TensorNetwork on TensorFlow: A Spin Chain Application Using Tree Tensor Networks
Strongly Correlated Electrons
2019-05-07 v1 Machine Learning
High Energy Physics - Theory
Computational Physics
Machine Learning
Abstract
TensorNetwork is an open source library for implementing tensor network algorithms in TensorFlow. We describe a tree tensor network (TTN) algorithm for approximating the ground state of either a periodic quantum spin chain (1D) or a lattice model on a thin torus (2D), and implement the algorithm using TensorNetwork. We use a standard energy minimization procedure over a TTN ansatz with bond dimension , with a computational cost that scales as . Using bond dimension we compare the use of CPUs with GPUs and observe significant computational speed-ups, up to a factor of , using a GPU and the TensorNetwork library.
Cite
@article{arxiv.1905.01331,
title = {TensorNetwork on TensorFlow: A Spin Chain Application Using Tree Tensor Networks},
author = {Ashley Milsted and Martin Ganahl and Stefan Leichenauer and Jack Hidary and Guifre Vidal},
journal= {arXiv preprint arXiv:1905.01331},
year = {2019}
}
Comments
All code can be found at https://github.com/google/tensornetwork