Re: NIST Digital Library of Mathematical Functions

[Gerard Jungman <jungman at lanl dot gov>]

Been there, seen that. The NIST guys might be done in about five years,
but that is a pretty optimistic estimate, given how far they have
come in the last couple years. As for the web stuff and the e-handbook,
well, whatever. It seems to me people just want to be able to compute
the function values using a straightforward programming interface,
with a backend which is flexible with regard to precision and arithmetic
modes, and which is stable and correct. I define the project of
"Abramowitz+Stegun for the 21st Century" to be just that: all the
functions, no baloney. I don't consider a web-based form to be
a useful interface, if that is part of what they are hoping
to do. Besides, who needs plots of J_{5/2}(x) anyway?

They are good guys, and they are the people to do it, but they
are not going to do anything unless they get real funding to
fight a big-time war. Meanwhile, our guerilla actions continue.
Consequently, I am way ahead of them already, even though the
current specfunc code falls short of my goals of genericity,
coverage, and flexibility with respect to arithmetic implementation.
By the time I get started on the GSL++ version, which will be
highly generic and flexible with regard to the underlying arithmetic
implementation, the war against ignorance and proprietary numerical
code will be almost over.

By the way, if any of those guys are listening, I really do hope
they succeed. I just wish they had done it already so I wouldn't
have to do it myself.

--
G. Jungman


Mark Galassi wrote:
> 
> Amigos, this note might be interesting in relation to our special
> function library.
> 
> ------- Start of forwarded message -------
> Date: Mon, 4 Oct 1999 09:41:22 -0400 (EDT)
> Message-Id: <199910041341.JAA53402@corkie.nrl.navy.mil>
> From: Liam Healy <Liam.Healy@nrl.navy.mil>
> To: rosalia@lanl.gov
> Subject: NIST Digital Library of Mathematical Functions
> Reply-to: Liam Healy <Liam.Healy@nrl.navy.mil>
> 
> Mark,
> 
> Thank you for your interesting talk on Saturday about the GNU
> Scientific Library.  I mentioned to you the NIST project to
> update/replace Abramowitz and Stegun.  It is called "Digital Library
> of Mathematical Functions" (DLMF) and their web page is
> 
> http://math.nist.gov/DigitalMathLib/
> 
> The "ebook" is almost completely empty, the exception being the
> chapter on the Airy functions (Chapter 11).  They do say that they
> have secured funding.  In their "overview" section they say:
> 
> "Clearly, more emphasis should be placed on computational
> aspects. Ideally, static tables of numerical values should be replaced
> by software which can generate custom tables on demand. Similarly for
> graphical representations. Will technology and resources permit us to
> do this? What is the proper role for symbolic software? These kinds of
> questions will be considered with the assistance of qualified
> professionals from around the world."
> 
> It seems like your project would mesh very nicely with their goals.
> 
> Liam
> 
> ------- End of forwarded message -------