SearcharxivSearch

arXiv subjects

Geno Nikolov

Publications and source records attributed to Geno Nikolov.

At least 19 recordsLinked to original sources

An inequality for Jacobi polynomials: a complement to Finite Increment Theorem

Let $P=P_n^{(\alpha,\beta)}$ be the $n$-th degree Jacobi polynomial, which is orthogonal in $[-1,1]$ with respect to the weight function $(1-x)^{\alpha}(1+x)^{\beta}$, $\alpha,\beta>-1$. For parameters $(\alpha,\beta)$ satisfying either $\alpha\geq\beta\geq 1/2$ or $\alpha\geq 1/2$, $\beta=-1/2$, we prove the inequality $$ P(1)-P(x)\geq P^{\prime}(x)\,(1-x),\quad x\in [0,1], $$ which may be viewed as a complement to Finite Increment Theorem for Jacobi polynomials.

math.CA

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

In two papers from 1995 P. K\"{o}hler 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 $r-1$ with double equispaced 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 $r$. For $r=3,\,4$, 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 $r=3$, and to point out to some error estimates in both cases $r=3$ and $r=4$, which are easier to evaluate and could be sharper than those which involve the uniform norm of the $r$-th derivative of the integrand. \medskip \noindent \textbf{Keywords and Phrases:} Spline functions, monosplines, Peano representation of linear functionals, definite quadrature formulae, error estimation of quadratures, Bernoulli polynomials.\medskip \noindent \textbf{Mathematics Subject Classification 2020:} 41A55, 65D30, 65D32.

math.NA

Simple bounds for the extreme zeroes of Jacobi polynomials

Some new bounds for the extreme zeroes of Jacobi polynomials are obtained with an elementary approach. A feature of these bounds is their simple forms, which make them easy to work with. Despite their simplicity, our lower bounds for the largest zeroes of Gegenbauer polynomials are compatible with some of the best hitherto known results.

math.CA

Modified Trapezoidal Product Cubature Rules. Definiteness, Monotonicity and a Posteriori Error Estimates

We study two modifications of the trapezoidal product cubature formulae, approximating double integrals over the square domain $[a,b]^2=[a,b]\times [a,b]$. Our modified cubature formulae use mixed type data: except evaluations of the integrand on the points forming a uniform grid on $[a,b]^2$, they involve two or four univariate integrals. An useful property of these cubature formulae is that they are definite of order $(2,2)$, that is, they provide one-sided approximation to the double integral for real-valued integrands from the class $$ \mathcal{C}^{2,2}[a,b]=\{f(x,y)\,:\,\frac{\partial^4 f}{\partial x^2\partial y^2}\ \text{continuous and does not change sign in}\ (a,b)^2\}. $$ For integrands from $\mathcal{C}^{2,2}[a,b]$ we prove monotonicity of the remainders and derive a-posteriori error estimates.

math.NA

Estimates for the largest critical value of $T_n^{(k)}$

Here we study the quantity $$ τ_{n,k}:=\frac{|T_n^{(k)}(ω_{n,k})|}{T_n^{(k)}(1)}\,, $$ where $T_n$ is the $n$-th Chebyshev polynomial of the first kind and $ω_{n,k}$ is the largest zero of $T_n^{(k+1)}$. Since the absolute values of the local extrema of $T_n^{(k)}$ increase monotonically towards the end-points of $[-1,1]$, the value $τ_{n,k}$ shows how small is the largest critical value of $\,T_n^{(k)}\,$ relative to its global maximum $\,T_n^{(k)}(1)$. This is a continuation of the recent paper \cite{NNS2018}, where upper bounds and asymptotic formuae for $τ_{n,k}$ have been obtained on the basis of Alexei Shadrin's explicit form of the Schaeffer--Duffin pointwise majorant for polynomials with absolute value not exceeding $1$ in $[-1,1]$. We exploit a result of Knut Petras \cite{KP1996} about the weights of the Gaussian quadrature formulae associated with the ultraspherical weight function $w_λ(x)=(1-x^2)^{λ-1/2}$ to find an explicit (modulo $ω_{n,k}$) formula for $τ_{n,k}^2$. This enables us to prove a lower bound and to refine the upper bounds for $τ_{n,k}$ obtained in \cite{NNS2018}. The explicit formula admits also a new derivation of the assymptotic formula in \cite{NNS2018} approximating $τ_{n,k}$ for $n\to\infty$. The new approach is simpler, without using deep results about the ordinates of the Bessel function, and allows to better analyze the sharpness of the estimates.

