On some general lacunary interpolation problems (Q1125970)

From MaRDI portal
scientific article
Language Label Description Also known as
English
On some general lacunary interpolation problems
scientific article

    Statements

    On some general lacunary interpolation problems (English)
    0 references
    0 references
    0 references
    0 references
    0 references
    23 February 1997
    0 references
    Consider a system \(x_1< \cdots< x_n\) of nodes in the interval \([-1,1]\) and a system \(0=m_0< m_1< \cdots< m_{q-1}\) of natural numbers. The \((0,m_1,\dots,m_n)\) interpolation problem consists in finding a polynomial \(L(x)\), of degree \(\leq nq-1\), such that \(L^{ (m_i)} (x_k)=y_{k,i}\), where \(y_{k,i}\) are arbitrary given numbers, \(k=1,\dots,n\); \(i=1,\dots,q -1\). The problem is called regular if it is uniquely solvable for any set \(\{y_{k,i}\}\) of data. In this case the problem reduces to the determination of the fundamental polynomials \(L_{\nu, m_j} (x)\) of degree at most \(nq-1\), verifying the conditions \(L^{(m_i)}_{\nu,m_j} (x_k)= \delta_{ij} \delta_{\nu k}\), \(\nu,k=1,\dots,n\); \(i,j=1,\dots,q-1\). Assuming \(L_{\nu,m_i} (x)\) known, for \(\nu=1,\dots,n\); \(i= p+1,\dots,q-1\), where \(m_p< q\leq m_{p+1}\), the remaining fundamental polynomials can be given (by an explicit formula) in terms of these known fundamental polynomials and the fundamental Hermite interpolation polynomials for the \((0,1,2,\dots,q-1)\) interpolation problem. The problem of finding the polynomials \(L_{\nu, m_i}\), \(\nu =1,\dots,n\); \(i=p +1,\dots,q-1\), is more complicated. Supposing that the interpolation nodes are the roots of a primitive of the Legendre polynomial \(P_{n-1} (x)\), the authors consider three cases, namely \((0,1,\dots,r-2,r)\), \((0,1,\dots,r-3,r)\), and \((0,1,\dots,r-3, r-1,r)\), show that these problems are regular and give explicit formulae for the last fundamental polynomials \(L_{\nu, m_{q-1}}\). They give also estimates for the Hermite fundamental polynomials as well as for \(L_{\nu,m_i}\). As application, they prove that, in the case of (0,2,3) interpolation, \(|f(x) - R_n(f,x) |=O(\omega_f(1/n) \log n)\), where \(R_n(f,x)=\sum^n_{\nu=1} f(x_k) L_{\nu,0} (x)\).
    0 references
    0 references
    lacunary interpolation
    0 references
    Hermite interpolation
    0 references
    0 references