RE: [boost] C++ committee actions - More statistical math functions?

From: Paul A. Bristow (boost_at_[hidden])
Date: 2003-11-05 13:09:21

| -----Original Message-----
| From: boost-bounces_at_[hidden]
| [mailto:boost-bounces_at_[hidden]]On Behalf Of Beman Dawes
| Sent: Monday, November 03, 2003 7:38 PM
| To: boost_at_[hidden]
| Subject: [boost] C++ committee actions
|
| The C++ committee met last week. Actions which may be of interest to
| Boosters:

| * The C (yes, C, not C++) committee decided the previous week to include
| the Special Functions from the C++ Library TR as a C library. The intent is
| to maintain as close to 100% compatibility between the two Special Function
| libraries as is possible. Certain minor changes were made to the TR library
| to facilitate compatibility.

Is now the time to (re-)start discussing MORE 'special functions' like the ones
that I prepared (but did not submit) a proposal for 'math function essential for
statistics' ?

My previous proposal and a brief introduction attached.

One possible starter might be to approach Stephen Moshier, author of a popular C
library which does most of the numerical work, to get his permission to re-skin
his code - mainly renaming functions to conform to std conventions.  However, if
the C men are going to use it, I am unclear of the requirements to get
compatibility.

Walter Brown was very keen not to try to introduce too much in his choice of
functions (flawed IMHO) for fear of frightening the vendors who would frighten
the TR.  If we have at least a Boost free-issue library, neither group need have
fear, but vendors would be free to improve on Moshier's code if they can/wish.

Views on if and how to progress this?

Paul

Paul A Bristow, Prizet Farmhouse, Kendal, Cumbria, LA8 8AB  UK
[phone]   Mobile [phone]
mailto:pbristow_at_[hidden]


attached mail follows:

Following discussions with Walter Brown, who made a previous proposal for some
math functions (adding to the C99 functions proposed by P J Plauger),
I have prepared a draft of a new proposal to add a good number of other 'special
functions' which I believe have even greater practical utility (but perhaps more
difficulty in implementation).

The functions are essential for any statistical work assessing probabilities. I
believe this is essential to get useful information even from means & standard
deviations.

The difficult function is the incomplete beta, but is implicitly used by most of
the other functions.

The C99 math functions are also required.

The attached zipped HTML DRAFT document gives some background, references,
rationale, proposed functions and suggested names (which put clarity before
curtness, diverging from math greek letters to longer C++ Standard style names.

I have drawn heavily from an existing C implementation called Cephes by S
Moshier which might provide a C++ implementation if a Boost licence for a
repackaged C++ version could be obtained from the author.

The proposed functions are overloaded for float, double and long double
following the C99 to C++98 proposal, without using exceptions.  Although, this
is not the best C++ solution, I am now resigned that it is the best compromise
to maximize C and C++ compatibility.

Comments, corrections, criticism etc welcome.

Paul

Paul A Bristow, Prizet Farmhouse, Kendal, Cumbria, LA8 8AB  UK
[phone]   Mobile [phone]
Mobile mailto:pabristow_at_[hidden]
mailto:pbristow_at_[hidden]