MathGroup Archive 2000

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

Search the Archive

Some more questions Part II: Mathematica

  • To: mathgroup at smc.vnet.net
  • Subject: [mg22188] Some more questions Part II: Mathematica
  • From: "Jordan Rosenthal" <jr at ece.gatech.edu>
  • Date: Thu, 17 Feb 2000 01:24:24 -0500 (EST)
  • Organization: Georgia Institute of Technology, Atlanta GA, USA
  • Sender: owner-wri-mathgroup at wolfram.com

Hi all,

This group has been so helpful I couldn't resist asking some more questions.
I've split them up into two messages, one for my general questions and the
other strictly Mathematica.

These are the Mathematica questions:

1)  Consider the following expression:

   intQ = FullSimplify[Integrate[Simplify[(1 - Cos[m*w])*(1 - Cos[n*w])],
{w, 0, wp}]]

  Out =
   wp +
       ((m*Cos[n*wp])/(m^2 - n^2) - 1/m)*Sin[m*wp] + ((n*Cos[m*wp])/(n^2 -
m^2) - 1/n)*Sin[n*wp]

In the output there are some common terms which I wanted to make obvious.
so I did the following:

   intQ /. (n_ Cos[m_ wp]/(n_^2 - m_^2) - 1/n_)Sin[n_ wp] -> g[n, m]

   Out =
      wp + g[m, n] + g[n, m]

Now my question.  I would like to be able to define g[n_,m_] according to
the replacement rule I just used above.  Easy enough, but how can I register
this fact with Simplify so that it uses it during the simplification
process?  I assume I need to define it with an upvalue in some way, but I
get protection complaints and am hesitant about using unprotect before
finding out if I'm on the right track.  So am I?

2) When converting formulas to other languages (say C for example), it would
be nice to match similar terms as I did above manually; this would make it
easy to write modular functions that can be reused.  Is there a
ComplexityFunction anyone can think of which, when given to
Simplify/FullSimplify, will return a formula with the number of matching
terms maximized (or number of dissimilar terms minimized)?  It might have an
interesting application for compilers.  (Is this ill-defined?)

Thanks,

Jordan





  • Prev by Date: Re: Re: Simple Problem (I suppose...)
  • Next by Date: Re: sum of recursive fn: solving for n
  • Previous by thread: Global precision and accuracy.
  • Next by thread: Integrating Probability Surface_HOW?