Jacobi-Sobolev orthogonal polynomials: asymptotics and a Cohen type inequality (Q387366)

The authors study the Jacobi-Sobolev orthogonal polynomials \(\{Q_n^{(\alpha,\beta,N)}\}_{n\geq 0}\), orthogonal w.r.t. \[ \langle f,g\rangle_S=\sum_{j=0}^N\,\int_{-1}^1\,f^{(j)}(x)g^{(j)}(x)d\mu_{\alpha,\beta}(x), \] where \(\alpha,\beta>-1\) and \(\lambda_j\geq 0\) \((0\leq j\leq N-1)\), \(\lambda_N>0\), \(N\geq 1\). For \(N=1\), \(\lambda_1=1\), the ordinary Jacobi polynomials arise; as normalization, the paper uses \(P_n^{(\alpha,\beta)}(1)= {n+\alpha\choose n}\). Moreover, the authors reintroduce (cf. the papers in the list of references) the polynomials \[ R_{n,N}^{(\alpha,\beta)}(x)=P_n^{(\alpha-N,\beta-N)}(x), \] with \((\alpha,\beta)\) omitted if there is no confusion. In Section 2, known properties of these polynomials and their monic forms are given. The Sections 3 and 4 contain the new results for the polynomials \(\{Q_n^{(\alpha,\beta,N)}\}_{n\geq 0}\). { Proposition 3.1:} \(Q_n^{(\alpha,\beta,N)}=R_{n,N}^{(\alpha,\beta)}(x)\) and an explicit expression for \(R_{n,N}^{(\alpha,\beta)}(x)\) in terms of \(Q_j^{(\alpha,\beta,N)}\) \((N-2n+2s\leq j\leq N)\), when there exists \(0\leq j\leq N-1\) such that \(\lambda_j\not= 0\), where \(s=\min\{j\,:\,\lambda_j\not= 0\}\), \(0\leq j\leq N\). { Proposition 3.2:} an explicit expression of \(Q_n^{(\alpha,\beta,N)}\) in terms of \(R_{j,N}^{(\alpha,\beta)}\), \(0\leq j\leq n\). { Proposition 3.3:} Mehler-Heine type formula, inner strong asymptotics and outer relative asymptotics for \(Q_n^{(\alpha,\beta,N)}\). { Proposition 3.4:} estimates for the Sobolev norms of the \(Q_n^{(\alpha,\beta,N)}\) for \(\alpha,\beta\geq -1/2\). In Section 4, the Fourier expansion of a function is studied \[ \sum_{k=0}^{\infty}\,\hat{f}(k)Q_k(x)\text{ with }\hat{f}(k)={\langle f,Q_k\rangle_S\over ||Q_k||^2_{W^{N,2}}},\;k\geq 0. \] For a function \(f\in W^{N,p}\) and a fixed sequence \(\{c_{k,n}\}_{k=0}^n\) \((n\in\mathbf{N}\cup\{0\}\) of real numbers with \(c_{n,k}= o(n^{ k-n}c_{n,n}),\;k=n-2N,\ldots, n-1\), operators \({\mathcal T}_n\) are defined by \[ {\mathcal T}_n(f)=\sum_{k=0}^n\,c_{k,n}\hat{f}(k)Q_k, \] and in { Theorem 4.1}, a lower bound for the norm (in \(W^{N,p}\)) of these operators is given in the case \(\alpha,\beta\geq -1/2\), \(\max\{\alpha,\beta\}>-1/2\).