Skip to main content

solve_matrix_in

Function solve_matrix_in 

Source
pub fn solve_matrix_in<T>(
    session: &mut dyn Engine,
    a: &Matrix<T>,
    b: &Matrix<T>,
) -> Result<Matrix<T>, BackendLinalgError>
where T: BackendLinalgScalar + Copy, Tensor: From<TypedTensor<T>>,
Expand description

Solve A X = B on a caller-held engine.

Same kernel and validation as solve_matrix, entered through the given session instead of the process-global one. The scalar conversion differs for the narrow types: f64 and Complex64 are solved directly on both routes, so the results agree bit for bit, but this entry solves f32 and Complex32 in their own precision, while solve_matrix promotes them to f64 / Complex64 and narrows the result. For the narrow types the two routes can therefore differ in the last bits; MatrixSolveScalar’s session hook applies the global route’s promotion.

§Errors

Returns BackendLinalgError under the same conditions as solve_matrix.