A two-line algorithm for proving \(q\)-hypergeometric identities (Q1378409): Difference between revisions

In an earlier paper [\textit{Y. Lily}, J. Math. Anal. Appl. 198, No. 3, 856-878 (1996; Zbl 0857.33002)] the author showed that hypergeometric identies \(\sum_kF(n,k)=f(n), n\geq n_0\) can be proved by checking a finite number of consecutive values of \(n\), id est for \(n\in\{n_0,n_0+1,\ldots,n_1\}\). Moreover, a first (theoretical) estimate of \(n_1\) was given in terms of the coefficients in \(F(n,k)\). The paper under review can be seen as a \(q\)-analogue of this earlier result. The algorithm is based on the \(q\)-analogue of Zeilberger's construction for proving hypergeometric identities [see \textit{T. H. Koornwinder}, J. Comput. Appl. Math. 48, No. 1-2, 91-111 (1993; Zbl 0797.65011)]. This construction is based on Sister Celine's method [see \textit{D. Zeilberger}, J. Math. Anal. Appl. 85, 114-145 (1982; Zbl 0485.05003) and also the marvellous book of \textit{M. Petkovsek, H. S. Wilf} and \textit{D. Zeilberger}, \(A=B\), Wellesley, MA, A.K. Peters (1996, Zbl 0848.05002)]. The first ``a priori'' estimate for \(n_1\) is too big and complicated to be of pratical use, but it might be possible that these estimates can be made smaller as the author indicates. More on algorithmic ways of proving hypergeometric identities can be found in [\textit{H. S. Wilf} and \textit{D. Zeilberger}, Invent. Math. 108, 575-634 (1992; Zbl 0782.05009)].