Source code for pyproximal.proximal.L21_plus_L1
import numpy as np
from pyproximal import ProxOperator
from pyproximal.ProxOperator import _check_tau
[docs]class L21_plus_L1(ProxOperator):
r"""L21 + L1 norm proximal operator.
Proximal operator of the :math:`L_{2,1} + L_1` mixed-norm:
:math:`f(\mathbf{X}) = \sigma \rho \|\mathbf{X}\|_1 +
\sigma (1 - \rho) \|\mathbf{X}\|_{2,1}`
Parameters
----------
sigma : :obj:`int`, optional
Multiplicative coefficient of :math:`L_{2,1} + L_1` mixed-norm
rho : :obj:`int`, optional
Balancing between sparsity of :math:`L_1` and grouping of :math:`L_{2,1}`
Notes
-----
The proximal operator of the :math:`L_{2,1} + L_1` mixed-norm is simply the
product of each individual proximal operator [1]_.
.. [1] Gramfort, Alexandre, Daniel Strohmeier, Jens Haueisen, Matti Hamalainen,
and Matthieu Kowalski. "Functional brain imaging with M/EEG using structured
sparsity in time-frequency dictionaries." In Biennial International Conference
on Information Processing in Medical Imaging, pp. 600-611. Springer, Berlin,
Heidelberg, 2011.
"""
def __init__(self, sigma=1.0, rho=0.8):
super().__init__(None, False)
self.sigma = sigma
self.rho = rho
def __call__(self, x):
return self.rho * self.sigma * np.sum(np.abs(x)) + \
(1 - self.rho) * self.sigma * np.sum(np.sqrt(np.sum(x ** 2, axis=0)))
@_check_tau
def prox(self, x, tau, axis=0):
thresh = self.sigma * tau
l1 = np.maximum(np.abs(x) - thresh * self.rho, 0)
# Axis defines what dimension to perform grouping over
aux_l21 = np.sqrt(np.sum(np.maximum(
np.abs(x) - thresh * self.rho, 0) ** 2, axis=axis))
l21 = np.maximum(1 - thresh * (1 - self.rho) / aux_l21, 0)
x = np.nan_to_num(x / np.abs(x)) * l1 * l21
return x