arXiv subjects

Alexei Shadrin

Publications and source records attributed to Alexei Shadrin.

11 recordsLinked to original sources

arXiv2021

Fast and stable approximation of analytic functions from equispaced samples via polynomial frames

We consider approximating analytic functions on the interval $[-1,1]$ from their values at a set of $m+1$ equispaced nodes. A result of Platte, Trefethen \& Kuijlaars states that fast and stable approximation from equispaced samples is generally impossible. In particular, any method that converges exponentially fast must also be exponentially ill-conditioned. We prove a positive counterpart to this `impossibility' theorem. Our `possibility' theorem shows that there is a well-conditioned method that provides exponential decay of the error down to a finite, but user-controlled tolerance $\epsilon > 0$, which in practice can be chosen close to machine epsilon. The method is known as \textit{polynomial frame} approximation or \textit{polynomial extensions}. It uses algebraic polynomials of degree $n$ on an extended interval $[-\gamma,\gamma]$, $\gamma > 1$, to construct an approximation on $[-1,1]$ via a SVD-regularized least-squares fit. A key step in the proof of our main theorem is a new result on the maximal behaviour of a polynomial of degree $n$ on $[-1,1]$ that is simultaneously bounded by one at a set of $m+1$ equispaced nodes in $[-1,1]$ and $1/\epsilon$ on the extended interval $[-\gamma,\gamma]$. We show that linear oversampling, i.e., $m = c n \log(1/\epsilon) / \sqrt{\gamma^2-1}$, is sufficient for uniform boundedness of any such polynomial on $[-1,1]$. This result aside, we also prove an extended impossibility theorem, which shows that such a possibility theorem (and consequently the method of polynomial frame approximation) is essentially optimal.

math.NA↗

arXiv2017

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

We study the quantity $$ \tau_{n,k}:=\frac{|T_n^{(k)}(\omega_{n,k})|}{T_n^{(k)}(1)}\,, $$ where $T_n$ is the Chebyshev polynomial of degree $n$, and $\omega_{n,k}$ 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 $\tau_{n,k}$ 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 Erd\H{o}s--Szeg\H{o}, Eriksson and Nikolov in several directions. Firstly, we show that the sequence $\,\{\tau_{n,k}\}_{n=k+2}^{\infty}$ is monotonically decreasing in $n$, hence derive several sharp estimates, in particular $$ \tau_{n,k} \le \begin{cases} \tau_{k+4,k} = \frac{1}{2k+1}\,\frac{3}{k+3}\,, & n \ge k+4\, \tau_{k+6,k} = \frac{1}{2k+1}\, (\frac{5}{k+5})^2 \beta_k\,, & n \ge k+6\,, \end{cases} $$ where $\beta_k < \frac{2+\sqrt{10}}{5} \approx 1.032$. We also obtain an upper bound which is uniform in $n$ and $k$, and that implies in particular $$ \tau_{n,k} \approx \big(\frac{2}{e}\big)^k, \quad n \ge k^{3/2}; \qquad \tau_{n,n-m} \approx \big(\frac{em}{2}\big)^{m/2} n^{-m/2}; \qquad \tau_{n,n/2} \approx \big(\frac{4}{\sqrt{27}}\big)^{n/2}. $$ Finally, we derive the exact asymptotic formulae for the quantities $$ \tau_k^{*} := \lim_{n\to\infty}\tau_{n,k} \quad \mbox{ and }\quad \tau_m^{**} := \lim_{n\to\infty} n^{m/2} \tau_{n,n-m}\,, $$ which show that our upper bounds for $\tau_{n,k}$ and $\tau_{n,n-m}$ are asymptotically correct with respect to the exponential terms given above.

math.CA↗

arXiv2017

Markov $L_2$-inequality with the Laguerre weight

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

math.CA↗

arXiv2017

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

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

math.CA↗

arXiv2016

Optimal sampling rates for approximating analytic functions from pointwise samples

We consider the problem of approximating an analytic function on a compact interval from its values at $M+1$ distinct points. When the points are equispaced, a recent result (the so-called impossibility theorem) has shown that the best possible convergence rate of a stable method is root-exponential in $M$, and that any method with faster exponential convergence must also be exponentially ill-conditioned at a certain rate. This result hinges on a classical theorem of Coppersmith & Rivlin concerning the maximal behaviour of polynomials bounded on an equispaced grid. In this paper, we first generalize this theorem to arbitrary point distributions. We then present an extension of the impossibility theorem valid for general nonequispaced points, and apply it to the case of points that are equidistributed with respect to (modified) Jacobi weight functions. This leads to a necessary sampling rate for stable approximation from such points. We prove that this rate is also sufficient, and therefore exactly quantify (up to constants) the precise sampling rate for approximating analytic functions from such node distributions with stable methods. Numerical results -- based on computing the maximal polynomial via a variant of the classical Remez algorithm -- confirm our main theorems. Finally, we discuss the implications of our results for polynomial least-squares approximations. In particular, we theoretically confirm the well-known heuristic that stable least-squares approximation using polynomials of degree $N < M$ is possible only once $M$ is sufficiently large for there to be a subset of $N$ of the nodes that mimic the behaviour of the $N$th set of Chebyshev nodes.

