MathGroup Archive 1995

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

Search the Archive

Re: Re: FindMinimum and NIntegrate

  • Subject: [mg2864] Re: [mg2745] Re: FindMinimum and NIntegrate
  • From: sci20312 at leonis.nus.sg (CHAN MUN CHOONG)
  • Date: 30 Dec 1995 03:44:35 -0600
  • Approved: usenet@wri.com
  • Distribution: local
  • Newsgroups: wri.mathgroup
  • Organization: Wolfram Research, Inc.
  • Sender: daemon at wri.com

  Thank you for enlightening me as to what methods the two functions uses.
But after I read the book 'Numerical Recipes in C' about the Brent's method,
I am wondering how FindMinimum chooses a bracketing triplet of abscissas, 
how large is a parabolic step and what is the tolerance for this function.
In the book their tolerance is no smaller than the square root of the 
machine's floating-point precision.
  Hehe..... I can't ask anything on  the Options of function NIntegrate yet
because I have not read up on Gauss-Kronrod points yet. But I wonder how far 
is one step-size.
  I think it is about time someone writes a book to explain the inner 
workings of most of the functions? I have read most books on Mathematica 
and they all write about how we can apply it. Personally I think that 
built-in functions is a wonderful idea.....but I also believe that we 
should understand the inner workings should we try to use it. At least 
NAG or IMSL do publish their listings and explain how their functions 
actually evaluate parameters that someone put in.


  • Prev by Date: Re: Mma speed trials (Win'95)
  • Next by Date: Short form for FortranForm
  • Previous by thread: Re: Re: FindMinimum and NIntegrate
  • Next by thread: Files required to interface mma with C.