Conjectured inequalities for Jacobi polynomials and their largest zeros (Q2458309)

Let \(P_{n}^{(\alpha,\beta)}(x)\) be the Jacobi polynomials of degree \(n\) and parameters \((\alpha,\beta)\). The authors consider inequalities related to the largest zeros of Jacobi polynomials and also inequalities involving the scaled Jacobi polynomials having the value \(1\) at \(x=1\). All inequalities are only conjectured to hold, but compelling evidence is provided, both numerical and analytic, in support of their validity. More specifically, let \(x_{n}^{(\alpha)}=\cos \theta_{n}^{(\alpha)}\), \(0<\theta_{n}^{(\alpha)}<\pi\), be the largest zero of the Jacobi polynomial \(P_{n}^{(\alpha,\alpha-1)}(x)\), \(\alpha>0\). Then the authors consider the inequality \[ n\theta_{n}^{(\alpha)}<(n+1)\theta_{n+1}^{(\alpha)},\tag{1} \] for which they state the following Conjecture 1: Given \(\alpha>0\), there are two alternatives: either (1) holds for all \(n=1,2,\ldots\), or (1) is false for \(n=1\). In other words, the validity of (1) for \(n=1\) implies the validity of (1) for all \(n\geq 1\). The next conjecture involves the scaled Jacobi polynomials \(\widetilde{P}_{n}^{(\alpha,\beta)}(x) :=\frac{P_{n}^{(\alpha,\beta)}(x)} {P_{n}^{(\alpha,\beta)}(1)}\) and the inequality \[ \widetilde{P}_{n}^{(\alpha,\alpha-1)}\Big(\cos \frac{\theta}{n}\Big) <\widetilde{P}_{n+1}^{(\alpha,\alpha-1)}\Big(\cos \frac{\theta}{n+1}\Big),\tag{2} \] which is studied in the two intervals \(0<\theta<\theta_{1}^{(\alpha)}\) and \(0<\theta<\pi\) (\(*\)). Conjecture 2: Given \(\alpha>0\), there are two alternatives for each of the two intervals (\(*\)): either (2) holds for all \(n=1,2,\ldots\) and all \(\theta\) in the respective interval, or (2) is false for \(n=1\) and some \(\theta\) in the respective interval. In other words, the validity of (2) for \(n=1\) implies the validity of (2) for all \(n\geq 1\). The authors state analogous conjectures for general Jacobi polynomials. Let \(x_{n}^{(\alpha,\beta)}=\cos \theta_{n}^{(\alpha,\beta)}\), \(0<\theta_{n}^{(\alpha,\beta))}<\pi\), be the largest zero of the Jacobi polynomial \(P_{n}^{(\alpha,\beta)}(x)\), \(\alpha>-1\), \(\beta>-1\). Conjecture 3: Given \(\alpha>-1,\;\beta>-1\), there are two alternatives: either \[ n\theta_{n}^{(\alpha,\beta)}< (n+1)\theta_{n+1}^{(\alpha,\beta)}, \tag{3} \] holds for all \(n=1,2,\ldots\), or (3) is false for \(n=1\). In other words, the validity of (3) for \(n=1\) implies its validity for all \(n\geq 1\). Then the authors consider the inequality \[ \widetilde{P}_{n}^{(\alpha,\beta)}\Big(\cos \frac{\theta}{n}\Big)<\widetilde{P}_{n+1}^{(\alpha,\beta)}\Big(\cos \frac{\theta}{n+1}\Big),\tag{4} \] on either of the two intervals \(0<\theta<\theta_{1}^{(\alpha,\beta)}\) and \(0<\theta<\pi\) (\(**\)). Conjecture 4: Given \(\alpha>-1\), \(\beta>-1\), there are two alternatives for each of the two intervals (\(**\)): either (4) holds for all \(n=1,2,\ldots\) and all \(\theta\) in the respective interval, or (4) is false for \(n=1\) and some \(\theta\) in the respective interval. In other words, the validity of (4) for \(n=1\) implies the validity of (4) for all \(n\geq 1\). Conjectures 3 and 4 can be combined into the following Conjecture 5: Given \(\alpha>-1,\;\beta>-1\), if (4) holds for \(n=1\) and \(0<\theta\leq\theta_{1}^{(\alpha,\beta)}\), then (4) holds for \(0<\theta\leq n\theta_{n}^{(\alpha,\beta)}\) for all \(n=1,2,\ldots\). Note that inequality (4) has been proved by the reviewer in the cases where \[ (\alpha, \beta)\in \Biggl\{\biggl(\frac{1}{2},\frac{1}{2}\biggr), \biggl(\frac{1}{2},-\frac{1}{2}\biggr), \biggl(-\frac{1}{2}, \frac{1}{2}\biggr)\Biggr\}, \] for all \(n=1,2,\dots\) and \(0<\theta<\pi\), in the recently published paper [``On a conjectured inequality of Gautschi and Leopardi for Jacobi polynomials'', Numer. Algorithms 44, No. 3, 249--253 (2007; Zbl 1123.26016)]. The case \(\alpha=\beta=-1/2\) of (4) is trivial as both of its sides are identically equal to \(\cos \theta\).