Multivariate Hypergeometric Terms
A multivariate function \(f(z_1,\dots, z_k)= f(z)\) from \(\mathbb{Z}^k\) to be a field \(K\) is a hypergeometric term if for each \(i\in \{1,\dots, k\}\) there exist nonzero polynomials \(A_i(z)\) and \(B_i(z)\) in \(K[z]\) such that \(A_i(z)f(z)= B_i(z) f(z+e_i)\) for all \(z\in \mathbb{Z}^k\). Here \(e_1,\dots, e_k\) is the standard basis for \(\mathbb{Z}^k\). The multivariate hypergeometric terms that arise in practice have the form \[ f(z)= \gamma_1^{z_1}\cdots \gamma_k^{z_k} \frac {C(z) \prod_{i=1}^p (m_i)_{v_i z+r_i}} {D(z) \prod_{j=1}^q (n_j)_{w_j z+s_i}} \] where \(\gamma_1,\dots, \gamma_k\in K\), \(C\) and \(D\) are polynomials in \(K[z]\), the \(v_i\) and \(w_j\) are in \(\mathbb{Z}^k\), the \(r_i\) and \(s_j\) are in \(\mathbb{Z}\), the \(m_i\) and \(n_j\) are in \(K\), and \((m)_r= m(m+1)\cdots (m+r-1)\) is the Pochhammer symbol. The question is: do all hyprgeometric terms have this form? We prove that if \(K\) is an algebraically closed field of characteristic 0, and the hypergeometric term is honest (meaning that each pair \(A_i\) and \(B_i\) above can be taken to be relatively prime), then such an expression for \(f\) exists piecewise (the domain of \(f\) can be expressed as a finite union of hyperplanes and polyhedral regions such that on each polyhedral region, there exists an expression for \(f\) as above). We use this result to settle the discrete part of a conjecture [Invent. Math. 103, 575-634 (1992; Zbl 0739.05007)] of \textit{H. S. Wilf} and \textit{D. Zeilberger} by showing that a holonomic hypergeometric term is piecewise proper, which means roughly that expressions for \(f\) exist on polyhedral regions as above, and in addition we can take \(D(z)=1\). The above results depend on a theorem that gives the structure of rational solutions of a system of equations satisfied by the term ratios of a hypergeometric term. The author was unaware that this theorem (our Theorem 2.8.4) is due to M. Sato and that its two variable case is due to O. Ore. We give a new proof of Sato's theorem using a lemma of Gosper. We solve an unrelated problem of Cameron. A set \(S\) is sum-free if \(S+S\subset S^c\), complete if \(S+S\supset S^c\) and symmetric if \(S=-S\). We solve Cameron's problem by showing that for all sufficiently large moduli \(m\) there exists a sum-free complete set \(S\) in \(\mathbb{Z}/m\mathbb{Z}\) that is not symmetric and we observe that such sets have the property that \(|S+S|>|S-S|\), which is relevant to an erstwhile conjecture of Conway disproved by Marica.
- Proof of the Wilf-Zeilberger conjecture for mixed hypergeometric terms
- On the structure of multivariate hypergeometric terms.
- \(k\)-free recurrences of double hypergeometric terms
- Multiplicative decompositions of multivariate q-hypergeometric terms
- Multivariable hypergeometric functions
- Rational normal forms and minimal decompositions of hypergeometric terms
This page was built for publication: Multivariate Hypergeometric Terms
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4806202)