[Date Index] [Thread Index] [Author Index]

Re: Re: c code generation

• To: mathgroup at smc.vnet.net
• Subject: [mg4449] Re: [mg4362] Re: c code generation
• From: vschaer at uia.ua.ac.be (Dennis.Verschaeren)
• Date: Mon, 22 Jul 1996 04:40:52 -0400
• Sender: owner-wri-mathgroup at wolfram.com

It must be said that you can not use a floating-point implementation in 
C to verify Mathematica's exact arithmetic, unless you implement a 
library in C, which enables you to overload all operations in rational 
arithmetic (I do not mention symbolic computations).

You can try to use floating-point numbers in Mathematica to verify an 
algorithm in C, but then again, I must admit that is very unclear and
at least not transparant to the user, when Mathematica sends its 
floating-point numbers directly to the hardware (like single or double 
floating-point numbers in C), or handles them in software.

----------------------------Dennis Verschaeren--------------------------

University of Antwerp (Campus UIA)       desk            32  3 820 24 03
InfoProc Numeric Interest Group          secr            32  3 820 24 01
Dept. of Math. & Comp. Science           telecopier      32  3 820 24 21
Universiteitsplein 1                     e-mail:
B-2610 Wilrijk - Antwerp (Belgium)       Dennis.Verschaeren at uia.ua.ac.be

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




==== [MESSAGE SEPARATOR] ====