autopdex.solver.damped_jacobi_relaxation

autopdex.solver.damped_jacobi_relaxation(hvp_fun, diag, x_0, rhs, damping_factor=0.3333333, **kwargs)

Damped Jacobi smoother (experimental).

Parameters:

• hvp_fun (callable) – Hessian vector product function.

• diag (jnp.ndarray) – Diagonal of the Hessian matrix.

• x_0 (jnp.ndarray) – Initial guess for the solution.

• rhs (jnp.ndarray) – Right-hand side vector of the linear system.

• damping_factor (float) – Damping factor for the iterations (<=0.5 guarantees a good smoother).

• **kwargs (dict) – Additional keyword arguments for customization.

Returns:

Solution vector x.

Return type:

array