Mathematica 9 is now available
Services & Resources / Wolfram Forums
-----
 /
MathGroup Archive
2004
*January
*February
*March
*April
*May
*June
*July
*August
*September
*October
*November
*December
*Archive Index
*Ask about this page
*Print this page
*Give us feedback
*Sign up for the Wolfram Insider

MathGroup Archive 2004

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

Search the Archive

Re: long evaluation time

  • To: mathgroup at smc.vnet.net
  • Subject: [mg51221] Re: [mg51208] long evaluation time
  • From: Andrzej Kozlowski <akoz at mimuw.edu.pl>
  • Date: Sat, 9 Oct 2004 04:18:33 -0400 (EDT)
  • Sender: owner-wri-mathgroup at wolfram.com

What do you think it should not take long?

The answer is obviously 0 at {0,0,0}, which can be easily proved by 
most reasonably competent human beings. Nminimize can also do it:


NMinimize[{(x^2 + x)/(y^2 + 1) + (z^2 + z)/(x^2 + 1) +
     (y^2 + y)/(z^2 + 1), x > 0 && y > 0 && z > 0},
   {x, y, z}]


{0., {x -> 0., y -> 0., z -> 0.}}


though nothing NMinimize does constitutes a proof. But Minimize returns 
only exact results,and when it returns an answer then (barring of 
course bugs) it is the equivalent of a proof. To solve this kind of 
non-linear problem it relies on a famous algorithm, due to Collins, 
known as Cylindirical Algebraic Decomposition. However, the running 
time of this algorithm is doubly exponential in the number of 
variables, so you rarely can expect it to deal with more than 2 
variable problems. If you reduce the number of variables by 
substituting something for one of them,  you will get an answer in 
reasonable time, vis:


Minimize[{(x^2 + x)/(y^2 + 1) + (z^2 + z)/(x^2 + 1) +
      (y^2 + y)/(z^2 + 1), x > 0 && y > 0 && z > 0} /.
    z -> 1/2, {x, y}]

Minimize::"wksol":"Warning: There is no minimum in the region described 
by
the contraints; returning a result on the boundary.

{3/4, {y -> 0, x -> 0}}





On 8 Oct 2004, at 15:55, _n3g wrote:

> Hi everybody!
>
> Can anyone explain me, why it takes so long to evaluate
>
> Minimize[
>
> (x + x^2)/(1 + y^2) + (y + y^2)/(1 + z^2) + (z + z^2)/(1 + x^2),
>
> x > 0 && y > 0 && z > 0,
>
> {x,y,z}
>
> ]
>
> ?
>
> thx, _n3g
>
>
Andrzej Kozlowski
Chiba, Japan
http://www.akikoz.net/~andrzej/
http://www.mimuw.edu.pl/~akoz/


  • Prev by Date: Re: Re: Colors in ScatterPlot3D?
  • Next by Date: Re: no line PlotStyles with ParametricPlot3D ?
  • Previous by thread: Re: long evaluation time
  • Next by thread: Maximum likelihood estimator