We present a linear program that is capable of determining whether a set of correlations can be captured by a local realistic model. If the correlations can be described by such a model, the linear program outputs a joint probability distribution that produces the given correlations. If the correlations cannot be described under the assumption of local realism, the program outputs a Bell inequality violated by the correlations.
@article{arxiv.0905.2950,
title = {A linear program for testing local realism},
author = {Matthew B. Elliott},
journal= {arXiv preprint arXiv:0905.2950},
year = {2009}
}