MathGroup Archive 1999

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

Search the Archive

Precision

  • To: mathgroup at smc.vnet.net
  • Subject: [mg17929] Precision
  • From: "Peltio" <peltDOT.ioNOS at PAMiolDOT.it>
  • Date: Sat, 5 Jun 1999 02:56:01 -0400
  • Organization: Peltio Inc.
  • Sender: owner-wri-mathgroup at wolfram.com

Not too long ago , in the Group appeared a message regarding a way to get
rid of near zero results due to round off at machine precision.
The function

f[s_,n_]:= BesselK[0,n(1-s)]-BesselK[0,n]BesselI[0,n(1-s)]/BesselI[0,n]

was not giving the expected 0 value for s=0. Somehow, using the Chop
function was not useful, even though it resulted in the right series.
I was wondering if the following method, that is based on a fast way to
rationalize previously suggested in this NG, could represent a viable
solution:

preciseF[s_,n_]:=f[s,SetPrecision[n,Infinity]]

Or is it too demanding from the point of view of the resulting performance?
I'm posting this since I did not see further answer to the original post,
and all my previous posts got lost due (I think) to my server's problems.

Regards,
Peltio
peltio AT usa DOT net







  • Prev by Date: WAVELET PACKET TRANSFORM
  • Next by Date: ReadListBinary -- Is this a bug?
  • Previous by thread: WAVELET PACKET TRANSFORM
  • Next by thread: ReadListBinary -- Is this a bug?