On the error bounds of the Gauss-type quadrature formulae associated with spaces of parabolic and cubic spline functions with double equidistant knots

Geno Nikolov, Petar B. Nikolov · 2026

Read On the error bounds of the Gauss-type quadrature formulae associated with spaces of parabolic and cubic spline functions with double equidistant knots by Geno Nikolov, Petar B. Nikolov (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2606-25586.txt.

Introduction

Approximating an integral by a weighted sum of function values is quadrature. Gauss type rules tuned to spline spaces are known to be nearly unbeatable for smooth integrands. Geno Nikolov and Petar Nikolov revisit the error constants of those rules in 2026 for parabolic and cubic splines with doubled, evenly spaced knots. Earlier work built the rules for both degrees and estimated how large the error constant can be. The machinery goes back to nineteen ninety five results on asymptotic optimality.

For the cubic case the pair refines the error constant estimate. For both degrees they point out alternative bounds that are easier to compute. Some are sharper than the classical ones, which depend on the largest value of a high derivative of the integrand. The new forms run through Peano kernel representations and Bernoulli polynomials, standard tools they deploy to squeeze the constants.

Catalog record

CREATORGeno Nikolov, Petar B. Nikolov
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE34 KB
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2606.25586

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

On the error bounds of the Gauss-type quadrature formulae associated with spaces of parabolic and cubic spline functions with double equidistant knots Geno Nikolov, Petar B. Nikolov Abstract . In two papers from 1995 P. Köhler and G. Nikolov showed that Gauss–type quadrature formulae associated with spaces of spline functions with equidistant knots are asymptotically optimal in certain Sobolev classes of functions. In particular, Gauss–type quadratures associated with the spaces of spline functions of degree r − 1 r-1 with double equispaced knots are asymptotically optimal definite quadrature formulae of order r r when r r is even, and it is conjectured that the asymptotical optimality property persists also in the case of odd r r . For r = 3 , 4 r=3,\,4 , these quadrature formulae have been constructed by G. Nikolov, who also proved estimates for their error constants. The aim of this note is to refine the estimates for the error constant in the case r = 3 r=3 , and to point out to some error estimates in both cases r = 3 r=3 and r = 4 r=4 , which are easier to evaluate and could be sharper than those which involve the uniform norm of the r r -th derivative of the integrand. Keywords and Phrases: Spline functions, monosplines, Peano representation of linear functionals, definite quadrature formulae, error estimation of quadratures, Bernoulli polynomials. Mathematics Subject Classification 2020: 41A55, 65D30, 65D32. 1. Introduction and statement of the results A standard way to evaluate approximately the definite integral I[f]:=\int\limits_{0}^{1}f(x)\,dx is to use quadrature formulae, which are linear functionals of the form (1.1) Q[f]=\sum_{i=1}^{n}a_{i}f(\tau_{i}),\quad 0\leq\tau_{1}<\cdots<\tau_{n}\leq 1. We start with introducing some notation and definitions. Throughout this paper, \pi_{m} stands for the set of algebraic polynomials of degree not exceeding m . A quadrature formula Q is said to have algebraic degree of precision m (in short, ADP(Q)=m ) if m is the largest non-negative integer such that its remainder functional R[Q;f]:=I[f]-Q[f] vanishes on \pi_{m} . The Sobolev classes of functions W^{r}_{p}[0,1] , \,(r\in\mathbb{N},\ p\geq 1) , are defined by W^{r}_{p}[0,1]:=\{f\in C^{r-1}[0,1]\,:\,f^{(r-1)}\mbox{ loc. abs. cont.