TTNOpt: Tree tensor network package for high-rank tensor compression

We have developed TTNOpt, a software package that utilizes tree tensor networks (TTNs) for quantum spin systems and high-dimensional data analysis. TTNOpt provides efficient and powerful TTN computations by locally optimizing the network structure, guided by the entanglement pattern of the target tensors. For quantum spin systems, TTNOpt searches for the ground state of Hamiltonians with bilinear spin interactions and magnetic fields, and computes physical properties of these states, including the variational energy, bipartite entanglement entropy (EE), single-site expectation values, and two-site correlation functions. Additionally, TTNOpt can target the lowest-energy state within a specified subspace, provided that the Hamiltonian conserves total magnetization. For high-dimensional data analysis, TTNOpt factorizes complex tensors into TTN states that maximize fidelity to the original tensors by optimizing the tensors and the network. When a TTN is provided as input, TTNOpt reconstructs the network based on the EE without referencing the fidelity of the original state. We present three demonstrations of TTNOpt: (1) Ground-state search for the hierarchical chain model with a system size of $256$. The entanglement patterns of the ground state manifest themselves in a tree structure, and TTNOpt successfully identifies the tree. (2) Factorization of a quantic tensor of the $2^{24}$ dimensions representing a three-variable function where each variant has a weak bit-wise correlation. The optimized TTN shows that its structure isolates the variables from each other. (3) Reconstruction of the matrix product network representing a $16$-variable normal distribution characterized by a tree-like correlation structure. TTNOpt can reveal hidden correlation structures of the covariance matrix.

💡 Research Summary

TTNOpt is a Python‑based software library that leverages tree tensor networks (TTNs) for two distinct but related tasks: (i) variational ground‑state searches of quantum spin Hamiltonians and (ii) high‑dimensional tensor compression for data‑analytic applications. The central premise of the package is that the entanglement structure of the target object dictates the optimal network topology. While matrix‑product networks (MPNs, also known as tensor trains) are well suited for one‑dimensional short‑range entanglement, they become inefficient for states with long‑range or hierarchical correlations because a fixed bond dimension χ would have to grow exponentially. TTNs, by contrast, have a loop‑free tree topology that can be reshaped locally so that each bond carries only the amount of entanglement required by the underlying Schmidt decomposition. Consequently, accurate representations can be achieved with modest χ.

The authors adopt the local‑bond‑reconnection algorithm originally proposed in Ref.