Compressed sensing with preconditioning for sparse recovery with subsampled matrices of Slepian prolate functions (Q467114): Difference between revisions

In this paper, the author discusses the efficient recovery of a smooth function \(f\in L^2(-1,1)\) which has a sparse expansion in the orthonormal basis of prolate spheroidal wave functions (PSWF) \(\varphi_k\) \((k=0,1,\ldots)\). Following the general theory of \textit{H. Rauhut} [in: Theoretical foundations and numerical methods for sparse recovery, Vienna, Austria, 2009. Berlin: de Gruyter, 1--92 (2010; Zbl 1208.15027)], an \(L^{\infty}\) bound for \(|\varphi_k|\) \((k=0,\ldots,N-1)\) is presented. Such a bound provides the restricted isometry property (RIP) for the measurement matrix \(\Phi = (\varphi_k (t_j))_{j=1,k=0}^{m,N-1}\) with \(m\) randomly distributed samples \(t_j\in [-1,\,1]\). As know, RIP of \(\Phi\) leads to exact recovery of \(f\) by means of \(\ell^1\) minimization. Later the author extends the preconditioning technique for sparse Legendre polynomial expansions introduced by \textit{H. Rauhut} and \textit{R. Ward} [J. Approx. Theory 164, No. 5, 517--533 (2012; Zbl 1239.65018)] to sparse PSWF expansions. Numerical examples illustrate the results.