Limits of elliptic hypergeometric biorthogonal functions (Q2339504): Difference between revisions

Elliptic hypergeometric functions constitute a generalization of hypergeometric (resp. \(q\)-hypergeometric) series when the ratio of two consecutive terms is not a rational function of \(n\) (resp. \(q^{n}\)) but an elliptic function of \(n\). Many important identities for hypergeometric functions have been generalized to the elliptic level (see, for instance, the survey paper [\textit{V. P. Spiridonov}, Russ. Math. Surv. 63, No. 3, 405--472 (2008); translation from Usp. Mat. Nauk 63, No. 3, 3--72 (2008; Zbl 1173.33017)]. A family of biorthogonal elliptic hypergeometric functions, which can be considered as elliptic versions of the Askey-Wilson polynomials, has been analyzed in [\textit{V. Spiridonov} and \textit{A. Zhedanov}, C. R. Math. Acad. Sci., Soc. R. Can. 22, No. 2, 70--76 (2000; Zbl 0974.33016)], where the measure corresponds to the integrand of the elliptic beta integral. Such biorthogonal functions are related by interchanging two of their parameters a situation that does not appear for Askey-Wilson polynomials and determine the location of poles of the biorthogonal functions. In the paper under review, the authors deal with different limits of biorthogonal systems of functions which include the entire \(q\)-Askey scheme as well as many pairs of rational functions which are biorthogonal to each other, families of rational functions biorthogonal to families of polynomials, and families of polynomials biorthogonal on the unit circle. Based on the approach by \textit{F. J. Van De Bult} and \textit{E. M. Rains} [SIGMA, Symmetry Integrability Geom. Methods Appl. 5, Paper 059, 31 p. (2009; Zbl 1188.33025)], tiling the polytope \(P^{0}\) with three different tiles and taking into account each face of any of these tiles corresponds to some limit of the beta integral evaluation, the picture is extended by associating to each simplicial face a system of biorthogonal functions, with orthogonality measure given by the appropriate beta integral limit. One system of biorthogonal functions is a limit of another one if the associated face contains the face of the another system in its boundary. If two faces differ by shifts along vectors in the root lattice of \(E_{6}\), then the corresponding biorthogonal functions are the same, though the two faces give two different measures. If \(\tilde{R}_{n} (z; t_{0}: t_{1},t_{2},t_{3};u_{0},u_{1};q;p)\) is the family of elliptic hypergeometric series \(_{12}V_{11}\) which is biorthogonal to the same functions with the \(u\)-parameters interchanged, by rescaling the variable \(z\) and the parameters \(t\) and \(u,\) by powers of \(p\), then under some conditions there exist the limits of such functions when \(p\) tends to zero and they are generically non-zero. Thus the authors focus the attention on those limit cases assuming they depend on the variable \(z\) as well as the limit of the bilinear form \(\langle \tilde{R}_{n}, \tilde{R}_{m} \rangle\) makes a sense. These two conditions yield a classification of 38 systems of biorthogonal functions and for each system at least one explicit measure for the bilinear form is given. The \(q\)-Askey scheme appears in the degeneration scheme taking into account that orthogonal polynomials are just the special case of biorthogonal rational functions, where the poles of the rational functions have disappeared to infinity and the a priori two different families on both sides of the bilinear form have reduced to identical functions. Pastro polynomials see [\textit{P. I. Pastro}, J. Math. Anal. Appl. 112, 517--540 (1985; Zbl 0582.33010)] constitute a nice example of biorthogonal polynomials, not included in the \(q\)-Askey scheme, which appear in the degeneration scheme.