A method for finding recurrence relations for hypergeometric polynomials directly from the series expansions of the polynomials.
The method is effective and easily implemented, but usually slower than Zeilberger's
algorithm. Given a sum 
, the method operates by finding a recurrence
of the form
 |
by proceeding as follows (Petkovšek et al. 1996, p. 59):
1. Fix trial values of 
and 
.
2. Assume a recurrence formula of the above form where 
are to be solved for.
3. Divide each term of the assumed recurrence by 
and reduce every ratio 
by simplifying the ratios of its constituent
factorials so that only rational functions in
and 
remain.
4. Put the resulting expression over a common denominator, then collect the numerator as a polynomial in 
.
5. Solve the system of linear equations that results after setting the coefficients of each power of 
in the numerator to 0 for the unknown coefficients
.
6. If no solution results, start again with larger 
or 
.
Under suitable hypotheses, a "fundamental theorem" (Verbaten 1974, Wilf and Zeilberger 1992, Petkovšek et al. 1996) guarantees that this algorithm
always succeeds for large enough 
and 
(which can be estimated in advance). The theorem also generalizes
to multivariate sums and to 
- and multi-
-sums (Wilf and Zeilberger 1992, Petkovšek et al.
1996).
") Multisum/Integral Identities." Invent. Math. 108,
575-633, 1992.