007-11-06 16:58:46 +00:00
John Maddock
f9f9023daa
Disabled long double support for Intel on Linux prior to version 10.
...
[SVN r40780]
2007-11-05 12:07:47 +00:00
John Maddock
098513f585
Added workaround for Sunpro not compiling the traits classes.
...
Added workaround for platforms where numeric_limits<>::denorm_min() does not return a denorm!
Updated Solaris error rates.
[SVN r40727]
2007-11-03 18:32:54 +00:00
John Maddock
25a040f5c2
Hopefully, take care of the remaining Tru64 failures: adjust >= test to > in some of the tests, and fix one expected error level for the incomplete beta function.
...
[SVN r40719]
2007-11-03 12:36:55 +00:00
John Maddock
4dad82d26a
Added needed #include and using declaration.
...
[SVN r40693]
2007-11-02 18:26:47 +00:00
John Maddock
e875aee99a
Changed concept code so there are no null-references any more (the code breaks on EDG based compilers otherwise).
...
[SVN r40499]
2007-10-27 11:27:19 +00:00
Paul A. Bristow
f731fd35d0
Removed obselete reference to parameter arity. Previous deletion caused lots of errors. Hope this is now OK.
...
[SVN r40402]
2007-10-24 08:56:38 +00:00
Paul A. Bristow
8cf0e97cb7
Removed obselete reference to parameter arity
...
[SVN r40369]
2007-10-23 18:54:39 +00:00