betaFeedback

WorksMathematics

Orthogonal polynomials on two or more intervals: the Sprugnoli context

Abstract

We show how the group of Sprugnoli arrays on three elements can be used to define families of polynomials that are orthogonal on two intervals, and we indicate how the group can be extended in the hierarchy of Sprugnoli groups to yield families of orthogonal polynomials with more complex supports. In particular, we examine the defining three term recurrences and the Jacobi matrices of these families of orthogonal polynomials, while detailing the Sprugnoli matrix structure of the coefficient matrices of the polynomial families. We also consider integral representations of the principal moments on the defining intervals. Sprugnoli arrays are shown to relate to periodic recurrences where the parameters remain cyclically constant.

Introduction

The theory of orthogonal polynomials has a long and rich history, with many entry points, and many areas of application [9, 14, 15, 19, 20, 24, 32, 35]. In this paper, which looks principally at simple families of polynomials that are orthogonal on two or three intervals, we concentrate on four principal aspects. These are

  • The structure of the coefficient array of the family

  • The corresponding three-term recurrence

  • The corresponding Jacobi matrix, and its role as the production matrix of the inverse of the coefficient array

  • The generating function of the principal moment sequence of the family, and its integral representation on the intervals involved in the case of two intervals.

Orthogonal polynomials supported on several intervals

There is a long history of research into orthogonal polynomials defined on more than one interval [1, 2, 2, 16, 25, 26], and more generally on arcs and curves [11, 22, 23, 27, 28].

The periodicity of the recurrence coefficients plays a central role in this context. This is evidenced in the associated Jacobi matrices. When the periodic coefficients remain constant (do not depend on nn, for instance), major simplifications occur. This is the context of the present paper. Taking inspiration from the case of Chebyshev-type polynomials and their association with Riordan arrays [5, 30], we examine orthogonal polynomials on two and three intervals defined by the extension of the Riordan group by the Sprugnoli groups [4]. A major link between the elements of these groups and orthogonal polynomials is that these group elements are lower-triangular matrices determined by their production matrices, paralleling the fact that the Jacobi matrices of orthogonal polynomials are determined by the coefficient array of the polynomials.

Recap on the Riordan group and Chebyshev-like orthogonal polynomials

Given a family of orthogonal polynomials Pn(x)P_n(x), the coefficient array CC of the family is the matrix C=(cn,k)C=(c_{n,k}) where

cn,k=[xk]Pn(x).c_{n,k}=[x^k]P_n(x).

Here, [xk][x^k] is the linear functional that acts on power series in xx to return the coefficient of xkx^k. As Pn(x)P_n(x) is of degree nn, the matrix CC is a lower-triangular matrix. Because Pn(x)P_n(x) is of exact degree nn, the matrix CC is invertible and the inverse of this matrix is associated with the moments of the family of orthogonal polynomials. In certain circumstances, the lower-triangular matrix CC can have a special structure which simplifies its analysis and applications. For instance, we can ask about the generating functions of the columns of the matrix CC (when viewed as sequences), and about those of the columns of its inverse. In certain cases, there is a structure to the sequence of these generating functions.

An example of this is given by looking at the Chebyshev polynomials of the second kind Un(x)U_n(x) [21, 29]. When we examine the coefficient array of the family of generalized scaled Chebyshev polynomials given by

Pn(x)=(b)nUn(x−a2b)−α(b)n−1Un−1(x−a2b)−β(b)n−2Un−2(x−a2b)P_n(x)=(\sqrt{b})^n U_n\left(\frac{x-a}{2\sqrt{b}}\right)-\alpha(\sqrt{b})^{n-1}U_{n-1}\left(\frac{x-a}{2\sqrt{b}}\right)-\beta(\sqrt{b})^{n-2}U_{n-2}\left(\frac{x-a}{2\sqrt{b}}\right)

we find that the generating functions of the columns of this matrix form a geometric series (of power series). This indicates that the coefficient array is an element of the Riordan group [6, 30, 31]. Specifically, it is given by the element

(1−αx−βx21+ax+bx2,x1+ax+bx2)(1)\left(\frac{1-\alpha x-\beta x^2}{1+ax+bx^2},\frac{x}{1+ax+bx^2}\right) \tag*{(1)}

of the Riordan group [6, 30, 31].

To describe this group, we recall that F0={∑n=0∞anxn∣a0≠0}\mathcal{F}_0=\left\{\sum_{n=0}^{\infty}a_nx^n\mid a_0\ne0\right\} and that F1={∑n=0∞anxn∣a0=0, a1≠0}\mathcal{F}_1=\left\{\sum_{n=0}^{\infty}a_nx^n\mid a_0=0,\ a_1\ne0\right\} and in general

Fr={∑n=r∞anxn∣ar≠0}.\mathcal{F}_r=\left\{\sum_{n=r}^{\infty}a_nx^n\mid a_r\ne0\right\}.

