diff options
Diffstat (limited to 'Wrappers/Python')
| -rw-r--r-- | Wrappers/Python/ccpi/optimisation/functions/L1Norm.py | 88 |
1 files changed, 7 insertions, 81 deletions
diff --git a/Wrappers/Python/ccpi/optimisation/functions/L1Norm.py b/Wrappers/Python/ccpi/optimisation/functions/L1Norm.py index 79040a0..37c2016 100644 --- a/Wrappers/Python/ccpi/optimisation/functions/L1Norm.py +++ b/Wrappers/Python/ccpi/optimisation/functions/L1Norm.py @@ -60,7 +60,7 @@ class L1Norm(Function): y = 0 if self.b is not None: - y = 0 + (self.b * x).sum() + y = 0 + self.b.dot(x) return y def proximal(self, x, tau, out=None): @@ -95,98 +95,24 @@ class L1Norm(Function): return ScaledFunction(self, scalar) -#import numpy as np -##from ccpi.optimisation.funcs import Function -#from ccpi.optimisation.functions import Function -#from ccpi.framework import DataContainer, ImageData -# -# -############################# L1NORM FUNCTIONS ############################# -#class SimpleL1Norm(Function): -# -# def __init__(self, alpha=1): -# -# super(SimpleL1Norm, self).__init__() -# self.alpha = alpha -# -# def __call__(self, x): -# return self.alpha * x.abs().sum() -# -# def gradient(self,x): -# return ValueError('Not Differentiable') -# -# def convex_conjugate(self,x): -# return 0 -# -# def proximal(self, x, tau): -# ''' Soft Threshold''' -# return x.sign() * (x.abs() - tau * self.alpha).maximum(0) -# -# def proximal_conjugate(self, x, tau): -# return x.divide((x.abs()/self.alpha).maximum(1.0)) - -#class L1Norm(SimpleL1Norm): -# -# def __init__(self, alpha=1, **kwargs): -# -# super(L1Norm, self).__init__() -# self.alpha = alpha -# self.b = kwargs.get('b',None) -# -# def __call__(self, x): -# -# if self.b is None: -# return SimpleL1Norm.__call__(self, x) -# else: -# return SimpleL1Norm.__call__(self, x - self.b) -# -# def gradient(self, x): -# return ValueError('Not Differentiable') -# -# def convex_conjugate(self,x): -# if self.b is None: -# return SimpleL1Norm.convex_conjugate(self, x) -# else: -# return SimpleL1Norm.convex_conjugate(self, x) + (self.b * x).sum() -# -# def proximal(self, x, tau): -# -# if self.b is None: -# return SimpleL1Norm.proximal(self, x, tau) -# else: -# return self.b + SimpleL1Norm.proximal(self, x - self.b , tau) -# -# def proximal_conjugate(self, x, tau): -# -# if self.b is None: -# return SimpleL1Norm.proximal_conjugate(self, x, tau) -# else: -# return SimpleL1Norm.proximal_conjugate(self, x - tau*self.b, tau) -# - -############################################################################### - - - - if __name__ == '__main__': from ccpi.framework import ImageGeometry import numpy - N, M = 40,40 + N, M = 400,400 ig = ImageGeometry(N, M) scalar = 10 - b = ig.allocate('random_int') - u = ig.allocate('random_int') + b = ig.allocate('random') + u = ig.allocate('random') f = L1Norm() f_scaled = scalar * L1Norm() - + f_b = L1Norm(b=b) f_scaled_b = scalar * L1Norm(b=b) - # call - + # call + a1 = f(u) a2 = f_scaled(u) numpy.testing.assert_equal(scalar * a1, a2) |