Consider the problem of estimating the entries of a large matrix, when the
observed entries are noisy versions of a small random fraction of the original
entries. This problem has received widespread attention in recent times,
especially after the pioneering works of Emmanuel Cand\`{e}s and collaborators.
This paper introduces a simple estimation procedure,