tomobar.data_fidelities#
Data fidelities to be used with the iterative methods such as FISTA, ADMM.
- tomobar.data_fidelities.grad_data_term(self, x: <cp.ndarray>, b: <cp.ndarray>, use_os: bool, sub_ind: int, w: <cp.ndarray> | None = None, w_sum: <cp.ndarray> | None = None) <cp.ndarray>[source]#
Calculation of the gradient of the data fidelity term :param x: the current solution/volume as a 3D CuPy array. :type x: cp.ndarray :param b: projection data, either post-log for LS-type methods, or pre-log for Poisson likelihood (KL). :type b: cp.ndarray :param use_os: True when OS-reconstruction is enabled. :type use_os: bool :param sub_ind: index for the ordered-subset approach. :type sub_ind: int :param indVec: Array of indices for the OS-model. :type indVec: Optional, cp.ndarray :param w: weights for Penalised-Weighted LS. :type w: Optional, cp.ndarray
- Returns:
gradient of the data fidelity as a 3D CuPy array.
- Return type:
cp.ndarray