Entanglement witnesses with variable number of local measurements

Abstract

We present a class of entanglement identifiers which has the following experimentally friendly feature: once the expectation value of the identifier exceeds some definite limit, we can conclude the state is entangled, even if not all measurements defining the identifier have been performed. These identifiers are in the form of sums of non-negative functions of correlations in a quantum state, mostly squares of correlations, and we illustrate their use and strengths on various examples.