he_ring::ntt::dyn_convolution

Trait DynConvolutionAlgorithm

Source 
pub trait DynConvolutionAlgorithm<R>
where
    R: ?Sized + RingBase,
{
    // Required methods
    fn compute_convolution_dyn(
        &self,
        lhs: &[R::Element],
        rhs: &[R::Element],
        dst: &mut [R::Element],
        ring: &R,
    );
    fn supports_ring_dyn(&self, ring: &R) -> bool;
}
Expand description

Trait for algorithms that compute convolutions. This mirrors ConvolutionAlgorithm, but is dyn-compatible. This is useful if you want to create a ring but only know the type of the convolution algorithm at runtime.

Wrap a dyn DynConvolutionAlgorithm<R> in DynConvolutionAlgorithmConvolution to use it as a ConvolutionAlgorithm.

Required Methods§

Source

fn compute_convolution_dyn( &self, lhs: &[R::Element], rhs: &[R::Element], dst: &mut [R::Element], ring: &R, )

Computes dst[i] += sum_j lhs[j] * rhs[i - j], where the sum runs over these indices that do not cause an out-of-bounds.

For implementation purposes, we requrie dst.len() >= lhs.len() + rhs.len() (not only dst.len() >= lhs.len() + rhs.len() - 1, which would be enough to include lhs[lhs.len() - 1] * rhs[rhs.len() - 1]).

Source

fn supports_ring_dyn(&self, ring: &R) -> bool

Implementors§