Trait RuntimeAdjointLinearOperator 

pub trait RuntimeAdjointLinearOperator<R>: RuntimeLinearOperator<R>
where
    R: Runtime,
{
    // Required method
    fn apply_adjoint_on(
        &self,
        runtime: &R,
        input: &R::Buffer,
        output: &mut R::Buffer,
    ) -> Result<()>;
}

Runtime-aware adjoint action for reverse-mode scientific rules.

This stays separate from RuntimeLinearOperator so existing execution backends remain source-compatible. A GPU or distributed backend implements this trait with its native adjoint kernel; the CPU blanket implementation delegates to LinearOperator::apply_adjoint.

Required Methods

fn apply_adjoint_on( &self, runtime: &R, input: &R::Buffer, output: &mut R::Buffer, ) -> Result<()>

Compute output = A† * input in runtime-owned storage.

Implementors

impl<O> RuntimeAdjointLinearOperator<CpuRuntime> for O

where O: LinearOperator + ?Sized,