From: kst@thomsoft.com (Keith Thompson)
Subject: Re: Ada function sqrt(x)
Date: 1996/09/26
Date: 1996-09-26T00:00:00+00:00 [thread overview]
Message-ID: <DyC78x.Kt6@thomsoft.com> (raw)
In-Reply-To: R.dewar.843335145@schonberg
In <R.dewar.843335145@schonberg> dewar@cs.nyu.edu (Robert Dewar) writes:
[...]
> In fact, the square root of 9 must be exactly 3.
I can't find an RM reference for this requirement.
For implementations that don't support the optional Numerics Annex
(Annex G), the accuracy requirements on the mathematical functions
are nearly nonexistent. Even Annex G only requires a relative error
bound of 2.0*EF.Float_Type'Model_Epsilon (where EF.Float_Type is the
floating-point type in question), which still allows Sqrt(9.0) to be
slightly above or below 3.0.
Of course a decent implementation of Sqrt *should* return exactly 3.0
for Sqrt(9.0).
--
Keith Thompson (The_Other_Keith) kst@thomsoft.com <*>
TeleSoft^H^H^H^H^H^H^H^H Alsys^H^H^H^H^H Thomson Software Products
10251 Vista Sorrento Parkway, Suite 300, San Diego, CA, USA, 92121-2706
FIJAGDWOL
next prev parent reply other threads:[~1996-09-26 0:00 UTC|newest]
Thread overview: 9+ messages / expand[flat|nested] mbox.gz Atom feed top
[not found] <01bba50c$7c1a2760$dc014dc6@MountainNet>
1996-09-18 0:00 ` Ada function sqrt(x) Pascal Obry
1996-09-20 0:00 ` Keith Thompson
1996-09-20 0:00 ` Roderick Chapman
1996-09-21 0:00 ` Robert Dewar
1996-09-21 0:00 ` Robert Dewar
1996-09-26 0:00 ` Keith Thompson [this message]
1996-09-22 0:00 ` Matthew Heaney
1996-09-22 0:00 ` Matthew Heaney
1996-09-24 0:00 W. Wesley Groleau (Wes)
replies disabled
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox