Exceptional Meixner and Laguerre orthogonal polynomials (Q2016139)

Exceptional orthogonal polynomials have been introduced in \textit{D. Gómez-Ullate} et al. [J. Math. Anal. Appl. 359, No. 1, 352--367 (2009; Zbl 1183.34033)] (see also \textit{D. Gómez-Ullate} et al. [J. Approx. Theory 162, No. 5, 987--1006 (2010; Zbl 1214.34079)]) as sequences of polynomials with a finite number of gaps in their degrees, but which are orthogonal with respect to a positive Borel measure \(\mu\) and constitute a complete orthonormal basis of the corresponding \(L^{2}(\mu)\) space. The first examples of exceptional orthogonal polynomials appear in the literature in the framework of the standard Bochner classification for polynomial eigenfunctions of second order linear differential operators that characterizes the so called classical orthogonal polynomials (Hermite, Laguerre and Jacobi) when a finite number of gaps appear in the degree of such polynomials. The study of exceptional orthogonal polynomials has known an increasing interest in the last seven years not only from theoretical purposes but also by their applications in super-symmetric (SUSY) transformations, quasi-exact solvability and shape-invariant potentials (see [\textit{C. Quesne}, ``Exceptional orthogonal polynomials, exactly solvable potentials ans supersymmetry'', J. Phys. A 41, 392001--392007 (2008)] as well as [\textit{S. Odake} and \textit{R. Sasaki}, ``Infinitely many shape invariant potentials and new orthogonal polynomials'', Phys. Lett. B 679, 414--417(2009)] among others). For second order linear difference operators, there is a characterization of the so called discrete orthogonal polynomials (Charlier, Meixner, Krawtchouk and Hahn) as eigenfunctions of such operators without gaps in their degrees, see \textit{O. E. Lancaster} [Am. J. Math. 63, 185--207 (1941; JFM 67.0304.03)]. Thus, it seems a natural question to extend the concept of exceptional polynomials to the discrete case. This task was started by the author of the paper under review in [\textit{A. J. Durán}, J. Approx. Theory 182, 29--58 (2014; Zbl 1298.33016)]. Therein, two basic ideas as the concept of duality between two sequences of polynomials and the Christoffel transformations of measures, respectively, constitute the key tools for the analysis of the first example of discrete exceptional orthogonal polynomials (Charlier) and, as a consequence of a natural limit process, one obtains Hermite exceptional polynomials. In the paper under review, the author deals with the Meixner case. For a pair \(\mathcal{F}=(F_{1},F_{2})\) of finite sets of positive integers, let denote by \(k_{i}\) the cardinal of \(F_{i}\), \(i=1,2\), \(k= k_{1}+ k_{2}\), and let us introduce the nonnegative number \[ u_{\mathcal{F}}= \sum_{f\in F_{1}} f + \sum_{f\in F_{2}} f- \frac{(k_{1}+ 1) k_{1}}{2}- \frac{k_{2}(k_{2}-1)}{2} \] as well as the infinite set of nonnegative integers \[ \sigma_{\mathcal{F}}= \{u_{\mathcal{F}} + n ; n=1, 2,\dots\}\backslash \{u_{\mathcal{F}}+ f, f\in F_{1} \}. \] Notice that one has a finite number of gaps in such an infinite set. In the next step, a sequence of polynomials \((m^{a,c;\mathcal{F}}_{n}(x))_{n\in \sigma_{\mathcal{F}}}\) is defined by the \(k\)- Casoratian of the Meixner polynomials \(m^{a,c}_{n- u_{\mathcal{F}}}(x)\), \(m^{a,c}_{f}(x)\), \(f\in F_{1}\), and \(m^{1/a,c}_{f}(x)\), \(f\in F_{2}\). As a first result, the author proves that this sequence and the sequence of polynomials orthogonal with respect to the Christoffel transformation \(\rho^{\mathcal{F}}_{a,c} = \pi(x) \rho_{a,c}\) of the Meixner orthogonality measure \[ \rho_{a,c}= \sum_{x=0}^{\infty} \frac {a^{x} \Gamma(x+c)}{x!} \delta_{x}, \] with \[ \pi(x)= \prod_{f\in F_{1}} (x- f) \prod_{f\in F_{2}} (x+c+ f), \] are dual sequences. As a consequence, the polynomials \(m^{a,c;\mathcal{F}}_{n}(x),\) \(n\in \sigma_{\mathcal{F}},\) are eigenfunctions of a second order linear difference operator with rational coefficients (Theorem 3.3). The connection with the Darboux transformation based on the factorization of this operator in two first order difference operators is stated in Lemma 3.4. On the other hand, if \(c\) is a real and is not a nonpositive number let denote \(\hat{c}= \max \{- [c], 0 \}\) and let assume that the rational function \(\frac{\pi(x)}{(x+c)_ {\hat{c}}}\) is nonnegative in the set of positive integer numbers (admissibility condition for \(c\) and \(\mathcal{F}\)), then in Theorem 4.3 it is proved that the sequence of polynomials \((m^{a,c;\mathcal{F}}_{n}(x))_{n\in \sigma_{\mathcal{F}}}\) is orthogonal with respect to the positive measure \[ \omega^{\mathcal{F}}_{a,c} = \sum_{x=0}^{\infty} \frac {a^{x} \Gamma(x+c+k)}{x! \Omega^{a,c}_{\mathcal{F}}(x) \Omega^{a,c}_{\mathcal{F}}(x+1)}\delta_{x}. \] \(\Omega^{a,c}_{\mathcal{F}}\) denotes the \(k\)-Casoratian of the Meixner polynomials \(m^{a,c}_{f}(x)\), \(f\in F_{1}\), and \(m^{1/a,c}_{f}(x)\), \(f\in F_{2}\). Furthermore, \((m^{a,c;\mathcal{F}}_{n}(x))_{n\in \sigma_{\mathcal{F}}}\) constitutes a complete orthonormal basis in the \(L^{2}(\omega^{\mathcal{F}}_{a,c})\) space (Theorem 4.4). Further, considering the dilation in the variable \(x\rightarrow x/(a-1)\), writing \(c= \alpha+1,\) and taking limit when \(a\) tends to \(1\) in the \(k\)-Casoratian expression of \(m^{a,c,\mathcal{F}}_{n} (x),\) the author proves that the corresponding sequence of polynomials are eigenfunctions of a second order linear differential operator with rational functions as coefficients (Theorem 5.2). Finally, assuming the admissibility of \(\alpha+1\) and \(\mathcal{F}\), the author deduces the orthogonality of such limit polynomials with respect to a positive weight \(\omega_{\alpha; \mathcal{F}}\) that is a positive rational perturbation of the Laguerre weight \(x^{\alpha + k} e^{-x}\) as well as they constitute a complete orthonormal basis in the corresponding \(L^{2}(\omega_{\alpha; \mathcal{F}})\) space (Theorem 6.3). Thus, one has a sequence of Laguerre \(k\)-exceptional orthogonal polynomials. As a conclusion, this contribution gives an original and constructive approach to generate exceptional Meixner polynomials. Exceptional Laguerre polynomials appear by a limit process. Notice that the approach is the same as in the Charlier case but the technicalities are higher.