Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]
Groups > comp.lang.python > #91345
| From | Dennis Lee Bieber <wlfraed@ix.netcom.com> |
|---|---|
| Subject | Re: a more precise distance algorithm |
| Date | 2015-05-27 23:04 -0400 |
| Organization | IISS Elusive Unicorn |
| References | <b2e66a94-7a89-4be9-bbf7-9434396cc178@googlegroups.com> <5563e453$0$12990$c3e8da3$5496439d@news.astraweb.com> <CADzuzGNggKfgBiq3RMtWk3GBF3ynbkjm2brdXj4cuUQwQ5yQkg@mail.gmail.com> |
| Newsgroups | comp.lang.python |
| Message-ID | <mailman.114.1432782300.5151.python-list@python.org> (permalink) |
On Wed, 27 May 2015 14:00:24 -0400, Brian Blais <bblais@gmail.com>
declaimed the following:
>
>where all of the methods deviate at the 13/14 decimal place.
>
Historically, double precision floating point is considered to have
14-15 significant /digits/ TOTAL.
Internally, IEEE floating point hardware tends to work with 80bit
floats -- so as long as the intermediate computations stay in the FPU for
the majority of the calculation one may get a more precise result when
downconverting to 64-bit doubles. But if the algorithm (or runtime) tends
to pull values out of the FPU and reloads them for the next stage, one
loses all the excess precision, and the resultant truncations may multiple
to give an overall result of less than 14 significant digits.
--
Wulfraed Dennis Lee Bieber AF6VN
wlfraed@ix.netcom.com HTTP://wlfraed.home.netcom.com/
Back to comp.lang.python | Previous | Next — Previous in thread | Next in thread | Find similar | Unroll thread
a more precise distance algorithm ravas <ravas@outlook.com> - 2015-05-25 12:21 -0700
Re: a more precise distance algorithm felix <felix@epepm.cupet.cu> - 2015-05-25 16:06 -0400
Re: a more precise distance algorithm Christian Gollwitzer <auriocus@gmx.de> - 2015-05-25 22:27 +0200
Re: a more precise distance algorithm ravas <ravas@outlook.com> - 2015-05-25 14:03 -0700
Re: a more precise distance algorithm Gary Herron <gary.herron@islandtraining.com> - 2015-05-25 13:20 -0700
Re: a more precise distance algorithm ravas <ravas@outlook.com> - 2015-05-25 14:05 -0700
Re: a more precise distance algorithm Steven D'Aprano <steve@pearwood.info> - 2015-05-26 13:11 +1000
Re: a more precise distance algorithm ravas <ravas@outlook.com> - 2015-05-25 21:13 -0700
Re: a more precise distance algorithm Gary Herron <gherron@digipen.edu> - 2015-05-25 22:09 -0700
Re: a more precise distance algorithm ravas <ravas@outlook.com> - 2015-05-25 22:49 -0700
Re: a more precise distance algorithm Christian Gollwitzer <auriocus@gmx.de> - 2015-05-26 07:33 +0200
Re: a more precise distance algorithm Brian Blais <bblais@gmail.com> - 2015-05-27 14:00 -0400
Re: a more precise distance algorithm Oscar Benjamin <oscar.j.benjamin@gmail.com> - 2015-05-27 23:03 +0100
Re: a more precise distance algorithm Dennis Lee Bieber <wlfraed@ix.netcom.com> - 2015-05-27 23:04 -0400
Re: a more precise distance algorithm Ian Kelly <ian.g.kelly@gmail.com> - 2015-05-25 22:42 -0600
Re: a more precise distance algorithm ravas <ravas@outlook.com> - 2015-05-25 21:59 -0700
Re: a more precise distance algorithm random832@fastmail.us - 2015-05-26 09:40 -0400
Re: a more precise distance algorithm random832@fastmail.us - 2015-05-26 09:51 -0400
Re: a more precise distance algorithm Robin Becker <robin@reportlab.com> - 2015-05-27 14:02 +0100
csiph-web