MathGroup Archive 2008

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

Search the Archive

Re: Machine-precision Expfunction

  • To: mathgroup at smc.vnet.net
  • Subject: [mg93605] Re: Machine-precision Expfunction
  • From: Bill Rowe <readnews at sbcglobal.net>
  • Date: Mon, 17 Nov 2008 06:18:56 -0500 (EST)

On 11/16/08 at 7:02 AM, rschmied at gmail.com (Roman) wrote:

>answering my own question:

>MachinePrecisionExp = Compile[{{x, _Complex}}, Exp[x]]

>Any thoughts on this?

Your original post seemed to start with a number specified to
have lower precision than machine precision. All you are doing
above is converting the number to machine precision. That is the
result you get with your function above should be exactly the
equivalent of

Exp[x//N]

There is an obvious problem with converting a low precision
number to a high precision number. The value of the bits added
to make the low precision a higher precision number is not well defined.

But if the end goal is to obtain a machine precision number, why
specify the original number to be other than machine precision
in the first place?


  • Prev by Date: Re: Creating a Banner for presentation
  • Next by Date: Re: Machine-precision Exp[] function
  • Previous by thread: Re: Working with vectors symbolically
  • Next by thread: Re: Machine-precision Expfunction