Crate tenferro_ext_burn 

Bridge between the Burn deep learning framework and tenferro tensor network operations.

This crate allows Burn tensors to be used with tenferro’s einsum and tensor network contraction routines, enabling seamless integration of tensor network methods into Burn-based deep learning pipelines.

Burn tensors are treated as row-major boundary values. The bridge normalizes them into tenferro’s internal column-major canonical layout for computation, then materializes row-major buffers again when exporting back to Burn.

Examples

use burn::backend::NdArray;
use burn::tensor::Tensor;
use tenferro_ext_burn::einsum;

// Matrix multiplication via einsum
let a: Tensor<NdArray<f64>, 2> = Tensor::ones([3, 4], &Default::default());
let b: Tensor<NdArray<f64>, 2> = Tensor::ones([4, 5], &Default::default());
let c: Tensor<NdArray<f64>, 2> = einsum("ij,jk->ik", vec![a, b]);

Re-exports

pub use convert::burn_to_tenferro;

pub use convert::tenferro_to_burn;

Modules

backward

Backward-mode support for TensorNetworkOps on Burn’s autodiff backend.

convert

Conversion utilities between Burn tensor primitives and tenferro tensors.

forward

Forward-mode (inference) implementation of TensorNetworkOps for the NdArray backend.

Enums

Error

Error type for Burn/tenferro bridge failures.

Traits

TensorNetworkOps

Trait for backends that support tenferro tensor network operations.

Functions

einsum

High-level infallible einsum convenience wrapper.

try_einsum

Fallible high-level einsum on Burn tensors, dispatching to the backend’s TensorNetworkOps::tn_einsum implementation.

Type Aliases

Result

Result type for Burn/tenferro bridge operations.