Crate tensor4all_simplett 

Tensor Train (MPS) library for numerical tensor networks.

A tensor train (TT), also known as a Matrix Product State (MPS), decomposes a high-dimensional tensor into a chain of rank-3 cores:

T[i0, i1, ..., iL-1] = A0[i0] * A1[i1] * ... * A_{L-1}[i_{L-1}]

where each Ak[ik] is a matrix of shape (r_{k-1}, r_k) and the bond dimensions r_k control the approximation accuracy.

Main types

Type	Purpose
TensorTrain	Primary tensor train container
AbstractTensorTrain	Common interface (evaluate, sum, norm)
CompressionOptions	Controls compression accuracy/cost trade-off
TTCache	Caches partial contractions for repeated evaluation
SiteTensorTrain	Center-canonical form for sweeping algorithms
VidalTensorTrain	Vidal canonical form with explicit singular values
Tensor3 / Tensor3Ops	Rank-3 core tensors and their operations

Typical workflow

use tensor4all_simplett::{TensorTrain, AbstractTensorTrain, CompressionOptions};

// 1. Create a tensor train (e.g. constant function)
let tt = TensorTrain::<f64>::constant(&[2, 3, 2], 1.0);
assert_eq!(tt.len(), 3);

// 2. Evaluate at a specific multi-index
let value = tt.evaluate(&[0, 1, 1]).unwrap();
assert!((value - 1.0).abs() < 1e-12);

// 3. Sum over all indices: 1.0 * 2 * 3 * 2 = 12.0
let sum = tt.sum();
assert!((sum - 12.0).abs() < 1e-10);

// 4. Compress to reduce bond dimensions
let compressed = tt.compressed(&CompressionOptions::default()).unwrap();
let val2 = compressed.evaluate(&[0, 1, 1]).unwrap();
assert!((val2 - 1.0).abs() < 1e-10);

Re-exports

pub use cache::TTCache;

pub use canonical::center_canonicalize;

pub use canonical::SiteTensorTrain;

pub use compression::CompressionMethod;

pub use compression::CompressionOptions;

pub use contraction::dot;

pub use contraction::ContractionOptions;

pub use error::Result;

pub use error::TensorTrainError;

pub use tensortrain::TensorTrain;

pub use traits::AbstractTensorTrain;

pub use traits::TTScalar;

pub use types::tensor3_from_data;

pub use types::tensor3_zeros;

pub use types::LocalIndex;

pub use types::MultiIndex;

pub use types::Tensor3;

pub use types::Tensor3Ops;

pub use vidal::DiagMatrix;

pub use vidal::InverseTensorTrain;

pub use vidal::VidalTensorTrain;

Modules

arithmetic

Arithmetic operations for tensor trains

cache

Cached tensor train evaluation

canonical

Canonical forms for tensor trains

compression

Compression algorithms for tensor trains

contraction

Contraction operations for tensor trains

error

Error types for tensor train operations

mpo

MPO (Matrix Product Operator) contraction algorithms

tensor

Fixed-rank tensor type backed by tenferro_tensor::TypedTensor<T>.

tensortrain

TensorTrain implementation

traits

Abstract traits for tensor train objects

types

Core types for tensor train operations

vidal

Vidal and Inverse tensor train representations