Re: Numerical issues with sin()/cos()

[Sergei Steshenko]

________________________________
From: Gordon Haverland <address@hidden>
To: address@hidden 
Sent: Wednesday, March 28, 2018 1:53 AM
Subject: Re: Numerical issues with sin()/cos()



I ran across a fairly recent (2008) paper on some of the "dark arts" of
finding good approximations to mathematical functions.  It does in part
look at sin(x).

The paper has a preferred citation:

Florent De Dinechin, Christoph Lauter. Optimizing polynomials for
floating-point implementation. 12 pages. 2008. <ensl-00260563>

I believe the URL is:

https://hal-ens-lyon.archives-ouvertes.fr/ensl-00260563


Gord

-----------------------------------------------------

Since Intel 80387 (see https://en.wikipedia.org/wiki/X86_instruction_listings 
-> 
https://en.wikipedia.org/wiki/X86_instruction_listings#Added_with_80387 ) sine 
and cosine functions are part of the instruction set, so if one wants fast 
implementation, he's limited by what his x86 (if it's the case) provides.

--Sergei.