Sharp recovery bounds for convex demixing, with applications (Q404302): Difference between revisions

In modern data-intensive science, it is common to observe a superposition of multiple information-bearing signals. Demixing refers to the challenge of separating out the constituent signals from the observation. This work presents a general framework for demixing based on convex optimization. The authors study the geometry of the optimization problem and they develop conditions that describe precisely when this method succeeds.