[Sollya-users] Fwd: information, features regarding remez and fplll

Sylvain CHEVILLARD sylvain.chevillard at inria.fr
Mon Jun 4 11:04:08 CEST 2018


[The lists server of Inria's forge has been experiencing attacks 
recently and the interface to allow messages from non members of the 
list is currently unavailable. I am forwarding this message manually.]

@Jon: thank you for your message. The appropriate list would rather be 
sollya-users at lists.gforge.inria.fr which I put in copy. Do not hesitate 
to subscribe to this list, if you wish. The traffic is usually low (less 
than a dozen of messages per year, I would say)

Best,

Sylvain Chevillard


-------- Forwarded Message --------
Subject: information, features regarding remez and fplll
Date: Wed, 30 May 2018 16:59:01 +1000
From: JonT <jtid4576 at uni.sydney.edu.au>
To: sollya-devl at lists.gforge.inria.fr.

Firstly - thanks for what looks like a great tool, I hope it will make 
my life easier.

I have a number of approximations to make, many of which I have 
explorerd in Mathematica and the results look promising. I now have 
several variations for which I'd like (semi) optimised floating point 
coefficients not just arbitrary precision coefficients.

It seems like sollya should help me find the optimised coefficients, but 
from my perusal of the online documentation, I dont quite see how.

I have three features I'd like to use that I cant see in the online 
documentation for sollya:

1) rational polynomial approximations, all the documentation suggests 
only simple polynomials

2) simple change-of-variable/transform

3) separate control of precision for different coefficients (Im NOT 
expecting automatic handling of complexities of variable precision of 
double-double)

It may be that all are present, but not directly referred to in the 
documentation.

I think a simple change of variable may be implemented by a list of 
terms like [| 1, h(x), h(x)^2, ... , h(x)^n |] but this will get tedious 
and messy with rational polynomials, and I'm not sure if the 
implementation caches the evaluation of h(x) in this scenario.

It may be that the best solution is not use remez in Sollya, but to 
import predetermined high precision coefficients and then use sollya 
(and indirectly fplll) to find (near) optimal floating point 
coefficients. But I also don't see documentation for this.

Or am I misinterpreting what sollya is intended to solve ?

thanking you in advance,

Jon Tidswell




More information about the Sollya-users mailing list