[Boost-users] Ann: Statistical distributions / Mathematical Special Functions

From: John Maddock (john_at_[hidden])
Date: 2006-10-04 13:43:00

This package consists of 3 related components, and we're now seeking further 
feedback on these before the final push prior to a review:

Statistical Distributions
~~~~~~~~~~~~~~~~~~~~~~~~~

Following feedback from the previous preview,  we have re-organised
so that distributions are now C++ classes.

See the docs and especially the tutorial for illustrations on how
this will be useful in real life.

We would especially welcome feedback from those with experience of
using statistics, and indeed from professional statisticians who are
probably not Boosters.

Mathematical Special Functions
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

These are currently focused on those functions that are useful for
statistical applications:

Floating point classification.
Gamma/lgamma/digamma/beta/erf/erfc
Ratios of gamma functions
Factorials
The incomplete gamma and beta functions.
The inverses of the gamma, beta, and error functions.
Derivatives of the incomplete gamma and beta functions.

Compared to the previous release, most of the changes involve accuracy and
quality of implementation issues.  Error handling has now been revised to
provide a comprehensive package-wide error handling scheme complete with
meaningful error messages.

Toolkit
~~~~~~~

Provides tools to assist in the implementation of numerical methods:

Infinite series evaluation.
Continued fraction evaluation.
Polynomial and rational function evaluation.
Root finding with derivatives (Newton/Halley/Schroeder methods).
Root finding without derivatives.
Function minimisation.

Misc Tools
~~~~~~~~~~

These are strictly experimental, but are used in the development of
approximations: they are provided to hopefully encourage others to develop
further special function implementations.  Tools include:

Polynomial arithmetic and manipulation.
Minimax approximations (the Remez algorithm).
Helper functions used in testing etc.

Documentation and source downloads are available online at 
www.johnmaddock.co.uk/toolkit

Regards,

John Maddock.