Commits

Author Commit Message Labels Comments Date
Bryan O'Sullivan
incompleteBeta: check x for NaN (bug 8)
Bryan O'Sullivan
Merge
Bryan O'Sullivan
Added tag 0.1.1.1 for changeset bf64c2b8cd67
Bryan O'Sullivan
Bump version to 0.1.1.1
Tags
0.1.1.1
Bryan O'Sullivan
Add tests to dist tarball
Aleksey Khudyakov
Fix case in invIncomplete when for small /s/ and /x it returned NaN instead of 0. E.g. before patch > invIncompleteGamma 0.01 0.2 = NaN Tests sometimes give spurious failure
Aleksey Khudyakov
Add missing module to the test suite. Fixes #7 Note that tests are disable at the moment
Aleksey Khudyakov
If either of parameters to incompleteGamma is NaN return NaN instead of looping
Bryan O'Sullivan
Disable tests, since they fail right now
Aleksey Khudyakov
Switch to simpler approximation later. It improver precision for logGamma from 13 digits to 15 digits around 1e6 at no visible performance penalty
Aleksey Khudyakov
More regular benchmark for logGamma* functions
Aleksey Khudyakov
Use functions to accept more random input
Aleksey Khudyakov
Test that incompleteBeta is in range
Aleksey Khudyakov
Fix bug in incompleteGamma Approximation for case s>1000 was calculated wrongly.
Aleksey Khudyakov
Add test to check that incomplete gamma is always in range
Bryan O'Sullivan
Added tag 0.1.1.0 for changeset c9130ac5dbe0
Aleksey Khudyakov
Add module documentation
Tags
0.1.1.0
Aleksey Khudyakov
Add bd0 function from S.Math It moved to N.SpecFunctions.Extra module since I think it's rarely used and it would just pollute namespace in N.SpecFunctions
Aleksey Khudyakov
Add strilingError. it was omitted somehow
Aleksey Khudyakov
Move module S.Constants to math-functions too
Bryan O'Sullivan
Added tag 0.1.0.0 for changeset 38bbc9ffdc3c
Bryan O'Sullivan
Make .cabal file consistent with my others
Tags
0.1.0.0
Bryan O'Sullivan
Fix references
Bryan O'Sullivan
Add a README
Aleksey Khudyakov
Use simple definition of factorial. Since all integer numbers below ~1e15 are exactly representable multiplication will give accurate results Also just return infinity if input is greater then 170. Multiplying all numbers from 2 to 1e6 isn't most effificent way to calculate IEEE infinity
Aleksey Khudyakov
Add benchmark for factorial
Aleksey Khudyakov
Fix benchmark
Aleksey Khudyakov
Add benchmark
Aleksey Khudyakov
Add test suite
Aleksey Khudyakov
cabalize
  1. Prev
  2. Next