fix demos

3 files changed, 77 insertions, 75 deletions

diff --git a/Wrappers/Python/ccpi/optimisation/functions/KullbackLeibler.py b/Wrappers/Python/ccpi/optimisation/functions/KullbackLeibler.py
index 51a008f..b53f669 100644
--- a/Wrappers/Python/ccpi/optimisation/functions/KullbackLeibler.py
+++ b/Wrappers/Python/ccpi/optimisation/functions/KullbackLeibler.py
@@ -105,18 +105,20 @@ class KullbackLeibler(Function):
             z = x + tau * self.bnoise
             return 0.5*((z + 1) - ((z-1)**2 + 4 * tau * self.b).sqrt())
         else:
-            z = x + tau * self.bnoise
-            out.fill( 0.5*((z + 1) - ((z-1)**2 + 4 * tau * self.b).sqrt()) )
-            #z_m = x + tau * self.bnoise -1
-            #self.b.multiply(4*tau, out=out)
-            #z_m.multiply(z_m, out=z_m)
-            #out += z_m
-            #out.sqrt(out=out)
-            #z_m.sqrt(out=z_m) 
-            #z_m += 2
-            #out *= -1
-            #out += z_m
-            #out *= 0.5
+#            z = x + tau * self.bnoise
+#            out.fill( 0.5*((z + 1) - ((z-1)**2 + 4 * tau * self.b).sqrt()) )
+                        
+            tmp1 = x + tau * self.bnoise - 1
+            tmp2 = tmp1 + 2
+            
+            self.b.multiply(4*tau, out=out)       
+            tmp1.multiply(tmp1, out=tmp1)
+            out += tmp1
+            out.sqrt(out=out)
+                        
+            out *= -1
+            out += tmp2
+            out *= 0.5
 
         
     
diff --git a/Wrappers/Python/wip/Demos/PDHG_TV_Denoising_Gaussian.py b/Wrappers/Python/wip/Demos/PDHG_TV_Denoising_Gaussian.py
index 1a3e0df..39bbb2c 100644
--- a/Wrappers/Python/wip/Demos/PDHG_TV_Denoising_Gaussian.py
+++ b/Wrappers/Python/wip/Demos/PDHG_TV_Denoising_Gaussian.py
@@ -32,7 +32,7 @@ from ccpi.optimisation.functions import ZeroFunction, L2NormSquared, \
 from skimage.util import random_noise
 
 # Create phantom for TV Gaussian denoising
-N = 300
+N = 100
 
 data = np.zeros((N,N))
 data[round(N/4):round(3*N/4),round(N/4):round(3*N/4)] = 0.5
@@ -46,9 +46,9 @@ n1 = random_noise(data.as_array(), mode = 'gaussian', mean=0, var = 0.05, seed=1
 noisy_data = ImageData(n1)
 
 # Regularisation Parameter
-alpha = 0.5
+alpha = 2
 
-method = '0'
+method = '1'
 
 if method == '0':
 
diff --git a/Wrappers/Python/wip/Demos/PDHG_TV_Denoising_Poisson.py b/Wrappers/Python/wip/Demos/PDHG_TV_Denoising_Poisson.py
index ccdabb2..4903c44 100644
--- a/Wrappers/Python/wip/Demos/PDHG_TV_Denoising_Poisson.py
+++ b/Wrappers/Python/wip/Demos/PDHG_TV_Denoising_Poisson.py
@@ -124,63 +124,63 @@ plt.show()
 
