MathGroup Archive 2011

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

Search the Archive

Algebraic substitution with PolynomialReduce

  • To: mathgroup at
  • Subject: [mg117078] Algebraic substitution with PolynomialReduce
  • From: Guido Walter Pettinari <coccoinomane at>
  • Date: Wed, 9 Mar 2011 06:59:18 -0500 (EST)

Dear group,

in order to perform algebraic substitutions in expressions, I usually
rely on replacementFunction, that is a very nice wrapper to
PolynomialReduce made by Daniel Lichtblau and Andrzej Kozlowski (see
links below).

However, I realized that replacementFunction is not enough when
dealing with multiple substitutions.  Take this expression (which
itself is a sub-expression of a more complicated one in my code):

terms = (2 kx1^2 kx2^2)/3 - (kx2^2 ky1^2)/3 + 2 kx1 kx2 ky1 ky2 - (
 kx1^2 ky2^2)/3 + (2 ky1^2 ky2^2)/3 - (kx2^2 kz1^2)/3 - (
 ky2^2 kz1^2)/3 + 2 kx1 kx2 kz1 kz2 + 2 ky1 ky2 kz1 kz2 - (
 kx1^2 kz2^2)/3 - (ky1^2 kz2^2)/3 + (2 kz1^2 kz2^2)/3

If I define  k1 = {kx1, ky1, kz1}  and k2 = {kx2, ky2, kz2},  then the
above is equal to the following combination of scalar products:

-(1/3 k1.k1 k2.k2 - (k1.k2)^2)

In order to explicitly reproduce the equality (and thus simplify the
equation), I tried to apply replacementFunction several times, with no
success.  So I turned to PolynomialReduce.  However, the following:

polylist = { k1.k1 k2.k2, (k1.k2)^2 };
PolynomialReduce[ terms, polylist, Join[k1, k2] ]  // Last

did not return a zero remainder.

By reading old messages in the mailing list, I see that this may be
due to the ordering of the variables in the third argument, i.e.
Join[ k1,k2 ].  Using  GroebnerBasis, I managed to prove that 'terms'
can be expressed in terms of the polynomials in 'polylist'.  In fact:

gb = GroebnerBasis[polylist, Join[k1, k2] ];
PolynomialReduce[terms, gb, Join[k1, k2] ] // Last

yields zero.

The question is:  how can I obtain the coefficients of my expression
('term') with respect to my polynomials ('polylist'),  given that I
already have the coefficients of 'terms' with respect to the Groebner
basis of 'polylist'?

Thank you very much for your consideration!

Best wishes,

Guido W. Pettinari


  • Prev by Date: Re: Please add me to this group
  • Next by Date: Re: evaluation-- one or many levels, your thoughts?
  • Previous by thread: how to resize the plot of two ellipses
  • Next by thread: Re: Algebraic substitution with PolynomialReduce