pub trait MatTransposeVec<V> {
// Required method
fn mat_transpose_vec(&self, x: &V) -> V;
}Expand description
Transposed matrix-vector product y = Aᵀ x. Lets least-squares
solvers form Jᵀ r without materializing Jᵀ.
§Contract
- Caller must: pass
xwhose length equalsself.nrows(). Backends panic on shape mismatch. - Implementor must: return a freshly allocated
Vof lengthself.ncols(), withy[j] = Σᵢ A[i, j] · x[i]. The op is a pure function of(self, x)and must agree withMatVecon the implicit transpose.
§Backends
Same backend coverage as MatVec.
Required Methods§
Sourcefn mat_transpose_vec(&self, x: &V) -> V
fn mat_transpose_vec(&self, x: &V) -> V
Compute Aᵀ x, allocating a fresh V of length self.ncols().