Services & Resources / Wolfram Forums
MathGroup Archive
*Archive Index
*Ask about this page
*Print this page
*Give us feedback
*Sign up for the Wolfram Insider

MathGroup Archive 1995

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

Search the Archive

FindMinimum and picewise linear functions

  • Subject: [mg2679] FindMinimum and picewise linear functions
  • From: sinan at (Sinan Karasu)
  • Date: Tue, 5 Dec 1995 02:26:41 -0500
  • Approved:
  • Distribution: local
  • Newsgroups: wri.mathgroup
  • Organization: University of Washington, Seattle

I define a piecewise linear function with If statements,
and then sample the function on a list. And then I try
to do an optimization on it (sum of squares...)
FindMinimum always complains about not being able
to find the gradient. Has anyone come up with a way 
to do this? 

 Let us say I define (0,1)->R^2 to be a triangle
Then I sample it on say {0.1,0.3,0.5,0.7,0.9} and then
I come up with another sampling, say {0.1,0.3,0.4,0.7,0.9}

Now if I do Sum[(triangle1samples[[i]]-triangle2samples[[i]])^2,{i,1,5}]
and then perturb {0.1,0.3,0.t+4,0.7,0.9} and do a minimization
of Sum over the second sample (i.e. t) FindMinimum complains
about not being able to take the gradient. I unserdtand why 
it does, the question is, can I circumvent this somehow????


 Redistribution by Microsoft Network is prohibited.

  • Prev by Date: Re: X-FrontEnd crashes upon certain keystrokes
  • Next by Date: Re: Smith Normal Form Algorithm
  • Previous by thread: Re: 3-D graphics to GIF
  • Next by thread: FindMinimum and picewise linear functions