Skip to main content

Crate burn_linalg

Crate burn_linalg 

Source
Expand description

Linear algebra operations for Burn tensors.

No execution backend is enabled by default. Select flex, wgpu, or another backend feature on this crate. When configuring through burn, enable both linalg and the backend feature, for example features = ["linalg", "flex"]. Enabling only burn/flex does not enable this crate’s backend implementations.

§Migration

Linear algebra previously lived at burn_tensor::linalg (and therefore burn_core::tensor::linalg). Those direct-crate paths are intentionally removed. Use this crate directly, or burn::linalg from the umbrella crate.

Enums§

Norm
Specifies the type of norm to compute.

Traits§

LinalgOps
Linear algebra operations supplied by a backend extension.

Functions§

cosine_similarity
Computes the cosine similarity between two tensors along a specified dimension.
det
Computes the determinant on the last two dimensions of the input tensor.
diag
Returns the diag of a matrix.
l0_norm
Computes the L0 norm of a tensor along a specified dimension.
l0_norm_dims
Computes the L0 norm of a tensor along specified dimensions.
l1_norm
Computes the L1 norm of a tensor along a specified dimension.
l1_norm_dims
Computes the L1 norm of a tensor along specified dimensions.
l2_norm
Computes the L2 norm of a tensor along a specified dimension.
l2_norm_dims
Computes the L2 norm of a tensor along specified dimensions.
lp_norm
Computes the general L(p) norm of a tensor along a specified dimension.
lp_norm_dims
Computes the general L(p) norm of a tensor along specified dimensions.
lu
Computes the LU decomposition of a square or rectangular matrix with partial pivoting.
matvec
Performs matrix-vector multiplication with optional batch dimensions.
max_abs_norm
Computes the L:INFINITY norm of a tensor along a specified dimension.
max_abs_norm_dims
Computes the L:INFINITY norm of a tensor along specified dimensions.
min_abs_norm
Computes the L:NEG_INFINITY norm of a tensor along a specified dimension.
min_abs_norm_dims
Computes the L:NEG_INFINITY norm of a tensor along specified dimensions.
outer
Computes the outer product for the last columns of 2 tensors.
outer_dim
Computes the outer product along a specific dimension, broadcasting over others.
qr
Computes the QR decomposition of a square or rectangular matrix using Householder reflections.
svd
Computes the singular value decomposition of a square or rectangular matrix.
trace
Computes the trace of a matrix.
vector_norm
Computes the vector norm of a tensor along a specified dimension.
vector_norm_dims
Computes the vector norm of a tensor along specified dimensions.
vector_normalize
Normalize a tensor versus its vector_norm.