For simplicity, we shall assume that the coefficients ana_n satisfy an∈Ca_n \in\mathbb{C}. Elements of F0\mathcal{F}_0 are multiplicatively invertible, and elements of F1\mathcal{F}_1 are compositionally invertible. The Riordan group is then the group of pairs (g,f)∈F0×F1(g,f) \in\mathcal{F}_0 \times\mathcal{F}_1 with the following product rule

(g(x),f(x))⋅(u(x),v(x))=(g(x)u(f(x)),v(f(x)))(2)(g(x),f(x))\cdot(u(x),v(x))=(g(x)u(f(x)),v(f(x))) \tag*{(2)}

and inverse

(g,f)−1=(1g(fˉ),fˉ(x)).(g,f)^{-1}=\left(\frac{1}{g(\bar{f})},\bar{f}(x)\right).

where fˉ\bar{f} is the compositional inverse of f∈F0f \in\mathcal{F}_0 (that is, fˉ\bar{f} is the solution uu of the equation f(u)=xf(u)=x for which u(0)=0u(0)=0). The identity of this group is (1,x)(1,x). To each element of this group we can associate in a unique way an invertible lower-triangular matrix (tn,k)(t_{n,k}) by means of

tn,k=[xn]g(x)f(x)k.t_{n,k}=[x^n]g(x)f(x)^k.

where [xn][x^n] is the functional on C[[x]]\mathbb{C}[[x]] that extracts the coefficient of xnx^n. Under this correspondence, the product (2) corresponds to ordinary matrix multiplication. The columns of (g,f)(g,f) have their generating functions given by the geometric series of power series

g,gf,gf2,gf3,gf4,…g,gf,gf^2,gf^3,gf^4,\ldots

The bi-variate generating function of the Riordan array (g(x),f(x))(g(x),f(x)) is given by g(x)1−yf(x)\frac{g(x)}{1-yf(x)}. The product rule (2) is the result of the following description of how a Riordan array acts on a power series.

(g(x),f(x))⋅h(x)=g(x)h(f(x)).(3)(g(x),f(x))\cdot h(x)=g(x)h(f(x)). \tag*{(3)}

This weighted composition is conventionally referred to as the “fundamental theorem of Riordan arrays”.

If M=(g,f)M=(g,f) is a Riordan array, then the matrix PM=M−1MˉP_M=M^{-1}\bar{M}, where Mˉ\bar{M} is the matrix MM with its first row removed, is called the production matrix of the Riordan array MM. We can recover M=(tn,k)M=(t_{n,k}) from PMP_M by

tn,k=(PMn)0,k.t_{n,k}=(P_M^n)_{0,k}.

For instance, the production matrix of the inverse of the Riordan array (1) begins

(a+α100000b+βa100000ba100000ba100000ba100000ba100000ba).\begin{pmatrix} a+\alpha& 1 & 0 & 0 & 0 & 0 & 0 \\ b+\beta& a & 1 & 0 & 0 & 0 & 0 \\ 0 & b & a & 1 & 0 & 0 & 0 \\ 0 & 0 & b & a & 1 & 0 & 0 \\ 0 & 0 & 0 & b & a & 1 & 0 \\ 0 & 0 & 0 & 0 & b & a & 1 \\ 0 & 0 & 0 & 0 & 0 & b & a \end{pmatrix}.

This is the Jacobi matrix of the corresponding family of orthogonal polynomials. In this case, we call the inverse matrix M=C−1M=C^{-1} the moment matrix, since the first column of this inverse matrix contains the moments of the orthogonal polynomials in question.

