On Jackknifed Greedy Algorithms and Their Applications in NMR

Multidimensional NMR experiments applied in e.g. protein NMR may take days or even weeks of expensive experimental time. Compressed sensing (CS) is a method to save this time: some measurements are skipped and then reconstructed with the help of special algorithms. The orthogonal matching pursuit (OMP) is one of the effective tools of compressed sensing field. In this paper we introduce a jackknife-type stopping criterion for OMP. Unlike the standard stopping criteria, it does not require an estimation of the noise level. We extend the concept to the lorentzian peaks matching pursuit (LPMP) algorithm.