 #%% Check with CVX solution
 
-#from ccpi.optimisation.operators import SparseFiniteDiff
-#
-#try:
-#    from cvxpy import *
-#    cvx_not_installable = True
-#except ImportError:
-#    cvx_not_installable = False
-#
-#
-#if cvx_not_installable:
-#
-#    ##Construct problem    
-#    u1 = Variable(ig.shape)
-#    q = Variable()
-#    
-#    DY = SparseFiniteDiff(ig, direction=0, bnd_cond='Neumann')
-#    DX = SparseFiniteDiff(ig, direction=1, bnd_cond='Neumann')
-#    
-#    # Define Total Variation as a regulariser
-#    regulariser = alpha * sum(norm(vstack([DX.matrix() * vec(u1), DY.matrix() * vec(u1)]), 2, axis = 0))
-#    
-#    fidelity = sum( u1 - multiply(noisy_data.as_array(), log(u1)) )
-#    constraints = [q>= fidelity, u1>=0]
-#        
-#    solver = ECOS
-#    obj =  Minimize( regulariser +  q)
-#    prob = Problem(obj, constraints)
-#    result = prob.solve(verbose = True, solver = solver)
-#
-#    
-#    diff_cvx = numpy.abs( pdhg.get_output().as_array() - u1.value )
-#        
-#    plt.figure(figsize=(15,15))
-#    plt.subplot(3,1,1)
-#    plt.imshow(pdhg.get_output().as_array())
-#    plt.title('PDHG solution')
-#    plt.colorbar()
-#    plt.subplot(3,1,2)
-#    plt.imshow(u1.value)
-#    plt.title('CVX solution')
-#    plt.colorbar()
-#    plt.subplot(3,1,3)
-#    plt.imshow(diff_cvx)
-#    plt.title('Difference')
-#    plt.colorbar()
-#    plt.show()    
-#    
-#    plt.plot(np.linspace(0,N,N), pdhg.get_output().as_array()[int(N/2),:], label = 'PDHG')
-#    plt.plot(np.linspace(0,N,N), u1.value[int(N/2),:], label = 'CVX')
-#    plt.legend()
-#    plt.title('Middle Line Profiles')
-#    plt.show()
-#            
-#    print('Primal Objective (CVX) {} '.format(obj.value))
-#    print('Primal Objective (PDHG) {} '.format(pdhg.objective[-1][0]))
-#
-#
-#
-#
-#
+from ccpi.optimisation.operators import SparseFiniteDiff
+
+try:
+    from cvxpy import *
+    cvx_not_installable = True
+except ImportError:
+    cvx_not_installable = False
+
+
+if cvx_not_installable:
+
+    ##Construct problem    
+    u1 = Variable(ig.shape)
+    q = Variable()
+    
+    DY = SparseFiniteDiff(ig, direction=0, bnd_cond='Neumann')
+    DX = SparseFiniteDiff(ig, direction=1, bnd_cond='Neumann')
+    
+    # Define Total Variation as a regulariser
+    regulariser = alpha * sum(norm(vstack([DX.matrix() * vec(u1), DY.matrix() * vec(u1)]), 2, axis = 0))
+    
+    fidelity = sum( u1 - multiply(noisy_data.as_array(), log(u1)) )
+    constraints = [q>= fidelity, u1>=0]
+        
+    solver = ECOS
+    obj =  Minimize( regulariser +  q)
+    prob = Problem(obj, constraints)
+    result = prob.solve(verbose = True, solver = solver)
+
+    
+    diff_cvx = numpy.abs( pdhg.get_output().as_array() - u1.value )
+        
+    plt.figure(figsize=(15,15))
+    plt.subplot(3,1,1)
+    plt.imshow(pdhg.get_output().as_array())
+    plt.title('PDHG solution')
+    plt.colorbar()
+    plt.subplot(3,1,2)
+    plt.imshow(u1.value)
+    plt.title('CVX solution')
+    plt.colorbar()
+    plt.subplot(3,1,3)
+    plt.imshow(diff_cvx)
+    plt.title('Difference')
+    plt.colorbar()
+    plt.show()    
+    
+    plt.plot(np.linspace(0,N,N), pdhg.get_output().as_array()[int(N/2),:], label = 'PDHG')
+    plt.plot(np.linspace(0,N,N), u1.value[int(N/2),:], label = 'CVX')
+    plt.legend()
+    plt.title('Middle Line Profiles')
+    plt.show()
+            
+    print('Primal Objective (CVX) {} '.format(obj.value))
+    print('Primal Objective (PDHG) {} '.format(pdhg.objective[-1][0]))
+
+
+
+
+