文摘
The many variants of the restricted isometry property (RIP) have proven to be crucial theoretical tools in the fields of compressed sensing and matrix completion. The study of extending compressed sensing to accommodate phaseless measurements naturally motivates a strong notion of restricted isometry property (SRIP), which we develop in this paper. We show that if rc">formulatext stixSupport mathImg" data-mathURL="/science?_ob=MathURL&_method=retrieve&_eid=1-s2.0-S1063520315000901&_mathId=si1.gif&_user=111111111&_pii=S1063520315000901&_rdoc=1&_issn=10635203&md5=9e845d43a7f1a13dc052e4ab51b3703b" title="Click to view the MathML source">A∈Rm×nr hidden"> satisfies SRIP and phaseless measurements rc">formulatext stixSupport mathImg" data-mathURL="/science?_ob=MathURL&_method=retrieve&_eid=1-s2.0-S1063520315000901&_mathId=si2.gif&_user=111111111&_pii=S1063520315000901&_rdoc=1&_issn=10635203&md5=8d96355d3d4e21cc96aef2869015f81c" title="Click to view the MathML source">|Ax0|=br hidden"> are observed about a k -sparse signal rc">formulatext stixSupport mathImg" data-mathURL="/science?_ob=MathURL&_method=retrieve&_eid=1-s2.0-S1063520315000901&_mathId=si3.gif&_user=111111111&_pii=S1063520315000901&_rdoc=1&_issn=10635203&md5=f07fd28a6b43db1912ba9ab3b9f0318f" title="Click to view the MathML source">x0∈Rnr hidden">, then minimizing the rc">formulatext stixSupport mathImg" data-mathURL="/science?_ob=MathURL&_method=retrieve&_eid=1-s2.0-S1063520315000901&_mathId=si4.gif&_user=111111111&_pii=S1063520315000901&_rdoc=1&_issn=10635203&md5=b99adb36838f4f52c7263494dd2cf488" title="Click to view the MathML source">ℓ1r hidden"> norm subject to rc">formulatext stixSupport mathImg" data-mathURL="/science?_ob=MathURL&_method=retrieve&_eid=1-s2.0-S1063520315000901&_mathId=si5.gif&_user=111111111&_pii=S1063520315000901&_rdoc=1&_issn=10635203&md5=476b51babe6416af12074752583f6a36" title="Click to view the MathML source">|Ax|=br hidden"> recovers rc">formulatext stixSupport mathImg" data-mathURL="/science?_ob=MathURL&_method=retrieve&_eid=1-s2.0-S1063520315000901&_mathId=si6.gif&_user=111111111&_pii=S1063520315000901&_rdoc=1&_issn=10635203&md5=51dac00f8e2d0ec01d8d235c8e8686b4" title="Click to view the MathML source">x0r hidden"> up to multiplication by a global sign. Moreover, we establish that the SRIP holds for the random Gaussian matrices typically used for standard compressed sensing, implying that phaseless compressed sensing is possible from rc">formulatext stixSupport mathImg" data-mathURL="/science?_ob=MathURL&_method=retrieve&_eid=1-s2.0-S1063520315000901&_mathId=si19.gif&_user=111111111&_pii=S1063520315000901&_rdoc=1&_issn=10635203&md5=40b29d277c217365e9bd411b91a3667d" title="Click to view the MathML source">O(klog(en/k))r hidden"> measurements with these matrices via rc">formulatext stixSupport mathImg" data-mathURL="/science?_ob=MathURL&_method=retrieve&_eid=1-s2.0-S1063520315000901&_mathId=si4.gif&_user=111111111&_pii=S1063520315000901&_rdoc=1&_issn=10635203&md5=b99adb36838f4f52c7263494dd2cf488" title="Click to view the MathML source">ℓ1r hidden"> minimization over rc">formulatext stixSupport mathImg" data-mathURL="/science?_ob=MathURL&_method=retrieve&_eid=1-s2.0-S1063520315000901&_mathId=si5.gif&_user=111111111&_pii=S1063520315000901&_rdoc=1&_issn=10635203&md5=476b51babe6416af12074752583f6a36" title="Click to view the MathML source">|Ax|=br hidden">. Our analysis also yields an erasure robust version of the Johnson–Lindenstrauss Lemma.