},\;\int_{0}^{1}\!|f^{(r)}(t)|^{p}\,dt<\infty\} (note that C^{r}[0,1]\subset W^{r}_{p}[0,1] for every p\geq 1 ). Henceforth, \|\cdot\| designates the supremum norm in [0,1] , and the usual L_{p}[0,1] -norm is shortly denoted by \|\cdot\|_{p} , \|f\|_{p}=\begin{cases}\Big(\int\limits_{0}^{1}|f(t)|^{p}\,dt\Big)^{1/p},&\text{ if }1\leq p<\infty,\\ {\rm vraisup}_{t\in[0,1]}\,|f(t)|,&\text{ if }p=\infty.\end{cases} If ADP(Q)=m\geq r-1 and f\in W^{r}_{1}[0,1] , then by Peano representation theorem for linear functionals (cf. [ 14 ] ), the remainder R[Q;f] can be written in the form (1.2) R[Q;f]=\int\limits_{0}^{1}K_{r}(Q;t)f^{(r)}(t)\,dt, where K_{r}(Q;t) is referred to as the r -th Peano kernel of Q and is given by (1.3) K_{r}(Q;t)=\frac{1}{(r-1)!}\,R\big[Q;(\cdot-t)_{+}^{r-1}\big], where (x)_{+}^{r-1}=\max\{x,0\}^{r-1} is the truncated power function. In literature, K_{r}(Q;t) is also termed as monospline of degree r . For quadrature formula Q in ( 1.1 ) the explicit form of K_{r}(Q;t) , t\in[0,1] , is (1.4) \displaystyle K_{r}(Q;t) \displaystyle=\frac{(1-t)^{r}}{r!}-\frac{1}{(r-1)!}\,\sum_{i=1}^{n}a_{i}(\tau_{i}-t)_{+}^{r-1} (1.5) \displaystyle=(-1)^{r}\Big\{\frac{t^{r}}{r!}-\frac{1}{(r-1)!}\,\sum_{i=1}^{n}a_{i}(t-\tau_{i})_{+}^{r-1}\Big\}\,. If f\in W^{r}_{p}[0,1] , then application of Hölder’s inequality to ( 1.2 ) implies the unimprovable error estimate \big|R[Q;f]\big|\leq c_{r,p}(Q)\,\|f^{(r)}\|_{p},\quad\text{where }c_{r,p}(Q)=\|K_{r}(Q;\cdot)\|_{q},\ \ \frac{1}{p}+\frac{1}{q}=1. Usually, c_{r,p}(Q) is called the error constant of Q in the Sobolev class W^{r}_{p}[0,1] . In what follows, the subscript ” n ” in Q_{n} is used to emphasize that Q_{n} is an n -point quadrature formula, i.e., a quadrature formula which has n nodes. Quadrature formulae Q_{n} with the smallest possible error constant c_{r,p}(Q_{n}) are called optimal quadrature formulae in W^{r}_{p}[0,1] . Without going into details, let us mention that the existence and uniqueness of optimal quadrature formulae in Sobolev classes of functions have been established by Bojanov [ 1 , 2 , 3 ] and Zhensykbaev [ 16 , 17 ] . In the present paper we study certain definite quadrature formulae. A quadrature formula Q_{n} is said to be definite of order r ( r\geq 1 ), if there exists a constant c_{r}(Q_{n})\neq 0 such that R[Q_{n};f]=c_{r}(Q_{n})\,f^{(r)}(\xi) for every f\in C^{r}[0,1] with some \xi\in[0,1] depending on the integrand f . More precisely, Q_{n} is called positive resp. negative definite quadrature formula of order r if c_{r}(Q_{n})>0 resp. c_{r}(Q_{n})<0 . Since c_{r}(Q_{n})=c_{r,\infty}(Q_{n}) if Q_{n} is positive definite and c_{r}(Q_{n})=-c_{r,\infty}(Q_{n}) if Q_{n} is negative definite, c_{r}(Q_{n}) will also be referred to as the error constant of Q_{n} . The importance of definite quadrature formulae of order r stems from the fact that they provide one-sided approximation to I[f] when f^{(r)} has a permanent sign in (0,1) . The midpoint and the trapezium quadrature formulae are best-known examples of positive resp. negative definite quadrature formulae of order two. Definite n -point quadrature formulae of order r with the smallest positive or the largest negative c_{r} are