Crate ariadnetor 

ariadnetor: tensor network library in Rust.

ariadnetor is the umbrella tensor library; it re-exports types and functions from the layers listed below into its own namespace. Each layer depends only on the layers listed earlier:

• ariadnetor_core — backend-agnostic abstractions (Scalar, ComputeBackend, EinsumExpr). The MemoryOrder layout type is intentionally not re-exported: the umbrella’s public API hides memory layout from end users.
• ariadnetor_native — NativeBackend: faer (+ optional hptt-rs transpose).
• ariadnetor_tensor — user-facing Tensor, DenseTensor, BlockSparseTensor, Sector, QNIndex.
• ariadnetor_linalg — backend-agnostic linear algebra over &Tensor (contract, svd, qr, eigh, expm, …).

ariadnetor_mps and ariadnetor_algorithms are separate consumer crates that depend on the leaf crates directly rather than on this umbrella; they are not re-exported here.

Example

use ariadnetor::DenseTensor;

let a = DenseTensor::<f64>::zeros(vec![2, 3]);
let b = DenseTensor::<f64>::zeros(vec![3, 2]);

assert_eq!(a.shape(), &[2, 3]);
assert_eq!(b.shape(), &[3, 2]);

Modules

expert

Expert layer: per-call ExecPolicy control.

Structs

BlockCoord

N-dimensional block coordinate.

BlockMeta

Metadata for a single block within a block-sparse tensor.

BlockScalars

Per-sector real scalar values keyed by fused sector.

Complex

A complex number in Cartesian form.

EinsumExpr

Parsed einsum expression with N inputs (indices as ASCII codes)

NativeBackend

Native backend using faer for GEMM and, with the hptt feature, HPTT for transpose (a naive kernel otherwise).

QNIndex

Quantum-number index for one tensor leg.

Tensor

Tensor wrapping a TensorData bundle.

TruncSvdParams

Parameters for truncated SVD.

U1Sector

U(1) symmetry sector (integer charge).

Z2Sector

Z₂ symmetry sector (values 0 or 1).

Enums

Direction

Leg direction for flux computation (see each variant).

ExecPolicy

Per-call execution policy for a compute backend operation.

LinalgError

Error from a linalg operation.

TensorError

Errors raised by tensor construction and manipulation.

Traits

BlockSparseHostOps

Host-defaulting method forms of the block-sparse explicit-backend operations.

ComputeBackend

Pluggable compute backend trait

DenseHostOps

Host-defaulting method forms of the dense explicit-backend operations.

LinalgContract

Sealed tensor-keyed dispatch trait for two-operand tensor contraction.

LinalgDecompose

Sealed tensor-keyed dispatch trait for the four tensor decompositions.

LinalgScale

Sealed tensor-keyed dispatch trait for diagonal scaling.

OpsFor

Compile-time marker: backend Self supports operations on storage flavor St. Implemented selectively per (backend, storage) pair, so a backend that cannot operate on a given storage simply omits the impl.

Scalar

Scalar type for tensor elements (sealed trait).

Sector

Abelian symmetry sector.

Functions

contract

Two-operand contraction over the einsum notation, using the supplied backend and the auto-selected execution policy.

diag_with_backend

Diagonal extraction / construction. The backend argument is accepted for API uniformity with the other twins; this operation needs no kernel, so it is unused here.

diagonal_scale

Per-slice diagonal scaling along axis, using the supplied backend.

eig_with_backend

General eigenvalue decomposition, using the supplied backend.

eigh_with_backend

Self-adjoint eigenvalue decomposition, using the supplied backend.

eigvals_with_backend

Eigenvalues-only general decomposition, using the supplied backend.

eigvalsh_with_backend

Eigenvalues-only self-adjoint decomposition, using the supplied backend.

einsum_with_backend

N-input Einstein summation, using the supplied backend.

expm_antihermitian_with_backend

Anti-Hermitian matrix exponential, using the supplied backend.

expm_hermitian_with_backend

Hermitian matrix exponential, using the supplied backend.

expm_with_backend

General matrix exponential, using the supplied backend.

inverse_with_backend

Matrix inverse via LU, using the supplied backend.

linear_combine

Linear combination of dense tensors: Σ_i coefs[i] * tensors[i].

lq

Thin LQ of a tensor reshaped as a matrix, using the supplied backend.

permute_with_backend

General axis permutation of a dense tensor, using the supplied backend.

qr

Thin QR of a tensor reshaped as a matrix, using the supplied backend.

solve_with_backend

Linear solve AX = B via LU, using the supplied backend.

svd

Thin SVD of a tensor reshaped as a matrix, using the supplied backend.

tensordot

Tensordot: contract lhs and rhs over the given axis pairs, emitting the output legs in their natural order (free left legs then free right legs, each in input axis order). The axis-pair face of contract, dispatched through the same tensor-keyed trait, so it serves both Dense and BlockSparse and returns a uniform tensor (rank-0 for a full contraction).

trace_with_backend

Partial trace over bond index pairs. The backend argument is accepted for API uniformity with the other twins; the partial trace needs no kernel, so it is unused here.

trunc_svd

Truncated SVD of a tensor reshaped as a matrix, using the supplied backend.

Type Aliases

BlockSparseEigResult

Result of a block-sparse general (non-Hermitian) eigenvalue decomposition: (eigenvalues, eigenvectors).

BlockSparseEighResult

Result of a block-sparse self-adjoint eigenvalue decomposition: (eigenvalues, eigenvectors).

BlockSparseQrResult

Result of a block-sparse QR or LQ decomposition.

BlockSparseSvdResult

Result of a block-sparse SVD: (U, S, Vt).

BlockSparseTensor

BlockSparse tensor alias.

BlockSparseTruncSvdResult

Result of a truncated block-sparse SVD: (U, S, Vt, trunc_err).

DenseTensor

Dense tensor alias.

EigResult

Result of a general eigenvalue decomposition: (eigenvalues, eigenvectors).

EighResult

Result of a self-adjoint eigenvalue decomposition: (eigenvalues, eigenvectors).

Host

The default host compute substrate, aliased so signatures name it through one stable alias rather than spelling the concrete backend type; repointing the substrate is then a one-line change.

LqResult

Result of a thin LQ decomposition: (L, Q).

QrResult

Result of a thin QR decomposition: (Q, R).

SvdResult

Result of a thin SVD decomposition: (U, S, Vt).

TruncSvdResult

Result of a truncated SVD decomposition: (U, S, Vt, trunc_err).