QuTree: A tree tensor network package
The Journal of Chemical Physics (2024)
In this paper, scientists introduce QuTree, a C++ library designed for tree tensor network methods.
The library offers tools for handling tensors, tensor trees, and related linear algebra functions, making it easier to develop approaches like the multilayer multiconfigurational time-dependent Hartree method or the density matrix renormalization group method and its variations. The computational overhead for managing the network structure is minimal, even when dealing with very large data sets and small tensors.
QuTree focuses on providing straightforward, high-level functions while still allowing users easy access to the underlying computational processes, which can help encourage new developments. The team used QuTree to calculate the eigenstates of coupled harmonic oscillators and run simulations of random circuits on a simulated quantum computer.
The study’s authors are Roman Ellerbrock, K. Grace Johnson, Stefan Seritan, Hannes Hoppe, J. H. Zhang, Tim Lenzen, Thomas Weike, Uwe Manthe, Todd J. Martínez, who did their work at Stanford University, SLAC National Accelerator Laboratory and Bielefeld University.