lib.cost module¶
COST FUNCTION
This module the class for the sf_deconvolveCost cost function.
| Author: | Samuel Farrens <samuel.farrens@gmail.com> |
|---|---|
| Version: | 1.1 |
| Date: | 20/10/2017 |
-
class
lib.cost.sf_deconvolveCost(y, grad, wavelet=None, weights=None, lambda_lowr=None, lambda_psf=1, mode='lowr', positivity=True, verbose=True)[source]¶ Bases:
objectCost function class for sf_deonvolve
This class implements the cost function for deconvolution
Parameters: - y (np.ndarray) – Input original data array
- operator (function) – Matrix operator function
- wavelet (class, optional) – Wavelet operator class (“sparse” mode only)
- weights (np.ndarray, optional) – Array of wavelet thresholding weights (“sparse” mode only)
- lambda_lowr (float, optional) – Low-rank regularization parameter (“lowr” mode only)
- lambda_psf (float, optional) – PSF estimate regularization parameter (“psf_unknown” grad_type only)
- mode (str {'lowr', 'sparse'}, optional) – Deconvolution mode (default is “lowr”)
- positivity (bool, optional) – Option to test positivity contraint (defult is “True”)
- verbose (bool) – Option for verbose output (default is “True”)
-
calc_cost(*args)[source]¶ Get cost function
This method calculates the cost
Parameters: x (np.ndarray) – Deconvolved data array Returns: Return type: float cost
-
grad_comp(x)[source]¶ Calculate gradient component of the cost
This method returns the l2 norm error of the difference between the original data and the data obtained after optimisation
Parameters: x (np.ndarray) – Deconvolved data array Returns: Return type: float gradient cost component
-
lowr_comp(x)[source]¶ Calculate low-rank component of the cost
This method returns the nuclear norm error of the deconvolved data in matrix form
Parameters: x (np.ndarray) – Deconvolved data array Returns: Return type: float low-rank cost component