Abstract:Atmospheric tomography, the problem of reconstructing atmospheric turbulence profiles from wavefront sensor measurements, is an integral part of many adaptive optics systems used for enhancing the image quality of ground-based telescopes. Singular-value and frame decompositions of the underlying atmospheric tomography operator can reveal useful analytical information on this inverse problem, as well as serve as the basis of efficient numerical reconstruction algorithms. In this paper, we extend existing singular value decompositions to more realistic Sobolev settings including weighted inner products, and derive an explicit representation of a frame-based (approximate) solution operator. These investigations form the basis of efficient numerical solution methods, which we analyze via numerical simulations for the challenging, real-world Adaptive Optics system of the Extremely Large Telescope using the entirely MATLAB-based simulation tool MOST.
What problem does this paper attempt to address?
This paper attempts to solve the problem of atmospheric turbulence profile reconstruction, specifically by reconstructing the atmospheric turbulence profile from wave - front sensor measurement data. Atmospheric turbulence can cause wave - front distortion of the light received by ground - based telescopes, thus affecting the imaging quality. The Adaptive Optics (AO) system is used to correct these wave - front distortions to improve image quality. However, in the case of multiple guide stars (including Natural Guide Stars (NGS) and Laser Guide Stars (LGS)), how to accurately reconstruct the entire atmospheric turbulence volume from wave - front sensor measurements is a complex and challenging problem.
### Main problem description
1. **Atmospheric turbulence reconstruction problem**:
- **Background**: Temperature fluctuations in the atmosphere lead to turbulence, which causes distortion of the incident light wave - front. If not corrected, these distortions will lead to image blurring and seriously degrade the image quality.
- **Objective**: Reconstruct the turbulence profile in the atmosphere through wave - front sensor measurements for real - time correction using a Deformable Mirror (DM).
2. **Limitations of existing methods**:
- Although existing reconstruction methods have been studied in detail numerically and theoretically, they do not provide new insights into the atmospheric turbulence reconstruction problem itself.
- Traditional methods such as the Least Mean Square Error method, the Back - Projection algorithm, and the Conjugate Gradient Iterative method, although effective, have limitations in practical applications, especially when dealing with multiple guide stars.
3. **New research directions**:
- The paper aims to provide a more in - depth theoretical analysis and efficient numerical solutions for the atmospheric turbulence reconstruction problem by introducing Singular Value Decomposition (SVD) and Frame Decomposition (FD).
- In particular, the paper extends the existing Singular Value Decomposition method to a more realistic Sobolev space setting and considers weighted inner products to better simulate the characteristics of atmospheric turbulence.
- At the same time, the paper derives an explicit representation of the frame decomposition, enabling the approximate solution operator to be implemented efficiently.
### Markdown representation of formulas
- **Definition of Sobolev space**:
\[
\langle u, v \rangle_{H^s(c_\ell \Omega_T, \gamma_\ell)}=\sum_{j,k\in\mathbb{Z}}\left(1 + \beta_{\ell,T}|(j, k)|^2\right)^s\langle u, w_{jk,\ell}\rangle_{L^2(c_\ell \Omega_T, \gamma_\ell)}\langle v, w_{jk,\ell}\rangle_{L^2(c_\ell \Omega_T, \gamma_\ell)}
\]
where \(\beta_{\ell,T}=\frac{\pi^2}{(c_\ell T)^2}\).
- **Periodic atmospheric reconstruction operator**:
\[
\tilde{A}^{(s)}:\bigoplus_{\ell = 1}^L H^s(c_\ell \Omega_T, \gamma_\ell)\to L^2(\Omega_T)^G,\quad\phi = (\phi_\ell)_{\ell = 1}^L\mapsto\varphi = ((\tilde{A}_g^{(s)}\phi))_{g = 1}^G
\]
where
\[
(\tilde{A}_g^{(s)}\phi)(r)=\sum_{\ell = 1}^L\phi_\ell(c_\ell r+\alpha_g h_\ell)
\]
- **Dual frame functions in frame decomposition**:
\[
\psi_{jk,\ell}=\left(1+\beta_{\ell,T}|(j, k)|