math.CA

Hardy's inequalities in finite dimensional Hilbert spaces

We study the behaviour of the smallest possible constants $d_n$ and $c_n$ in Hardy's inequalities $$ \sum_{k=1}^{n}\Big(\frac{1}{k}\sum_{j=1}^{k}a_j\Big)^2\leq d_n\,\sum_{k=1}^{n}a_k^2, \qquad (a_1,\ldots,a_n) \in \mathbb{R}^n $$ and $$ \int_{0}^{\infty}\Bigg(\frac{1}{x}\int\limits_{0}^{x}f(t)\,dt\Bigg)^2 dx \leq c_n \int_{0}^{\infty} f^2(x)\,dx, \ \ f\in \mathcal{H}_n, $$ for the finite dimensional spaces $\mathbb{R}^n$ and $\mathcal{H}_n:=\{f\,:\, \int_0^x f(t) dt =e^{-x/2}\,p(x)\ :\ p\in \mathcal{P}_n, p(0)=0\}$, where $\mathcal{P}_n$ is the set of real-valued algebraic polynomials of degree not exceeding $n$. The constants $d_n$ and $c_n$ are identified as the smallest eigenvalues of certain Jacobi matrices and the two-sided estimates for $d_n$ and $c_n$ of the form $$ 4-\frac{c}{\ln n}< d_n, c_n<4-\frac{c}{\ln^2 n}\,,\qquad c>0\, $$ are established.

math.CA

On the extreme zeros of Jacobi polynomials

By applying the Euler--Rayleigh methods to a specific representation of the Jacobi polynomials as hypergeometric functions, we obtain new bounds for their largest zeros. In particular, we derive upper and lower bound for $1-x_{nn}^2(λ)$, with $x_{nn}(λ)$ being the largest zero of the $n$-th ultraspherical polynomial $P_n^{(λ)}$. For every fixed $λ>-1/2$, the limit of the ratio of our upper and lower bounds for $1-x_{nn}^2(λ)$ does not exceed $1.6$. This paper is a continuation of [1].

math.CA

Some inequalities for Chebyshev polynomials

Askey and Gasper (1976) proved a trigonometric inequality which improves another trigonometric inequality found by M. S. Robertson (1945). Here these inequalities are reformulated in terms of the Chebyshev polynomial of the first kind $T_n$ and then put into a one-parametric family of inequalities. The extreme value of the parameter is found for which these inequalities hold true. As a step towards the proof of this result we establish the following complement to the finite increment theorem specialized to $T_n^{\prime}$: $$ T_n^{\prime}(1)-T_n^{\prime}(x)\geq (1-x)\,T_n^{\prime\prime}(x)\,,\qquad x\in [0,1]\,. $$ By a known expansion formula, this property is extended for the class of ultraspherical polynomials $P_n^{(λ)}$, $λ\geq 1$.

math.CA

Estimates for the best constant in a Markov $L_2$-inequality with the assistance of computer algebra

We prove two-sided estimates for the best (i.e., the smallest possible) constant $\,c_n(α)\,$ in the Markov inequality $$ \|p_n'\|_{w_α} \le c_n(α) \|p_n\|_{w_α}\,, \qquad p_n \in {\cal P}_n\,. $$ Here, ${\cal P}_n$ stands for the set of algebraic polynomials of degree $\le n$, $\,w_α(x) := x^α\,e^{-x}$, $\,α> -1$, is the Laguerre weight function, and $\|\cdot\|_{w_α}$ is the associated $L_2$-norm, $$ \|f\|_{w_α} = \left(\int_{0}^{\infty} |f(x)|^2 w_α(x)\,dx\right)^{1/2}\,. $$ Our approach is based on the fact that $\,c_n^{-2}(α)\,$ equals the smallest zero of a polynomial $\,Q_n$, orthogonal with respect to a measure supported on the positive axis and defined by an explicit three-term recurrence relation. We employ computer algebra to evaluate the seven lowest degree coefficients of $\,Q_n\,$ and to obtain thereby bounds for $\,c_n(α)$. This work is a continuation of a recent paper [5], where estimates for $\,c_n(α)\,$ were proven on the basis of the four lowest degree coefficients of $\,Q_n$.

