MathGroup Archive 1992

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

Search the Archive

FindRoot problem fixed

  • To: mathgroup at yoda.physics.unc.edu
  • Subject: FindRoot problem fixed
  • From: deutsch at bu-pub.bu.edu
  • Date: Wed, 16 Sep 92 11:22:35 -0400

Hello Mathgroup,

     Robby Villegas of WRI sent me mail this morning which fixed a
problem that one of our users here at Boston University was having
with FindRoot. Robby instructed me to set  Compile -> False in the
FindRoot argument list. This not only fixed the problem but reduced the
computation time from 33 secs to 10 sec. It is such a bizarre bug that
I wanted to mention it to the mathgroup. I have enclosed my original
question and the response from WRI.


David Deutsch
Boston University
Information Technology



-----------------------------------------------------------------------

Hello,

      I provide Mathematica support for users here at Boston
University. Recently a user has shown me a problem with FindRoot.

Using Timing[s]  where

s := N[ FindRoot[{d1v==0,d2v==0},{p2,p20},{m3,m30}] ,15]

and  d1v := a function of p2 and m3
     d2v := a function of p2 and m3


we observed that the time for solution was about 33 seconds which was
reasonable given the complexity of d1v and d2v. However, when we
repeated the same calculation we observed that the time more than
doubled. Further attempts showed that the time continued to increase
by a factor of 2. Are there any known problems with FindRoot that people
are aware of? Does anyone have any suggestions as to what could be
causing a problem of this type.
   

Thank you,

David Deutsch
Information Technology
Boston University
deutsch at bu-it.bu.edu

=====================================================================

Hi David,

This is a known bug in the compiler which crops up in some cases.
You can solve the problem by turning off the Compiled option for
FindRoot:

FindRoot[your arguments, Compiled->False]

I have forwarded the particular problem to the developer of Compiled.
This problem report is indexed as 92i.336, so if you wish to refer to
it in the future, please use this index.

Yours in truth,

Robby Villegas
Technical Support






  • Prev by Date: How does Mma work on a Mac Powerbook 170?
  • Next by Date: Has someone a Simplify that really simplifies?
  • Previous by thread: Re: How does Mma work on a Mac Powerbook 170?
  • Next by thread: Has someone a Simplify that really simplifies?