$\def\ket#1{\left|#1\right\rangle } \def\bra#1{\left\langle #1\right|}$ (This is part of a research problem)

The Setup: Consider a single particle on a finite 1D lattice with the Hilbert space spanned by the orthonormal position basis $\ket{n;x}$ for $n=0,1,...,D-1$. We define the lattice momentum basis by applying the lattice Fourier transform $$\ket{m;p}:=F\ket{m;x}:=\frac{1}{\sqrt{D}}\sum_{n=0}^{D-1}e^{i2\pi mn/D}\ket{n;x}.$$ We can then subdivide the lattice (in position space) into equidistant intervals of length $\omega$ and restrict the momentum basis to these position-space intervals, resulting in the definition of truncated momentum states $$\ket{m,k;p}:=\frac{1}{\sqrt{\omega}}\sum_{n\in\Omega_{k}}e^{i2\pi mn/D}\ket{n;x}$$ where $\Omega_{k}:=\left\{ \omega k,\omega k+1,..., \omega k+\omega-1 \right\} $ are the position indices of the $k$th interval.

We now focus on the subspace of a single interval spanned by $\left\{\ket{n;x}\right\}_{n\in\Omega_{0}}$. The truncated momentum basis are not pairwise orthogonal on this subspace as we can calculate $$\bra{m',0;p}\ket{m,0;p} =\frac{1}{\omega}\sum_{n\in\Omega_{0}}e^{i2\pi n\left(m-m'\right)/D}=\frac{1}{\omega}\frac{e^{i2\pi\left(m-m'\right)\omega/D}-1}{e^{i2\pi\left(m-m'\right)/D}-1}.$$

The Problem: From a combination of analysis and numerics I have reasons to believe that for $\omega>>1$ the following frame operator $S$ converges to a (unnormalized) projection $$S:=\sum_{m\in\Omega_{0}}\ket{m,0;p}\bra{m,0;p}\approx \frac{D}{\omega} \sum_{m=0...\gamma-1}\ket{m;p_{0}}\bra{m;p_{0}}$$ where $\gamma=\frac{\omega^2}{D}$ and I have introduced the native (orthonormal) momentum basis of the interval $$\ket{m;p_{0}}:=\frac{1}{\sqrt{\omega}}\sum_{n\in\Omega_{0}}e^{i2\pi mn/\omega}\ket{n;x}.$$ What would be a good strategy to argue / prove that the frame operator $S$ indeed converges to that projection?


  1. I am not showing my reasons to believe in the convergence because it is messy and long. The problem is about finding an effective line of reasoning.
  2. I believe that this problem should be known in the context of frame theory and signal processing. Any relevant results in that area are welcome.
  3. Since this is a research problem, I'm not expecting a solution. Any thoughts on how would one approach this problem will be greatly appreciated (but who knows, I won't be too surprised to see a solution from people in this community).

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.