math.CA

On the largest critical value of $T_n^{(k)}$

We study the quantity $$ τ_{n,k}:=\frac{|T_n^{(k)}(ω_{n,k})|}{T_n^{(k)}(1)}\,, $$ where $T_n$ is the Chebyshev polynomial of degree $n$, and $ω_{n,k}$ is the rightmost zero of $T_n^{(k+1)}$. Since the absolute values of the local maxima of $T_n^{(k)}$ increase monotonically towards the end-points of $[-1,1]$, the value $τ_{n,k}$ shows how small is the largest critical value of $\,T_n^{(k)}\,$ relative to its global maximum $\,T_n^{(k)}(1)$. In this paper, we improve and extend earlier estimates by Erdős--Szegő, Eriksson and Nikolov in several directions. Firstly, we show that the sequence $\,\{τ_{n,k}\}_{n=k+2}^{\infty}$ is monotonically decreasing in $n$, hence derive several sharp estimates, in particular $$ τ_{n,k} \le \begin{cases} τ_{k+4,k} = \frac{1}{2k+1}\,\frac{3}{k+3}\,, & n \ge k+4\, τ_{k+6,k} = \frac{1}{2k+1}\, (\frac{5}{k+5})^2 β_k\,, & n \ge k+6\,, \end{cases} $$ where $β_k < \frac{2+\sqrt{10}}{5} \approx 1.032$. We also obtain an upper bound which is uniform in $n$ and $k$, and that implies in particular $$ τ_{n,k} \approx \big(\frac{2}{e}\big)^k, \quad n \ge k^{3/2}; \qquad τ_{n,n-m} \approx \big(\frac{em}{2}\big)^{m/2} n^{-m/2}; \qquad τ_{n,n/2} \approx \big(\frac{4}{\sqrt{27}}\big)^{n/2}. $$ Finally, we derive the exact asymptotic formulae for the quantities $$ τ_k^{*} := \lim_{n\to\infty}τ_{n,k} \quad \mbox{ and }\quad τ_m^{**} := \lim_{n\to\infty} n^{m/2} τ_{n,n-m}\,, $$ which show that our upper bounds for $τ_{n,k}$ and $τ_{n,n-m}$ are asymptotically correct with respect to the exponential terms given above.

math.CA

Markov $L_2$-inequality with the Laguerre weight

Let $w_α(t) := t^α\,e^{-t}$, where $α> -1$, be the Laguerre weight function, and let $\|\cdot\|_{w_α}$ be the associated $L_2$-norm, $$ \|f\|_{w_α} = \left\{\int_{0}^{\infty} |f(x)|^2 w_α(x)\,dx\right\}^{1/2}\,. $$ By $\mathcal{P}_n$ we denote the set of algebraic polynomials of degree $\le n$. We study the best constant $c_n(α)$ in the Markov inequality in this norm $$ \|p_n'\|_{w_α} \le c_n(α) \|p_n\|_{w_α}\,,\qquad p_n \in \mathcal{P}_n\,, $$ namely the constant $$ c_n(α) := \sup_{p_n \in \mathcal{P}_n} \frac{\|p_n'\|_{w_α}}{\|p_n\|_{w_α}}\,. $$ We derive explicit lower and upper bounds for the Markov constant $c_n(α)$, as well as for the asymptotic Markov constant $$ c(α)=\lim_{n\rightarrow\infty}\frac{c_n(α)}{n}\,. $$

math.CA

Markov $L_2$ inequality with the Gegenbauer weight

For the Gegenbauer weight function $w_λ(t)=(1-t^2)^{λ-1/2}$, $λ>-1/2$, we denote by $\Vert\cdot\Vert_{w_λ}$ the associated $L_2$-norm, $$ \Vert f\Vert_{w_λ}:=\Big(\int_{-1}^{1}w_λ(t)f^2(t)\,dt\Big)^{1/2}. $$ We study the Markov inequality $$ \Vert p^{\prime}\Vert_{w_λ}\leq c_{n}(λ)\,\Vert p\Vert_{w_λ},\qquad p\in \mathcal{P}_n, $$ where $\mathcal{P}_n$ is the class of algebraic polynomials of degree not exceeding $n$. Upper and lower bounds for the best Markov constant $c_{n}(λ)$ are obtained, which are valid for all $n\in \mathbb{N}$ and $λ>-\frac{1}{2}$.