called optimal definite quadrature formulae. It is known that optimal definite quadrature formulae exist and are unique, cf. [ 6 , 11 , 15 ] and [ 5 , Chapter VII.8] . We denote by c_{n,r}^{+} and c_{n,r}^{-} the error constants of the optimal n -point definite quadrature formulae of order r : \displaystyle c_{n,r}^{+}:=\inf\{c_{r}(Q_{n})\;:\;Q_{n}\ \text{ is positive definite of order }r\}, \displaystyle c_{n,r}^{-}:=\sup\{c_{r}(Q_{n})\;:\;Q_{n}\ \text{ is negative definite of order }r\}. In [ 9 ] estimates have been established for the error constants of the Gauss-type quadrature formulae associated with the spaces of spline functions with double equidistant knots. These estimates in turn provide bounds for the error constants of the optimal definite quadrature formulae. Below we restate the main result from [ 9 ] , denoting by B_{r} the Bernoulli polynomial of order r with leading coefficient 1/r! . Theorem A. ( [ 9 ] , Theorem 1.1) (a) For even r with 2\leq r\leq 2n , there holds c_{n,r}^{+}\leq-\frac{B_{r}(j/2)}{(n+1-r/2)^{r}},\quad\textit{ if }\ r=4m+2j,\ j=0,1. (b) For even r with 2\leq r\leq 2n-2 , there holds c_{n,r}^{-}\geq-\frac{B_{r}(j/2)}{(n-r/2)^{r}},\quad\textit{ if }\ r=4m+2-2j,\ j=0,1. (c) For odd r with 1\leq r\leq 2n-1 , there holds c_{n,r}^{+}\leq\frac{\|B_{r}\|}{(n-(r-1)/2)^{r}}\quad\textit{ and }\ \ c_{n,r}^{-}\geq-\frac{\|B_{r}\|}{(n-(r-1)/2)^{r}}. Comparison with results of Lange [ 10 ] shows that Gauss-type quadrature formulae associated with the spaces of spline functions of degree r-1 with double equidistant knots are asymptotically optimal definite quadrature formulae of order r when r is even, and it is conjectured that the asymptotical optimality property persists also in the case of odd r . Two particular cases of Theorem A relevant to the object of this paper are (1.6) \displaystyle c_{n+1,3}^{+}\leq\frac{\sqrt{3}}{216n^{3}},\quad c_{n+1,3}^{-}\geq-\frac{\sqrt{3}}{216n^{3}}, (1.7) \displaystyle c_{n+1,4}^{+}\leq\frac{1}{720n^{4}}\,. The right-hand sides of the inequalities in ( 1.6 ) and ( 1.7 ) are in fact bounds for the error constants of Gauss-type quadrature formulae associated with the linear spaces of spline functions S_{n,3} and S_{n,4} , respectively, where for r\geq 3 and n\geq 2 , (1.8) \begin{split}&S_{n,r}=\{f\;:\;f\in C^{r-3}[0,1],\;f_{|(x_{k},x_{k+1})}\in\pi_{r-1},\ k=0,\ldots,n-1\},\\ &x_{k}=x_{k,n}:=\frac{k}{n},\quad k=0,\ldots,n.\end{split} The functions \{1,x,x^{2},(x-x_{1})_{+},(x-x_{1})_{+}^{2},\ldots,(x-x_{n-1})_{+},(x-x_{n-1})_{+}^{2}\} form a basis for S_{n,3} , therefore \dim S_{n,3}=2n+1 and the Gauss-type quadratures associated with S_{n,3} are left and right (n+1) -point Radau quadrature formulae. These quadrature formulae were found, among others, in [ 12 ] . Theorem B. ( [ 12 , Theorem 2] ) The right Radau quadrature formula associated with the space of parabolic splines S_{n,3} is Q_{n+1}^{R,r}[f]=\sum_{k=0}^{n-1}a_{k,n}f\Big(\frac{k+\theta_{k}}{n}\Big)+a_{n,n}f(1)\approx I[f], where \theta_{0}=1/3 , \displaystyle\theta_{k}=\frac{1-\theta_{k-1}}{5-6\theta_{k-1}},\quad k=1,\ldots,n-1, \displaystyle a_{k,n}=\frac{1}{6n\theta_{k}(1-\theta_{k})},\quad k=0,\ldots,n-1, and a_{n,n}=\frac{2-3\theta_{n-1}}{6n(1-\theta_{n-1})}. Q_{n+1}^{R,r} is negative definite quadrature formula of order three with error constant c_{3}\big(Q_{n+1}^{R,r}\big)=-\frac{\sqrt{3}}{216n^{3}}+O(n^{-4}). Remark 1 . The left Radau quadrature formula Q_{n+1}^{R,l} associated with S_{n,3} is obtained from Q_{n+1}^{R,r} by

Read the complete text