Distributing persistent homology via spectral sequences (Q6050215): Difference between revisions

Despite a better understanding of the theoretical and empirical properties, the calculation of persistent homology for large-scale data sets proved to be elusive for a long time. This paper provides a leap forward towards parallelising the computation of persistent homology based on spectral sequences. Building on the Mayer-Vietoris spectral sequence, the author presents an algorithm that enables calculations in a purely local fashion. To obtain the full persistent homology of the input data, a suitable cover of said data is required, which then enables the combination of all local results into a global outcome. Next to providing statements about issues arising from the spectral sequence approach and how to overcome them, the paper also provides a working implementation of the proposed algorithm.