math.CA

On the Markov inequality in the $L_2$-norm with the Gegenbauer weight

Let $w_λ(t) := (1-t^2)^{λ-1/2}$, where $λ> -\frac{1}{2}$, be the Gegenbauer weight function, let $\|\cdot\|_{w_λ}$ be the associated $L_2$-norm, $$ \|f\|_{w_λ} = \left\{\int_{-1}^1 |f(x)|^2 w_λ(x)\,dx\right\}^{1/2}\,, $$ and denote by $\mathcal{P}_n$ the space of algebraic polynomials of degree $\le n$. We study the best constant $c_n(λ)$ in the Markov inequality in this norm $$ \|p_n'\|_{w_λ} \le c_n(λ) \|p_n\|_{w_λ}\,,\qquad p_n \in \mathcal{P}_n\,, $$ namely the constant $$ c_n(λ) := \sup_{p_n \in \mathcal{P}_n} \frac{\|p_n'\|_{w_λ}}{\|p_n\|_{w_λ}}\,. $$ We derive explicit lower and upper bounds for the Markov constant $c_n(λ)$, which are valid for all $n$ and $λ$.

math.CA

On the $L_2$ Markov Inequality with Laguerre Weight

Let $w_α(t)=t^α\,e^{-t}$, $α>-1$, be the Laguerre weight function, and $|\cdot|_{w_α}$ denote the associated $L_2$-norm, i.e., $$ | f|_{w_α}:=\Big(\int_{0}^{\infty}w_α(t)| f(t)|^2\,dt\Big)^{1/2}. $$ Denote by ${\cal P}_n$ the set of algebraic polynomials of degree not exceeding $n$. We study the best constant $c_n(α)$ in the Markov inequality in this norm, $$ | p^{\prime}|_{w_α}\leq c_n(α)\,| p|_{w_α}\,,\quad p\in {\cal P}_n\,, $$ namely the constant $$ c_{n}(α)=\sup_{\mathop{}^{p\in {\cal P}_n}_{p\ne 0}}\frac{| p^{\prime}|_{w_α}}{| p|_{w_α}}\,, $$ and we are also interested in its asymptotic value $$ c(α)=\lim_{n\rightarrow\infty}\frac{c_{n}(α)}{n}\,. $$ In this paper we obtain lower and upper bounds for both $c_{n}(α)$ and $c(α)$. % Note that according to a result of P. Dörfler from 2002, $c(α)=[j_{(α-1)/2,1}]^{-1}$, with $j_{ν,1}$ being the first positive zero of the Bessel function $J_ν(z)$, hence our bounds for $c(α)$ imply bounds for $j_{(α-1)/2,1}$ as well.

math.CA

Asymptotically optimal definite quadrature formulae of $4$-th order

We construct several sequences of asymptotically optimal definite quadrature formulae of fourth order and evaluate their error constants. Besides the asymptotical optimality, an advantage of our quadrature formulae is the explicit form of their weights and nodes. For the remainders of our quadrature formulae monotonicity properties are established when the integrand is a 4-convex function, and a-posteriori error estimates are proven.

math.NA

On the Markov inequality in the $L_2$-norm with Gegenbauer weight

Let $w_λ(t)=(1-t^2)^{λ-1/2}$, $λ>-1/2$, be the Gegenbauer weight function, and $\Vert\cdot\Vert$ denote the associated $L_2$-norm, i.e., $$ \Vert f\Vert:=\Big(\int_{-1}^{1}w_λ(t)\vert f(t)\vert^2\,dt\Big)^{1/2}. $$ Denote by $\mathcal{P}_n$ the set of algebraic polynomials of degree not exceeding $n$. We study the best (i.e., the smallest) constant $c_{n,λ}$ in the Markov inequality $$ \Vert p^{\prime}\Vert\leq c_{n,λ}\,\Vert p\Vert,\qquad p\in \mathcal{P}_n, $$ and prove that $$ c_{n,λ}< \frac{(n+1)(n+2λ+1)}{2\sqrt{2λ+1}},\qquad λ>-1/2\,. $$ Moreover, we prove that the extremal polynomial in this inequality is even or odd depending on whether $n$ is even or odd.

math.CA

Inequalities for ultraspherical polynomials. Proof of a conjecture of I. Raşa

A recent conjecture by I. Raşa asserts that the sum of the squared Bernstein basis polynomials is a convex function in $[0,1]$. This conjecture turns out to be equivalent to a certain upper pointwise estimate of the ratio $P_n^{\prime}(x)/P_n(x)$ for $x\geq 1$, where $P_n$ is the $n$-th Legendre polynomial. Here, we prove both upper and lower pointwise estimates for the ratios $\big(P_n^{(λ)}(x)\big)^{\prime}/P_n^{(λ)}(x)$, $~x\geq 1$, where $P_n^{(λ)}$ is the $n$-th ultraspherical polynomial. In particular, we validate Raşa's conjecture.

math.CA