cg_solve

iskra.sparse_linalg.cg_solve(f: Callable[[Tensor], Tensor], b: Tensor, init: Tensor, maxiter: int, tol: float) → Tensor[SOURCE]

Matrix-free Conjugate Gradient solver for (I - J^T) u = b.

Parameters:

• f (Callable[[Tensor], Tensor]) – Computes J^T @ z (vector-Jacobian product)

• b (Tensor) – Right-hand side of linear system, same shape as u.

• init (Tensor) – Initial guess for u.

• maxiter (int) – Maximum number of iterations for the solver.

• tol (float) – Minimum tolerance the solver needs to reach before exiting.

Returns:

​ (Tensor) – Solution to the linear system (I - J^T) u = b.