Sobolev orthogonal polynomials of several variables on product domains (Q2238745)

From MaRDI portal





scientific article; zbMATH DE number 7418707
Language Label Description Also known as
default for all languages
No label defined
    English
    Sobolev orthogonal polynomials of several variables on product domains
    scientific article; zbMATH DE number 7418707

      Statements

      Sobolev orthogonal polynomials of several variables on product domains (English)
      0 references
      0 references
      0 references
      0 references
      2 November 2021
      0 references
      In [\textit{L. Fernández} et al., J. Comput. Appl. Math. 284, 202--215 (2015; Zbl 1312.33039)], polynomial bases orthogonal in terms of the weight \(w_{1} (x)w_{2}(y)\) supported on the rectangle \(A= [a_{1}, b_{1}]\times [a_{2},b_{2}]\) with respect to the inner product \[\langle f, g\rangle = \int _{A} \nabla f(x, y) \cdot \nabla g(x, y)w_{1} (x)w_{2}(y) dxdy+\lambda f(c_{1}, c_{2})g(c_{1} ,c_{2}),\] where \((c_{1},c_{2})\) is a fixed point of the two-dimensional Euclidean space, are studied. The authors focus the attention on two particular examples, the product of two Laguerre and the product of two Gegenbauer weight functions, respectively with different choices of the parameters. The motivation for using this kind of inner product comes from problems of approximation of solutions of partial differential equations that require control over the gradient. In the contribution under review, the authors deal with a natural extension of the above paper by analyzing algebraic properties of multivariate orthogonal polynomials with respect to a discrete-continuous Sobolev inner product \[\langle f, g\rangle _{S} = \sum_{j=0}^{\kappa-1} \lambda_{j} \nabla^{j}f(\textbf{p})\cdot \nabla^{j}g(\textbf{p})+ c\int_{\Omega} \nabla^{\kappa} f(\textbf{x})\cdot \nabla^{\kappa} g(\textbf{x}) W(\textbf{x}) d\textbf{x}.\] Here, \(\nabla^{j}f, j=0,1, \cdots, \kappa,\) denotes the column vector containing all the partial derivatives of order \(j\) of the function \(f\), the weight function \( W(\textbf{x}) = w_{1}(x_{1}), \cdots w_{d}(x_{d})\) is the product of \(d\) weight functions \(w_{k}(x)\) supported on intervals \([a_{k}, b_{k}], k=1, 2, \cdots, k, \) of the real line, \(\lambda_{j}> 0, j=0, \cdots, \kappa-1, \) \(c\) is the normalization constant of the weight \(W,\) and \(\textbf{p}\) is a point in the \(d\)-dimensional Euclidean space. In other words, they deal with a particular case of weight function supported on the parallelepiped \(\Omega= \{\textbf{x}=(x_{1}, x_{2}, \cdots, x_{d}), x_{k}\in (a_{k}, b_{k}) \}\). A connection formula between an orthogonal polynomial basis with respect to the discrete-continuous Sobolev inner product \(\langle \cdot, \cdot \rangle _{S}\) and an orthogonal polynomial basis with respect to the continuous component of the above inner product is deduced. Assuming each weight \(w_{k}, k=1, 2, \cdots, d, \) is classical (Hermite, Laguerre, Jacobi) and following the ideas developed when \(d = 2\) in [\textit{H. Dueñas Ruiz} et al., Integral Transforms Spec. Funct. 28, No. 12, 988--1008 (2017; Zbl 1379.33020)], a connection formula between the corresponding polynomial bases defined as column vectors is given. The matrix coefficients in such a connection formula are represented in a compact way. Finally, the above results are illustrated in the case \(d = 3\) for the Hermite-Hermite-Laguerre product weight function.
      0 references
      orthogonal polynomials
      0 references
      several variables
      0 references
      classical orthogonal polynomials
      0 references
      product domains
      0 references
      Sobolev polynomials
      0 references
      0 references
      0 references
      0 references
      0 references
      0 references
      0 references
      0 references

      Identifiers