How many linear measurements does it take to recover an object of interest? What makes this result truly relevant is that the The practical results are striking; in general, a From an algebraic standpoint, the measurement process will simultanuously discover the exact support and the component values of f from only the K measurements. In short, recovery via l1 minimization works, well, L1-MAGIC is a collection of MATLAB routines for sparse Download A nonlinear sampling theorem by: Emmanuel Candes, Justin Romberg, and Terence Tao To appear in IEEE Transactions on Information Theory, February 2006. The central result from this paper is that a sparse Fourier domain observations. More precisely, let f be a length-N discrete signal which has B nonzero components (we stress that the number and locations of the components are unknown a priori). We collect samples at K different frequencies which are randomly selected. Then for K on the order of B log N, we can recover f perfectly (with very high probability) through l1 minimization. Download Near-optimal signal recovery and the Uniform Uncertainty by: Emmanuel Candes and Terence Tao Submitted to IEEE Transactions on Information Theory, November 2004. This paper derives precise conditions for when Download Stability This paper demonstrates that the recovery procedure Download Statistical Estimation by: Emmanuel Candes and Terence Tao Submitted to IEEE Transactions on Information Theory, June 2005. When the errors made in the measurement process are Download Linear Decoding This paper demonstrates that in addition to recovering Download Finding Sparse Decompositions by: Emmanuel Candes and Justin Romberg To appear in Foundations of Computational Mathematics, 2006. This paper revisits the now classical application Download |

# l1-Magic

top