MathGroup Archive 2002

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

Search the Archive

Re: Precision....

  • To: mathgroup at
  • Subject: [mg33280] Re: [mg33266] Precision....
  • From: Tomas Garza <tgarza01 at>
  • Date: Wed, 13 Mar 2002 03:14:51 -0500 (EST)
  • References: <>
  • Sender: owner-wri-mathgroup at

I suggest you read the topics on numerical precision in The Book (1.1.2 and
3.1.4 - 3.1.5) or in the on-line help browser. The way Mathematica handles
precision is different from most conventional computer languages, so the
question about single- or double-precision is not relevant.

Tomas Garza
Mexico City
----- Original Message -----
From: "Balaji Srinivasan" <sbalaji at>
To: mathgroup at
Subject: [mg33280] [mg33266] Precision....

> Hi,
> I am a beginner to Mathematica, so I am sorry in case this is kind of
> trivial.
> I was trying to implement the PSLQ Integer Detection Algorithm in
> Mathematica 4.0. Was running into problems trying to implement something
> of the following nature:
> Have a bunch of matrices at an arbitrarily high precision level.
> Initially I wanted to convert them to say double precision
> (will N[x,16] do? or do I have to use SetAccuracy?)
> At this juncture I needed to carry out a bunch of steps at double
> precision.
> (will $MaxPrecision = 'something or the other' work here?)
> Next, back to a bunch of calculations at an arbitrarily high precision. Is
> there any easy and robust way of doing these steps.
> The double-precision steps are to save computation. Don't know if
> Mathematica would actually save time even if these are implemented..
> Thanks,
> Balaji

  • References:
  • Prev by Date: Re: Can anyone help about this inverse laplace transform problem?
  • Next by Date: Re: LU Decompsoition for singular matrices
  • Previous by thread: Precision....
  • Next by thread: Animation