Changelog for math-functions-0.3.4.4

Changes in 0.3.4.4

Support for future versions of GHC. Monoid and Monad instances now have canonical form

Fixed crash in invIncompleteBeta (#68) for some inputs initial approximation was computed incorrectly.

Dependency on vector-th-unbox is dropped. All instances are written by hand now.

New implementation for logGamma. Now it's precise within 2 ulps at full range including zeros at 1 and 2.
Bug with precision loss of invErf for parameters near zero is fixed.
Fixed loss of precision in bundled erf near zero. (Affect primarily GHCJS)
factorial for now uses lookup table
logFactorial is optimized a bit (less number of terms is used)
m_sqrt_eps constant added.
Module Numeric.SpecFunctions.Internal is exposed.
Many improvements for test suite. Tables of expected function values are generated with mpmath, error estimates improved. Test suite itself is migrated to tasty from test-framework.

Semigroup and Monoid instances added for data types from Numeric.Sum
API for finding roots of real functions reworked. 1) All algorithm parameters are now tweakable. 2) Functions for getting list of iterations added.
Foldable and Traversable instances for Root were added.

log1p and expm1 are simply reexported from GHC.Float. They're methods of Floating type class.
On windows expm1 is implemented in pure haskell for older GHCs.

logGamma now uses Lancsoz approximation and same as logGammaL. Old implementation of logGamma moved to Numeric.SpecFunctions.Extra.logGammaAS245.
Precision of logGamma for z<1 improved.
New much more precise implementation for incompleteGamma
Dependency on erf package dropped. erf and erfc just do direct calls to C.
Numeric.SpecFunctions.expm1 added
Numeric.SpecFunctions.log1pmx added.
logGammaCorrection exported in Numeric.SpecFunctions.Extra.
Module Numeric.Series added for working with infinite sequences, series summation and evaluation of continued fractions.
Module statistics: Statistics.Math.RootFinding copied to Numeric.RootFinding. Instances for binary and aeson dropped.
Root-finding using Newton-Raphson added
Numeric.MathFunctions.Comparison.ulpDelta added. It calculates signed distance between two doubles.
Other bug fixes.

Module statistics: Statistics.Function.Comparison moved to Numeric.MathFunctions.Comparison. Old implementation if within compared negative numbers incorrectly.
addUlps and ulpDistance added to Numeric.MathFunctions.Comparison.
relativeError and eqRelErr added to Numeric.MathFunctions.Comparison.
Precision of logFactorial is slightly improved.

logFactorial type is genberalized. It accepts any Integral type
Evaluation of polynomials using Horner's method where coefficients are store in lists added