Planar orthogonal polynomials as type I multiple orthogonal polynomials (Q6042956)

Let \(P_{n}(z)\) be the monic polynomial of degree \(n\) such that \[ \int_ {\mathbb{C}} P_{n} (z) \overline{z}^{k} e^{- |z|^{2}} |W(z)|^{2} d A(z)=0, \ \ \ k=0, 1, \dots, n-1.\] Here \(dA\) is the Lebesgue area measure of the complex plane and \(W(z)= \prod_{j=1}^{p} (z-a_{j})^{c_{j}}\), \(p\geq 1,\) where \(\{ c_{j}\}_{j=1}^{p}\) are positive real numbers and \(\{ a_{j}\}_{j=1}^{p}\) are complex numbers. The polynomial \(P_{n}(z)\) is said to be a planar orthogonal polynomial of degree \(n.\) Since the \(c_{j}\) are not necessarily integer numbers the branch cuts and the branches must be fixed so that \(W(z)\) be unambiguous. \textit{S.-Y. Lee} and \textit{M. Yang} [J. Phys. A, Math. Theor. 52, No. 27, Article ID 275202, 14 p. (2019; Zbl 1509.33010)] prove that given positive integer numbers \(n\), \(p,\) such that \(n= p \kappa +r\), \(0 \leq r <p\), then \(P_{n}(z)= q_{\mathfrak{n}}(z)\) is a type II orthogonal polynomial of order \(\mathfrak{n}= (\underbrace{\kappa+1, \dots, \kappa+1}_{r}, \underbrace{\kappa, \dots, \kappa}_{p-r})\). Indeed, such a type II multiple orthogonality is defined as \(\int_{\Gamma} q_{\mathfrak{n}}(z) z^{k} \chi_{\mathfrak{n}- \mathfrak{e}_{j}} (z) dz=0\), \(0 \leq k \leq n_{j}-1\), \(1\leq j\leq p\), where \(\Gamma\) is a simple closed curve with counterclockwise orientation that can be chosen as \(\overline{a_{1} a_{2}}\cup \overline{a_{2} a_{3}}\cup \overline{a_{p-1} a_{p}}\cup \overline{a_{p} a_{1}}\) and \(\chi_{\mathfrak{k}}(z)= W(z) \int _{\gamma} \prod_{j=1}^{p} (s- \overline{a}_{j})^{k_{j}} \overline{W}(s) e^{-zs} ds\), where \(\mathfrak{k}= (k_{1}, \dots, k_{p})\) and \(\gamma= \{ t \overline{z}, t\geq 0\}\). These planar polynomials appear as solutions of a Riemann-Hilbert problem in the framework of type II multiple orthogonality. Assuming all \(c_{j}\) are positive integer numbers, in the paper under review the authors provide two new characterizations of planar orthogonal polynomials. The first one is given in terms of a polynomial differential operator related to the polynomial \(W(z)\) in such a way \( W^{\ast} (D_{z})[ P_{n}(z) W(z)]= O(z^{n})\), \(z\rightarrow0\), where \(W^{\ast} (s)= \prod_{j=1}^{p} (s-\overline{a}_{j})^{c_{j}}\). In the second one, there exist polynomials \(Q_{j}(z)\), \(deg Q_{j}\leq c_{j}-1\), such that \(P_{n}(z) W(z) + \sum_{j=1}^{p} Q_{j}(z) e^{\overline{a_{j}} z}=O(z^{c_{1}+ \cdots + c_{p}})\), \(z\rightarrow0\). This can be regarded as a type I Hermite-Padé approximation problem at the origin. The proofs of them rely on a fundamental identity by Lee and Yang [loc. cit.]. Such an identity is proved in an alternative way by using elementary facts of complex analysis and the use of the Stokes Theorem. Indeed, the integral on the complex plane is written in polar coordinates, and, by analyticity, the angular integral is deformed to an integral over a closed contour around the origin. The planar orthogonality leads to three different but closely related Riemann-Hilbert problems for the type I multiple orthogonality as it is pointed out in the paper under review. Notice that there is a canonical correspondence between the Riemann-Hilbert problems for the characterization of type I and type II multiple orthogonalities as proved in [\textit{W. Van Assche} et al., NATO Sci. Ser. II, Math. Phys. Chem. 30, 23--59 (2001; Zbl 0997.42012)]. On the other hand, when \(c_{j}\), \(j=1, \dots, p,\) are integer numbers, the planar orthogonal polynomials can also be expressed as ratios of determinants, see [\textit{G. Akemann} and \textit{G. Vernizzi}, Nucl. Phys., B 660, No. 3, 532--556 (2003; Zbl 1030.82003)].