GPU implementation of singular value decomposition for high rank tensors

Authors

  • Kryzz Joshua Gonzaga de Leon National Institute of Physics, University of the Philippines Diliman
  • Francis N. C. Paraan National Institute of Physics, University of the Philippines Diliman

Abstract

Programming using the Python API (application programming interface) offers some advantages over using compiled languages. Here we implement a high rank tensor decomposition routine using the TensorFlow library which has native support for utilizing multi-core CPU, GPU, and TPU hardware. Specifically, a singular value decomposition algorithm was performed on a rank-5 tensor. The performance of this Python implementation was compared with a known C++ based library written specifically for tensor manipulations but without native GPU support. We report some use cases where the implementation on a consumer grade GPU was empirically faster than the C++ based library when the rank-5 tensor has more than 2 × 106 elements. With the acceptable performance of the implementation, it may be beneficial to have have a native implementation of tensor network operations on TensorFlow.

Downloads

Issue

Article ID

SPP-2018-PB-50

Section

Poster Session B (Complex Systems, Simulations, and Theoretical Physics)

Published

2018-05-31

How to Cite

[1]
KJG de Leon and FNC Paraan, GPU implementation of singular value decomposition for high rank tensors, Proceedings of the Samahang Pisika ng Pilipinas 36, SPP-2018-PB-50 (2018). URL: https://proceedings.spp-online.org/article/view/SPP-2018-PB-50.