From: HL
Subject: Numerical Recipes site has Common Lisp code for scientific computing.
Date:
Message-ID: <867izs41dq.fsf@agora.my.domain>
Dear *
It might be of interest to some here a little something I just bumped
into: the famous Numerical Recipes book series website sells a CD-ROM
with Numerical Recipes code in Common Lisp:
http://www.numerical-recipes.com/cdrom-blurb.html
Buying the CD-ROM grants you a "one screen" license to use the code.
It costs less than US$ 100.00. Versions for UNIX, Mac and Windows are
available.
(I can't comment on anything more since it hasn't arrived yet).
Best regards.
HL
HL wrote:
> Dear *
>
> It might be of interest to some here a little something I just bumped
> into: the famous Numerical Recipes book series website sells a CD-ROM
> with Numerical Recipes code in Common Lisp:
>
> http://www.numerical-recipes.com/cdrom-blurb.html
This is exciting -- please post details when you get the CDROM. The
blurb seems to be confusing about the Lisp content (what is the
difference between the first edition and the second ed). Most
importantly, does the lisp source include everything in the book?
Thank you.
> Buying the CD-ROM grants you a "one screen" license to use the code.
> It costs less than US$ 100.00. Versions for UNIX, Mac and Windows are
> available.
>
> (I can't comment on anything more since it hasn't arrived yet).
>
> Best regards.
>
> HL
remixer wrote:
> HL wrote:
> > Dear *
> >
> > It might be of interest to some here a little something I just bumped
> > into: the famous Numerical Recipes book series website sells a CD-ROM
> > with Numerical Recipes code in Common Lisp:
>
> > Buying the CD-ROM grants you a "one screen" license to use the code.
> > It costs less than US$ 100.00. Versions for UNIX, Mac and Windows are
> > available.
>
If it's anything like the other books, the numerical recipes are rather
amateurish. I'd avoid them like the plague.
Licensing also is a PITA.
best,
-tony
Well, I was going to resist replying to this, but since we're dog
piling on this now, I'll add my 2 cents in. My boss has the cd and I
went over the CL code pretty carefully, especially in relation to
Matlisp and Femlisp. I benchmarked all 3 on SBCL and found that both
Matlisp and Femlisp were way faster for basic linear algebra than the
NR CL routines. Additionally, the NR CL functions didn't seem to me to
be as nicely "Lispy" as either of Matlisp or Femlisp.
I think ultimately that using Matlisp, Femlisp, or stealing stuff out
of Maxima is much more reasonable these days, as well as being much
nicer in the license area.
Glenn
On 25 Sep 2006 13:48:06 -0700, <·············@specastro.com> wrote:
>
> My boss has the cd and I
> went over the CL code pretty carefully, especially in relation to
> Matlisp and Femlisp. I benchmarked all 3 on SBCL and found that both
> Matlisp and Femlisp were way faster for basic linear algebra than the
> NR CL routines.
NR did produce the same answers as Femlisp and Matlisp ? Besides
speed, the other common criticism about NR is that it's wrong.
'way faster' is vague, 2x, 3x, 20x ?
--
Reply-To email is ignored.
--
Posted via a free Usenet account from http://www.teranews.com
On Mon, 25 Sep 2006 17:58:47 -0700, GP lisper <········@CloudDancer.com> wrote:
> On 25 Sep 2006 13:48:06 -0700, <·············@specastro.com> wrote:
>>
>> My boss has the cd and I
>> went over the CL code pretty carefully, especially in relation to
>> Matlisp and Femlisp. I benchmarked all 3 on SBCL and found that both
>> Matlisp and Femlisp were way faster for basic linear algebra than the
>> NR CL routines.
>
> NR did produce the same answers as Femlisp and Matlisp ? Besides
> speed, the other common criticism about NR is that it's wrong.
>
> 'way faster' is vague, 2x, 3x, 20x ?
Before beeing too excited about the code, take note of how it is
developed. Basically it is a automatic translation from parts of the
Fortran files from the 1st edition of the book.
I would not consider it as comparable to matlisp. It is a collection
of numerical functions you can use.
Yes, it is fine for a lot of simple to moderate stuff. Sometimes
you need to solve a least squares problem, or a Spearman
correlation, or find a zero, and you need to do it now, in Lisp,
without loading a huge Fortran library with a FFI that never
quite seems to work.
BTW, here is another NR Lisp on the web:
http://www.phys.uu.nl/DU/num_recipes/lisp.1ed/kdo/readme.htm
http://www.phys.uu.nl/DU/num_recipes/lisp.1ed/senac/readme.htm
I have no idea what the license is. They apparently have not been shut
down by the NR authors or their publisher, yet.
> Before beeing too excited about the code, take note of how it is
> developed. Basically it is a automatic translation from parts of the
> Fortran files from the 1st edition of the book.
> I would not consider it as comparable to matlisp. It is a collection
> of numerical functions you can use.