Re: sin(x) for large x
From: Gerry Thomas (gfthomas_at_sympatico.ca)
Date: 05/18/04
- Next message: Gerry Thomas: "Re: Searching zeros of complex function"
- Previous message: Steven G. Kargl: "Re: Searching zeros of complex function"
- In reply to: glen herrmannsfeldt: "sin(x) for large x"
- Next in thread: bv: "fdlibm (was: sin(x) for large x)"
- Reply: bv: "fdlibm (was: sin(x) for large x)"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Date: Tue, 18 May 2004 00:03:03 -0400
"glen herrmannsfeldt" <gah@ugcs.caltech.edu> wrote in message
news:FRapc.12381$6f5.977509@attbi_s54...
> There is a discussion in comp.lang.c titled
> Sine code for ANSI C
>
> regarding the evaluation of the sin() function for large
> arguments. It might be more applicable in this
> group, so I thought I would ask here.
>
> What result do people expect from the sin() function for
> very large arguments, for example sin(1e100)?
>
> Hopefully the discussion won't get too nasty, but it
> does seem to be a contentious issue over there.
>
> -- glen
>
Glen is to be commended for bringing this issue here to c.l.f., the
newsgroup that lays claim to all things right in numerics (chortle,
chortle!).
Several years ago I suggested to the hawker Appleyard of
polyputthekettleonweallwantteaandbecchmarks that he include a test on
trignometrics and the exponential of large argument as a QA test of the
current Fortran offerings. Hell no, this bucks Herr Maine (whose
defensiveness of polyputthekettleonweallwantteaandbecchmarks Appleyard is
now clear, nodnod, winkwink, synomore, saynomore), uncovers vendor deciet
(sic deceit, WGF), Fortran standard technical committee bankruptcy, click
your heels!, stand erect!, and make the true Father-of-Fortran Yorkey
proud!, also read two chapters of Imperium, whip yourself but don't get
carried away, nodnod, winkwink, saynomore, saynomore, and get that wombat
out of Maine's eyes, he's got his eye on it!.
IMNHO, most practioneer of Fortran are the most naive suckers when it comes
to numerics: they know such noises as eps, tols, single, double, etc., but
they lack any deep understanding of fp numerics, particularly of its
pitfalls. The N(umerical) A(nalysis) fraternity have bent over backwards to
accommodate the uninitiated but the Fortran set would rather stick with
Numerical Recipes, Rodney Dangerfield's bible of NA, rather than face the
reality that Fortran 2++ (whatever) just doesn't give a fig about the
output of Fortran programs.
Today on c.l.f., we've had someone from .gov say more or less "well I
don't know, I'm all right Jack!" Lucky for him I'm not Jack! This is why
the Plugger's of this world can get buy with impunity on the 'the what the
idiots won't know won't hurt them' m.o. Evidently, this cavalier attitude
extends beyond Fortran. So, Mr Maine and your clients, you're exposed for
the charlatans that you are.
BTW, this from Cleve Moler: In Appendix 3 of 'Recent Progress of the Java
Grande Numerics Working Group, June 7, 1999':
1. "The exponential, sine and cosine functions in Microsoft's math library
are so inaccurate for large arguments that the library is unsuitable for
general-purpose use."
2. "(At the MathWorks, we use FDLIBM as the underlying library for MATLAB
on
all platforms. We did this primarily because we got tired of working around
the bugs, poor algorithms and inconsistent handling of exceptional cases
that we found in the libm's provided by other hardware and operating
systems
vendors.)"
So, for 1/2 decade the bimbo's of .gov have been going around telling
everyone to shutter up and pray with Fr Richard M., S.J. NOT, and his
coterie, for all is well and things turn out for the better and all that
polyanna crap. Be Jeasus.
-- E&OE Ciao, Gerry T. ______ "The old order changeth, yielding place to new, And God fulfils Himself in many ways, Lest one good custom should corrupt the world." -- Alfred Lord Tennyson, in Morte d'Arthur.
- Next message: Gerry Thomas: "Re: Searching zeros of complex function"
- Previous message: Steven G. Kargl: "Re: Searching zeros of complex function"
- In reply to: glen herrmannsfeldt: "sin(x) for large x"
- Next in thread: bv: "fdlibm (was: sin(x) for large x)"
- Reply: bv: "fdlibm (was: sin(x) for large x)"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Relevant Pages
|
