Take the 2-minute tour ×
MathOverflow is a question and answer site for professional mathematicians. It's 100% free, no registration required.

Consider a grid of points $T=\{t_0,t_1,\ldots,t_m\}$ with $0\le t_i\le 1$. I would like to find a function $f(t):[0,1]\rightarrow \mathbb{C}$ of the form \begin{equation*} f(t)=\sum_{k=-n}^n c_k e^{2\pi k it} \end{equation*}

(with n as small as possible) such that

1) $f(t_0)=1$, and $f(t_k)=0$ for $k\ge1$.

2) $|f(t)|\le \frac{c}{1+|t-t_0|^2}$. That is, $f(t)$ decays quickly away from $t_0$. Here, the distance $|t-t_0|$ is a circular distance ($|0.1-0.9|=0.2$).

I know that a condition on the interpolation nodes $t_0,t_1,\ldots,t_m$ like the ones appearing below is required

1) smallest constant $c$ such that

$min_{k,\ell}|t_k-t_\ell|\ge c/n$

suffices, (with the distance meant to be circular that is |0.9-0.1|=0.2).

or

2) more sophisticated conditions like:

$D_{m+1}(t_0,t_1,\ldots,t_m)$ needs to be small. The discrepancy of a a finite sequence of real numbers $x_1,x_2,\ldots,x_N\in[0,1]$ is defined as \begin{equation*} D_N(x_1,x_2,\ldots,x_N)=\underset{0\le\alpha<\beta\le 1}{sup}\bigg|\frac{A([\alpha,\beta);N)}{N}-(\beta-\alpha)\bigg|, \end{equation*} with $A([\alpha,\beta);N)$ denoting the number of $x_i$ such that $x_i\in[\alpha,\beta)$ (Based on section 2 of Uniform Distribution of Sequences by Kuipers and Niederreiter).

share|improve this question

1 Answer 1

up vote 3 down vote accepted

Some initial thoughts to get us started:

Consider $$ g(z) = \bigg ( \frac{\sin(2\pi(z - t_0))}{2\pi(z - t_0)} \cdot \prod_{k \neq 0} \frac{\sin(2\pi(z - t_k))}{\sin(2\pi(t_0 - t_k))} \bigg )^2 $$ Note that the Fourier transform of $g$ is compactly supported (by Paley-Wiener). Now define $$ h(x) := \sum_{n \in \mathbb{Z}} g(x+n) $$ Note that $h(t_0)=1$ and $h(t_k)=0$. Now apply Poisson summation and get $$ h(x) = \sum_{n \in \mathbb{Z}} \widehat{g}(n) e(x n) $$ Note that this is a finite sum because $\widehat{g}$ is compactly supported. Now also notice that $h$ decays like $\ll 1/(1+||z - t_0||)^2$ where $|| \cdot ||$ denotes your circular distance (I haven't checked carefully). The decay can be increased to an arbitrary power by taking higher powers in the definition of $g$ but the price to pay is larger support of $\widehat{g}$. Finding the smallest support of $\widehat{g}$ could possibly be a research level problem. I would suggest looking at Valeer survey on extremal functions and/or Chapter 1 of Montgomery's book "Ten lectures on the interface between analytic number theory and harmonic analysis".

EDIT: Try also looking for Levin's book "Lectures on entire functions". There is a chapter on interpolation and sine-like functions. This will be useful for another question you asked about the product over the $t_k$'s. Also take a look at some paper of Boas, since he was also concerned with product of the type that appear above (the product over $t_k$).

share|improve this answer
    
Thanks for your answer and references. I was actually thinking in a similar way this is was why I asked this question mathoverflow.net/questions/134762/…. I wanted to then multiple by a low-pass fast decaying kernel and use a similar trick to yours to get the answer here. The main issue is that I still need to bound the absolute value of the products of sines subject to a distance or discrepancy condition above. Any suggestion on how to do this? Please let me know of any other references that you think might be related. –  mohi Jun 27 '13 at 20:54

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.