Implementing the Singular Value Decomposition in the Helmholtz Analytics Toolkit HeAT

Kurzfassung

Singular value decomposition (SVD) is a fundamental tool in data science and often used as, e.g., a preprocessing step. When dealing with very large data sets as they often arise at the DLR, performing the analysis in a scalable way on HPC systems can be necessary; this also includes the computation of the SVD. In this talk we present work in progress regarding our implementation of a parallel SVD within the PyTorch- and mpi4py-based HPC-data analytics software HEAT (Helmholtz Analytics Toolkit) developed at DLR, JSC, and KIT (Götz et al., 2020 IEEE International Conference on Big Data, pp. 276-287).