Module validate 

Validation helpers shared across backends and exec layers.

Examples

use tenferro_tensor::validate::validate_nonsingular_u;
use tenferro_tensor::{Tensor, TypedTensor};

let t = Tensor::F64(TypedTensor::from_vec_col_major(vec![2, 2], vec![1.0, 0.0, 0.0, 1.0]).unwrap());
assert!(validate_nonsingular_u(&t).is_ok());

Traits

DiagSingularity

Trait for detecting singular or non-finite diagonal entries.

Functions

can_convert_dtype

Return whether public convert may change from into to.

check_singular_diagonal

Checks that every diagonal element of a (possibly batched) upper-triangular factor is non-singular and finite.

checked_shape_product

Compute a shape product with overflow reported as a typed tensor error.

matmul_config_for_shapes

Validate rank-2 matrix multiplication shapes and return its dot-general config.

promote_dtype

Promote two dtypes according to tenferro’s public dtype-promotion lattice.

validate_convert_dtype

Validate a public checked dtype conversion.

validate_nonsingular_u

Validates that the upper-triangular factor u of a matrix decomposition has no singular (zero) or non-finite diagonal entries.

validate_permutation_axes

Validate a full permutation for a tensor rank.

validate_unique_axes

Validate a subset of axes for a tensor rank.