From mboxrd@z Thu Jan  1 00:00:00 1970
X-Spam-Checker-Version: SpamAssassin 3.4.4 (2020-01-24) on polar.synack.me
X-Spam-Level: 
X-Spam-Status: No, score=-0.3 required=5.0 tests=BAYES_00,
	REPLYTO_WITHOUT_TO_CC autolearn=no autolearn_force=no version=3.4.4
X-Google-Thread: 103376,6d7a86bff9319841
X-Google-Attributes: gid103376,domainid0,public,usenet
X-Google-Language: ENGLISH,ASCII-7-bit
Path: 
 g2news1.google.com!news3.google.com!news4.google.com!feeder1-2.proxad.net!proxad.net!feeder2-2.proxad.net!newsfeed.arcor.de!newsspool3.arcor-online.net!news.arcor.de.POSTED!not-for-mail
From: "Dmitry A. Kazakov" <mailbox@dmitry-kazakov.de>
Subject: Re: Where I find Bessel function for Ada ?
Newsgroups: comp.lang.ada
User-Agent: 40tude_Dialog/2.0.15.1
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: 7bit
Reply-To: mailbox@dmitry-kazakov.de
Organization: cbb software GmbH
References: 
 <a8cfd61d-2b5e-4df9-a7a4-8a83f3709f04@s20g2000yqh.googlegroups.com>
  <ccf0cc26-f19d-4457-941e-de15ed0a38dd@j39g2000yqn.googlegroups.com>
 <ggjhp2$5ts$1@nntp.motzarella.org>
 <de0bec28-e6be-4ae7-b67d-b34a430bd4b1@f13g2000yqj.googlegroups.com>
 <111195bd-6d60-42e6-8ce1-bf7c27a2f665@u14g2000yqg.googlegroups.com>
Date: Wed, 26 Nov 2008 18:17:03 +0100
Message-ID: <5rukpfxv15it.17kah89egv5pw$.dlg@40tude.net>
NNTP-Posting-Date: 26 Nov 2008 18:17:06 CET
NNTP-Posting-Host: 8ee0eac2.newsspool2.arcor-online.net
X-Trace: 
 DXC=HH\iGk@b`G^Fm0Y?OE@2^XA9EHlD;3YcR4Fo<]lROoRQ^YC2XCjHcbY4XCjf8D;S`]DNcfSJ;bb[UIRnRBaCd<M^[Koa8449eBZE<?KI8]k0BQ
X-Complaints-To: usenet-abuse@arcor.de
Xref: g2news1.google.com comp.lang.ada:2816
Date: 2008-11-26T18:17:06+01:00
List-Id: <comp.lang.ada>

On Wed, 26 Nov 2008 08:12:21 -0800 (PST), Ludovic Brenta wrote:

> On Nov 26, 4:09 pm, Ken Thomas <k...@ecs.soton.ac.uk> wrote:

>> As the code for bessj0 is very short, translation to Ada is a small
>> exercise. On the other hand, the code seems to deal with only single
>> precision arithmetic so does not promise high accuracy.
> 
> I suppose this is easy to fix by making the Ada implementation
> generic. The user then instantiates it for any floating-point type
> with the required accuracy.

An ability to do so depends on the accuracy of the constants used in the
implementation.

A good-designed library should have the accuracy of the result within n *
T'Model_Epsilon or, better, within the range f(x)'Pred..f(x)'Succ. It is
non-trivial to do for a generic package. Short types will suffer accuracy
losses due to insufficient precision. Long precision types will due to
inaccuracy of the coefficients. Most likely in a generic body you will need
some "carrier" type of the precision higher than of the argument type. In
some cases T'Base can help. In others it cannot...

-- 
Regards,
Dmitry A. Kazakov
http://www.dmitry-kazakov.de