Crate tensor4all_treetn 

Tree Tensor Network (TreeTN) library for tensor4all.

This crate provides data structures and algorithms for working with tree tensor networks, a generalization of matrix product states (MPS) to tree-shaped graphs. Tree tensor networks are useful for representing quantum states and operators on systems with tree-like connectivity.

Key Types

• TreeTN: The main tree tensor network type, parameterized by tensor and node name types.
• DefaultTreeTN: A convenient alias for TreeTN<TensorDynLen, NodeIndex>.
• NamedGraph: A graph wrapper that maps node names to internal graph indices.

Features

• Canonicalization: Transform networks into canonical forms (unitary, LU, CI).
• Truncation: Compress networks using SVD-based truncation.
• Contraction: Contract two networks using zip-up or fit algorithms.
• Linear operators: Apply and compose linear operators on tree tensor networks.
• Linear solvers: Solve linear systems involving tree tensor network operators.

Re-exports

pub use algorithm::CanonicalForm;

pub use algorithm::CompressionAlgorithm;

pub use algorithm::ContractionAlgorithm;

pub use treetn::contraction;

pub use link_index_network::LinkIndexNetwork;

pub use named_graph::NamedGraph;

pub use node_name_network::CanonicalizeEdges;

pub use node_name_network::NodeNameNetwork;

pub use operator::apply_linear_operator;

pub use operator::are_exclusive_operators;

pub use operator::build_identity_operator_tensor;

pub use operator::compose_exclusive_linear_operators;

pub use operator::ApplyOptions;

pub use operator::ArcLinearOperator;

pub use operator::IndexMapping;

pub use operator::LinearOperator;

pub use operator::Operator;

pub use options::CanonicalizationOptions;

pub use options::RestructureOptions;

pub use options::SplitOptions;

pub use options::TruncationOptions;

pub use random::random_treetn;

pub use random::LinkSpace;

pub use site_index_network::SiteIndexNetwork;

pub use treetn::apply_local_update_sweep;

pub use treetn::factorize_tensor_to_treetn;

pub use treetn::factorize_tensor_to_treetn_with;

pub use treetn::get_boundary_edges;

pub use treetn::partial_contract;

pub use treetn::BoundaryEdge;

pub use treetn::LocalUpdateStep;

pub use treetn::LocalUpdateSweepPlan;

pub use treetn::LocalUpdater;

pub use treetn::PartialContractionSpec;

pub use treetn::ScheduledSwapStep;

pub use treetn::SwapOptions;

pub use treetn::SwapSchedule;

pub use treetn::TreeTN;

pub use treetn::TreeTopology;

pub use treetn::TruncateUpdater;

pub use linsolve::square_linsolve;

pub use linsolve::EnvironmentCache;

pub use linsolve::LinsolveOptions;

pub use linsolve::LinsolveVerifyReport;

pub use linsolve::NetworkTopology;

pub use linsolve::NodeVerifyDetail;

pub use linsolve::ProjectedOperator;

pub use linsolve::ProjectedState;

pub use linsolve::SquareLinsolveResult;

pub use linsolve::SquareLinsolveUpdater;

Modules

algorithm

Algorithm selection types for tensor train operations.

link_index_network

Link Index Network for bond/link index management.

linsolve

Linear equation solvers for Tree Tensor Networks.

named_graph

Named graph wrapper inspired by NamedGraphs.jl (https://github.com/mtfishman/NamedGraphs.jl)

node_name_network

Node Name Network - Graph structure for node name connections.

operator

Operator composition for Tree Tensor Networks.

options

Options and traits for TreeTN operations.

random

Random tensor network generation.

site_index_network

Site Index Network (inspired by ITensorNetworks.jl’s IndsNetwork)

treetn

Tree Tensor Network implementation.

Functions

tensor_train_to_treetn

Convert a linear-chain simple tensor train into a TreeTN with node names 0..n-1.

tensor_train_to_treetn_with_names

Convert a linear-chain simple tensor train into a TreeTN with explicit node names.

tensor_train_to_treetn_with_names_and_site_indices

Convert a linear-chain simple tensor train into a TreeTN with explicit node names and caller-provided site indices.

Type Aliases

DefaultTreeTN

Default TreeTN type using TensorDynLen as the tensor type.