Struct AdaptiveCompute 

pub struct AdaptiveCompute { /* private fields */ }

Adaptive GPU/CPU dispatcher for the legacy Cl(3,0,0) flat-batch helper.

Single multivector products always use the CPU amari-core baseline. batch_geometric_product accepts flat Cl(3,0,0) batches with 8 coefficients per multivector and uses GPU only when a context is available and the batch crosses GpuCliffordAlgebra::should_use_gpu.

Implementations

impl AdaptiveCompute

pub async fn new<const P: usize, const Q: usize, const R: usize>() -> Self

Create with optional GPU acceleration

pub async fn geometric_product<const P: usize, const Q: usize, const R: usize>( &self, a: &Multivector<P, Q, R>, b: &Multivector<P, Q, R>, ) -> Multivector<P, Q, R>

Perform geometric product, automatically choosing CPU or GPU

pub async fn batch_geometric_product( &self, a_batch: &[f64], b_batch: &[f64], ) -> Result<Vec<f64>, GpuError>

Batch geometric product with adaptive dispatch for flat Cl(3,0,0) inputs.