Module simd_kernel 

3D Geometric Algebra (Cl_3,0) SIMD kernel.

Split into focused submodules (CLAUDE.md §11):

• simd_backend — the [f32; 8] array kernels, basis sign tables, AVX2/FMA.
• types — the value types (Multivector, Grade, Rotor, Translator).
• operations — grade-aware products and rotor/translator application.

The public surface is re-exported verbatim so simd_kernel::* consumers are unaffected by the split.

Re-exports

pub use operations::apply_rotor;

pub use operations::apply_translator;

pub use operations::geometric_product;

pub use operations::is_simd_available;

pub use operations::outer_product;

pub use operations::rotor_from_angle_axis;

pub use operations::translator_from_displacement;

pub use simd_backend::multivector_geometric_product;

pub use simd_backend::multivector_outer_product;

pub use simd_backend::GaKernel;

pub use simd_backend::GA_SIMD_KERNEL;

pub use types::Grade;

pub use types::Multivector;

pub use types::Rotor;

pub use types::Translator;

Modules

operations

High-level geometric-algebra operations over the value types: products that track grades, and rotor/translator construction and application.

simd_backend

Low-level Cl(3,0) array kernels.

types

Geometric-algebra value types: Multivector, Grade, Rotor, Translator and their constructors / accessors / operators. Pure structured math over [f32; N] coefficient arrays — no SIMD, no graph.