Distributed Principal Component Analysis with Limited Communication

However, in many settings the large volume of data requires processing by multiple nodes, which has inspired significant work on reducing the distribution overheads of PCA, e.g.