References

  1. [1]N. I. Achieser, Über einige Funktionen, welche in zwei gegebenen Intervallen am wenigsten von Null abweichen, Bull. Acad. Sci. URSS, VII. 5, Nr. 9 (1932), 1163–1202.
  2. [2]N. I. Achieser, Über eine Eigenschaft der “elliptischen” Polynome, Comm. Kharkov. Math. Soc., 9 (1934), 3–8.
  3. [4]P. Barry, A new group in the Riordan family of matrix groups: the Sprugnoli group, https://arxiv.org/abs/2605.16633.
  4. [5]P. Barry and M. M. Mwfise, Classical and semi-classical orthogonal polynomials defined by Riordan arrays, and their moment sequences, J. Integer Seq., 21 (2018), Article 18.1.5.
  5. [6]P. Barry, Riordan Arrays: a Primer, Logic Press, 2017.
  6. [7]P. Barry, On the group of almost-Riordan arrays, arXiv:1606.05077v1, 2016.arxiv.org/abs/1606.05077
  7. [8]V. Bitonti, B. Deb, A. D. Sokal, Thron-Type Continued Fractions (T-Fractions) for Some Classes of Increasing Trees, Electron. J. Combin., 33 (2026) #P1.5.arxiv.org/abs/2412.10214
  8. [9]T. S. Chihara, An Introduction to Orthogonal Polynomials, Dover Publications, 2011.
  9. [10]C. Corsani, D. Merlini, and R. Sprugnoli, Left-inversion of combinatorial sums, Discrete Math., 180 (1998) 107–122.DOI
  10. [11]M. G. De Bruin, Polynomials orthogonal on a circular arc, J. Comput. Appl. Math., 31 (1990), 253–266.DOI
  11. [12]E. Deutsch, L. Ferrari and S. Rinaldi, Production matrices and Riordan arrays, Ann. Comb., 13 (2009), 65–85.arxiv.org/abs/math/0702638
  12. [13]P. Flajolet, Combinatorial aspects of continued fractions, Disc. Math., 306 (2006), 992–1021.
  13. [14]W. Gautschi, Orthogonal Polynomials, Oxford Science Publications, OUP, 2004.
  14. [15]L. Ya. Geronimus, Orthogonal Polynomials, Consultants Bureau, 1961.
  15. [16]J. S. Geronimus and W. Van Assche, Orthogonal polynomials with asymptotically periodic recurrence coefficients, J. Approx. Theory, 46 (1986), 251–283.
  16. [18]T.-X. He and R. Sprugnoli, Sequence characterization of Riordan arrays, Discrete Math., 309 (2009), 3962–3974.DOI
  17. [19]M. E. H. Ismail, Classical and Quantum Orthogonal Polynomials in one Variable, Cambridge University Press, 2009.
  18. [20]S. Khrushchev, Orthogonal Polynomials and Continued Fractions (From Euler’s Point of View), Cambridge University Press, 2008.
  19. [21]J. C. Mason and D. C. Handscomb, Chebyshev Polynomials, Chapman & Hall/CRC, Boca Raton, FL, 2003.
  20. [22]G. V. Milovanović and P. M. Rajković, On polynomials orthogonal on a circular arc, J. Comput. Appl. Math., 51 (1994) 1–13.DOI
  21. [23]G. V. Milovanović, Some nonstandard types of orthogonality (a survey), Filomat, 9 (1995), 517–542.
  22. [24]E. M. Nikishin and V. N. Sorokin, Rational Approximations and Orthogonality, American Mathematical Society, 1991.DOI
  23. [25]F. Peherstorfer, Elliptic orthogonal and extremal polynomials, Proc. London Math. Soc., s3-70 (1995), 605–624.DOI
  24. [26]F. Peherstorfer, Orthogonal- and Chebyshev polynomials on two intervals, Acta Math. Hung., 55 (1990), 245–278.DOI
  25. [27]F. Peherstorfer and R. Steinbauer, Orthogonal Polynomials on Arcs of the Unit Circle, I, J. Approx. Theory, 85 (1996), 140–184.DOI
  26. [28]F. Peherstorfer and R. Steinbauer, Orthogonal Polynomials on the Circumference and Arcs of the Circumference, J. Approx. Theory, 102 (2000), 96–119.DOI
  27. [29]T. J. Rivlin, Chebyshev Polynomials: From Approximation Theory to Algebra and Number Theory, John Wily & Son, 1990.
  28. [30]L. Shapiro, R. Sprugnoli, P. Barry, G.-S. Cheon, T.-X. He, D. Merlini, and W. Wang, The Riordan Group and Applications, Springer, 2022.DOI
  29. [31]L. W. Shapiro, S. Getu, W. J. Woan, and L. C. Woodson, The Riordan group, Discr. Appl. Math., 34 (1991), 229–239.
  30. [32]J. A. Shohat and J. D. Tamarkin, The Problem of Moments, American Mathematical Society, 1943.DOI
  31. [33]N. J. A. Sloane, The On-Line Encyclopedia of Integer Sequences. Published electronically at http://oeis.org, 2026.
  32. [34]N. J. A. Sloane, The On-Line Encyclopedia of Integer Sequences, Notices Amer. Math. Soc. 50 (2003), 912–915.
  33. [35]G. Szegő, Orthogonal Polynomials, American Mathematical Society, Providence, RI, 1975.
  34. [36]D. Sundarajan, The Discrete Fourier Transform, World Scientific Publishing Co. Pte. Ltd.
  35. [37]G. Viennot, A Combinatorial theory of general orthogonal polynomials with extensions and applications, in: Polynomes Orthogonaux et Applications, Lecture Notes in Mathematics, 1171 (1985), 139–157.DOI
  36. [38]H. S. Wall, Analytic Theory of Continued Fractions, AMS Chelsea Publishing, 2000.

Paper details

vlnc:W_zpwwk4m4
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Mathematics

Not recorded for this paper

View statement
Looking for related papers…
Contents