math.NA↗

arXiv2016

On the $L_2$ Markov Inequality with Laguerre Weight

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

math.CA↗

arXiv2015

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

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

math.CA↗

arXiv2013

On almost everywhere convergence of orthogonal spline projections with arbitrary knots

The main result of this paper is a proof that, for any $f \in L_1[a,b]$, a sequence of its orthogonal projections $(P_{\Delta_n}(f))$ onto splines of order $k$ with arbitrary knots $\Delta_n$, converges almost everywhere provided that the mesh diameter $|\Delta_n|$ tends to zero, namely \[ f \in L_1[a,b] \Rightarrow P_{\Delta_n}(f,x) \to f(x) \quad \mbox{a.e.} \quad (|\Delta_n|\to 0)\,. \] This extends the earlier result that, for $f \in L_p$, we have convergence $P_{\Delta_n}(f) \to f$ in the $L_p$-norm for $1 \le p \le \infty$.}

math.FA↗

arXiv2012

Landau--Kolmogorov inequality revisited

The Landau-Kolmogorov problem consists of finding the upper bound $M_k$ for the norm of intermediate derivative $|f^{(k)}|$, when the bounds $|f| \le M_0$ and $|f^{(n)}| \le M_n$, for the norms of the function and of its higher derivative, are given. Here, we consider the case of a finite interval, and when all the norms are the max-norms. Our interest to that particular case is motivated by the fact that there are good chances to add this case to a short list of Landau--Kolmogorov inequalities where a complete solution exists, i.e., a solution that covers all values of $n,k\in\N$ (and, for a finite interval, all values of $\sigma = M_n/M_0$). The main guideline here is Karlin's conjecture that says that, for all $n,k\in\N$ and all $\sigma>0$, the maximum of $|f^{(k)}|$ is attained by a certain Chebyshev or Zolotarev spline. So far, it has been proved only for small $n \ge 4$ with all $\sigma$, and for all $n$ with particular $\sigma = \sigma_n$. Here, we prove Karlin's conjecture in several further subcases: 1) all $n,k\in\N$ and all $0 < \sigma \le \sigma_n$ 2) all $n \in \N$, all $\sigma > 0$, with $k=1,2$ 3) all $\sigma > 0$, with $n < 10$ and $0 < k < n$.

math.NA↗

arXiv2012

On Markov-Duffin-Schaeffer inequalities with a majorant. II

We are continuing out studies of the so-called Markov inequalities with a majorant. Inequalities of this type provide a bound for the $k$-th derivative of an algebraic polynomial when the latter is bounded by a certain curved majorant $\mu$. A conjecture is that the upper bound is attained by the so-called snake-polynomial which oscillates most between $\pm \mu$, but it turned out to be a rather difficult question. In the previous paper, we proved that this is true in the case of symmetric majorant provided the snake-polynomial has a positive Chebyshev expansion. In this paper, we show that that the conjecture is valid under the condition of positive expansion only, hence for non-symmetric majorants as well.

math.NA↗

arXiv2012

A stability barrier for reconstructions from Fourier samples

We prove that any stable method for resolving the Gibbs phenomenon - that is, recovering high-order accuracy from the first $m$ Fourier coefficients of an analytic and nonperiodic function - can converge at best root-exponentially fast in $m$. Any method with faster convergence must also be unstable, and in particular, exponential convergence implies exponential ill-conditioning. This result is analogous to a recent theorem of Platte, Trefethen & Kuijlaars concerning recovery from pointwise function values on an equispaced $m$-grid. The main step in our proof is an estimate for the maximal behaviour of a polynomial of degree $n$ with bounded $m$-term Fourier series, which is related to a conjecture of Hrycak & Groechenig. In the second part of the paper we discuss the implications of our main theorem to polynomial-based interpolation and least-squares approaches for overcoming the Gibbs phenomenon. Finally, we consider the use of so-called Fourier extensions as an attractive alternative for this problem. We present numerical results demonstrating rapid convergence in a stable manner.

math.NA↗