Gram polynomials and the Kummer function (Q1270292): Difference between revisions

From MaRDI portal
Added link to MaRDI item.
RedirectionBot (talk | contribs)
Removed claims
Property / author
 
Property / author: Q1231998 / rank
Normal rank
 
Property / author
 
Property / author: Germund Dahlquist / rank
Normal rank
 
Property / author
 
Property / author: Kent Pearce / rank
Normal rank
 
Property / author
 
Property / author: Lothar Reichel / rank
Normal rank
 
Property / author
 
Property / author: Kendall C. Richards / rank
Normal rank
 
Property / reviewed by
 
Property / reviewed by: Doron S. Lubinsky / rank
Normal rank
 

Revision as of 06:49, 10 February 2024

scientific article
Language Label Description Also known as
English
Gram polynomials and the Kummer function
scientific article

    Statements

    Gram polynomials and the Kummer function (English)
    0 references
    2 March 2000
    0 references
    In this interesting paper, the authors investigate the Gram polynomials and deduce a sharp inequality for the Kummer function. Let \(m\geq 1\), and let \(x_j:= -1+{2j-1 \over m}\), \(1\leq j\leq m\) denote \(m\) equally spaced points in \((-1,1)\). Define the discrete inner product \((g,h):= {1\over m} \sum^m_{j=1} g(x_j) h(x_j)\). Fix \(n<m\). Then we may define orthonormal polynomials \(\{\varphi_j\}^n_{j=0}\), called the Gram polynomials, satisfying \((\varphi_j, \varphi_k)= \delta_{jk}\). The authors start with the explicitly known three term recurrence relation for the Gram polynomials, namely \[ \varphi_n(x) =2\alpha_{n-1} x\varphi_{n-1} (x)-{\alpha_{n-1} \over\alpha_{n-2}} \varphi_{n-2}(x)\quad \text{where} \quad\alpha_{n-1}: ={m\over n} \left( {n^2-1/4 \over m^2-n^2} \right)^{1/2} \] and \(\varphi_0 (x)=1\), \(\varphi_{-1} (x): =0\). They introduce the parameters \(\tau:=nm^{1/2}\) and \(t:=\tau-{1\over 2} m^{-1/2}\), and show as \(m\to\infty\), the recurrence relation becomes a differential equation \[ {d\varphi (t)\over dt^2}= \Bigl(t^2- \textstyle {1\over 4} t^{-2}-2 \zeta\Bigr) \varphi(t). \] Here \(\zeta\) is appropriately chosen. The general solution of this equation is given in terms of Kummer's function, and another special function. The Kummer function is used to provide an asymptotic expression for \(\varphi_{n-1} (1-\zeta/m)/ \sqrt{2 \sqrt m}\), as \(m\to\infty\). Then the authors go on to prove a sharp inequality for Kummer's function, which had been conjectured in 1985. More precisely, for \(\zeta\geq 0\), \(x\geq 0\), \(c\geq {1\over 2}\) \(F({1-\zeta \over 2}, c,x)\leq F({1\over 2}, c,x)\) where \(F\) is the confluent hypergeometric function. \(F(a,c,z): =\sum^\infty_{n=0} {(a)_n \over(c)_nn!} z^n.\) The symbol \((a)_n\) is of course Pochhammer's symbol, and the inequality fails for \(c<{1\over 2}\).
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    Gram polynomials
    0 references
    Kummer function
    0 references
    confluent hypergeometric function
    0 references