cr.sparse.sls.power_iterations_jit¶
- cr.sparse.sls.power_iterations_jit(operator, b, max_iters=100, error_tolerance=1e-06)¶
Computes the largest eigen value of a (symmetric) linear operator by power method
- Parameters
operator (cr.sparse.lop.Operator) – A symmetric linear operator \(A\)
b (jax.numpy.ndarray) – A user provided initial guess for the largest eigen vector
max_iters (int) – Maximum number of iterations
error_tolerance (float) – Tolerance for relative change in largest eigen value
- Returns
A named tuple containing the largest eigen value, corresponding eigen vector and the number of iterations for convergence
- Return type
The operator may accept multi-dimensional arrays as input. E.g. a 2D convolution operator will accept 2D images as input. In such cases, the eigen vector will also be a multi-dimensional array.
Note
This will not work for matrices with complex eigen values.