摘要
The Sobolev space HS(Rd) with s 〉 d/2 contains many important functions such as the bandlimited or rational ones. In this paper we propose a sequence of measurement functions { φj^r,k}∈C H^-S(R^d) to the phase retrieval problem for the real-valued functions in H^s(R^d). We prove that any real-valued function f ∈ H^s (Rd) can be determined, up to a global sign, by the phaseless measurements {|( f, φj^r,k}|}. It is known that phase retrieval is unstable in infinite dimensional spaces with respect to perturbations of the measurement functions. We examine a special type of perturbations that ensures the stability for the phase-retrieval problem for all the real-valued functions in Hs(Rd) ∩ C1(Rd), and prove that our iterated reconstruction procedure guarantees uniform convergence for any function f ∈ Hs (Rd)∩ C1 (Rd) whose Fourier transform f is L1-integrable. Moreover, numerical simulations are conducted to test the efficiency of the reconstruction algorithm.
The Sobolev space HS(Rd) with s 〉 d/2 contains many important functions such as the bandlimited or rational ones. In this paper we propose a sequence of measurement functions { φj^r,k}∈C H^-S(R^d) to the phase retrieval problem for the real-valued functions in H^s(R^d). We prove that any real-valued function f ∈ H^s (Rd) can be determined, up to a global sign, by the phaseless measurements {|( f, φj^r,k}|}. It is known that phase retrieval is unstable in infinite dimensional spaces with respect to perturbations of the measurement functions. We examine a special type of perturbations that ensures the stability for the phase-retrieval problem for all the real-valued functions in Hs(Rd) ∩ C1(Rd), and prove that our iterated reconstruction procedure guarantees uniform convergence for any function f ∈ Hs (Rd)∩ C1 (Rd) whose Fourier transform f is L1-integrable. Moreover, numerical simulations are conducted to test the efficiency of the reconstruction algorithm.
基金
supported by Natural Science Foundation of China(Grant Nos.61561006 and11501132)
Natural Science Foundation of Guangxi(Grant No.2016GXNSFAA380049)
the support from NSF under the(Grant Nos.DMS-1403400 and DMS-1712602)