MathGroup Archive 2000

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

Search the Archive

Re: Compiled

  • To: mathgroup at smc.vnet.net
  • Subject: [mg23853] Re: Compiled
  • From: Rob Knapp <rknapp at wolfram.com>
  • Date: Mon, 12 Jun 2000 01:17:45 -0400 (EDT)
  • Organization: Wolfram Research, Inc.
  • References: <8hstqk$dgu@smc.vnet.net>
  • Sender: owner-wri-mathgroup at wolfram.com

Bernd Brandt wrote:
> 
> Dear members,
> 
> I would like to ask something about Compile:
> Should Compiled -> False when using WorkingPrecison > 16
> 
> The manual is not clear about this, and it seems not to matter anyway.
> Maybe Compiled is automatically set to False when WorkingPrecison > 16?
> 
> However at Compiled this is stated:
> * You should set Compiled -> False if you need to use highprecision
> numbers.
> 
> At WorkingPrecsion:
> NIntegrate[1/x, {x, 1, 2},  WorkingPrecision -> 40,  Compiled -> False]
> 
> In 3.9.3 however, the same example without Compiled ->False:
> NIntegrate[1/x, {x, 1, 2}, WorkingPrecision -> 40]
> 

The Compiled option is ignored when doing computations with precision
higher than machine precision becuase the CompiledFunction objects in
Mathematica currently only deal with machine precision numbers.

Rob Knapp
Wolfram Research


  • Prev by Date: Re: Complex Numbers
  • Next by Date: Re: Re: joining lists.
  • Previous by thread: Compiled
  • Next by thread: ContourPlot doesn't work properly?