Stability in phase retrieval: Characterizing condition numbers and the optimal vector set
Mathematics of Computation · 14.11.2024 · 94 (356) · 2931-2960
DOI: 10.1090/mcom/4042 ↗Kurz gefasst
Fachartikel in der Zeitschrift Mathematics of Computation, online erschienen am 14.11.2024. Verfasst von Yu Xia und 2 weiteren Autor:innen. Die Arbeit wurde bisher 6-mal zitiert (Crossref) und nennt 33 Literaturangaben. Gefördert von: National Science Fund for Distinguished Young Scholars, National Natural Science Foundation of China.
Abstract (Originalsprache)
In this paper, we primarily focus on analyzing the stability property of phase retrieval by examining the bi-Lipschitz property of the map Φ A ( x ) = | A x | ∈ R + m \Phi _{\boldsymbol {A}}(\boldsymbol {x})=|\boldsymbol {A}\boldsymbol {x}|\in \mathbb {R}_+^m , where x ∈ H d \boldsymbol {x}\in \mathbb {H}^d and A ∈ H m × d \boldsymbol {A}\in \mathbb {H}^{m\times d} is the measurement matrix for H ∈ { R , C } \mathbb {H}\in \{\mathbb {R},\mathbb {C}\} . We define the condition number β A = U A L A \beta _{\boldsymbol {A}}=\frac {U_{\boldsymbol {A}}}{L_{\boldsymbol {A}}} , where L A L_{\boldsymbol {A}} and U A U_{\boldsymbol {A}} represent the optimal lower and upper Lipschitz constants, respectively. We establish the universal lower bound on β A \beta _{\boldsymbol {A}} by demonstrating that for any A ∈ H m × d {\boldsymbol {A}}\in \mathbb {H}^{m\times d} , β A ≥ β 0 H = { π π − 2 ≈ 1.659 a m p ; if H = R , 4 4 − π ≈ 2.159 a m p ; if H = C . \begin{equation*} \beta _{\boldsymbol {A}}\geq \beta _0^{\mathbb {H}}= \begin {cases} \sqrt {\frac {\pi }{\pi -2}}\approx 1.659 & \text {if $\mathbb {H}=\mathbb {R}$,}\\ \sqrt {\frac {4}{4-\pi }}\approx 2.159 & \text {if $\mathbb {H}=\mathbb {C}$.} \end{cases} \end{equation*} We prove that the condition number of a standard Gaussian matrix in H m × d \mathbb {H}^{m\times d} asymptotically matches the lower bound β 0 H \beta _0^{\mathbb {H}} for both real and complex cases. This result indicates that the constant lower bound β 0 H \beta _0^{\mathbb {H}} is asymptotically tight, holding true for both the real and complex scenarios. As an application of this result, we utilize it to investigate the performance of quadratic models for phase retrieval. Lastly, we establish that for any odd integer m ≥ 3 m\geq 3 , the harmonic frame A ∈ R m × 2 \boldsymbol {A}\in \mathbb {R}^{m\times 2} possesses the minimum condition number among all A ∈ R m × 2 \boldsymbol {A}\in \mathbb {R}^{m\times 2} . To the best of our knowledge, our findings provide the first universal lower bound for the condition number in phase retrieval. Additionally, we have identified the first optimal vector set in R 2 \mathbb {R}^2 for phase retrieval. We are confident that these findings carry substantial implications for enhancing our understanding of phase retrieval.
Förderung
- National Science Fund for Distinguished Young Scholars
- National Natural Science Foundation of China
Zitieren
APA
Xia, Y., Xu, Z., & Xu, Z. (2024). Stability in phase retrieval: Characterizing condition numbers and the optimal vector set. Mathematics of Computation, 94(356), 2931-2960. https://doi.org/10.1090/mcom/4042
Vancouver
Xia Y, Xu Z, Xu Z. Stability in phase retrieval: Characterizing condition numbers and the optimal vector set. Mathematics of Computation. 2024;94(356):2931-2960. doi:10.1090/mcom/4042
BibTeX
@article{Xia2024,
author = {Xia, Yu and Xu, Zhiqiang and Xu, Zili},
title = {Stability in phase retrieval: Characterizing condition numbers and the optimal vector set},
journal = {Mathematics of Computation},
year = {2024},
volume = {94},
number = {356},
pages = {2931--2960},
doi = {10.1090/mcom/4042}
}
EUVIMED