Re: Numerical precision problem
- To: mathgroup at smc.vnet.net
- Subject: [mg43104] Re: Numerical precision problem
- From: "Ersek, Ted R" <ErsekTR at navair.navy.mil>
- Date: Tue, 12 Aug 2003 04:43:14 -0400 (EDT)
- Sender: owner-wri-mathgroup at wolfram.com
Daniel Lichtblau suggested forcing the use of high precision arithmatic with
the following.
fhighprecision[r_,{t1_,t2_},{d1_,d2_},prec_]:= With[
{bigr=Rationalize[r,0], bigt1=Rationalize[t1,0],
bigt2=Rationalize[t2,0], bigd1=Rationlize[d1,0],
bigd2=Rationalize[d2,0]},
N[f[bigr,{bigt1,bigt2},{bigd1,bigd2}],prec]
]
---------------------------
I suggest that instead of
Rationalize[x, 0]
one use
SetPrecision[x, Infinity]
since SetPrecision is about 26 times faster using Mathematic 4.1.
In other cases you don't want to use SetPrecision because it typically
returns a rational number with many more digits in the numerator and
denominator. However the excessive digits in the rational approximation are
of no concern in this case.
--------------------
Regards,
Ted Ersek
Download Mathematica tips, tricks from
http://www.verbeia.com/mathematica/tips/Tricks.html