Nscore transform with conditional points

Additional_examples/example_nscore_transformation.py

@@ -13,6 +13,7 @@
 # data
 x = np.loadtxt('./Data/data1.dat')
 reference_log = x[:,1].reshape(-1, 1)
+n_ptos = reference_log.shape[0]
 
 # NSCORE TRANSFORMATION - Transform th non-parametric reference log/data to a Gaussian distributed varible
 referencelog_uniform = NonParametricToUniform(reference_log, reference_log)
@@ -27,12 +28,20 @@
 axs[2].hist(referencelog_gaussian[:,0], bins=15)
 
 # NSCORE BACK TRANSFORMATION - Sequential Gaussian Simulation of a non-parametric distribution
-xcoords = np.arange(reference_log.shape[0]).reshape((reference_log.shape[0],1))    
-cond_depth = np.array([]).reshape((0,1))    
-cond_values = np.array([]).reshape((0,1))    
+xcoords = np.arange(reference_log.shape[0]).reshape((reference_log.shape[0],1))
+
+n_ptos_cond = 20;    
+dcoords = np.random.choice(n_ptos, n_ptos_cond ,replace=False).reshape((n_ptos_cond,1))     
+cond_depth = xcoords[dcoords].reshape((n_ptos_cond,1))    
+cond_values = reference_log[dcoords].reshape((n_ptos_cond,1))    
+
+# Transform conditional points from non parametric to uniform
+cond_values_uniform = NonParametricToUniform(cond_values, reference_log)
+cond_values_gaussian = norm.ppf(cond_values_uniform) 
+
 
 # Simulate a Gaussian simulation with zero mean and std 1
-simulation_gaussian = SeqGaussianSimulation(xcoords, cond_depth, cond_values, 0.0, 1.0, 10.0, 'exp', 0)                      
+simulation_gaussian = SeqGaussianSimulation(xcoords, cond_depth, cond_values_gaussian, 0.0, 1.0, 20.0, 'sph', 0)                      
 # Transform from Gaussian to uniform
 simulation_uniform = norm.cdf(simulation_gaussian)
 # Then, transform from uniform to target non parametric distribution
@@ -44,6 +53,11 @@
 plt.plot(simulation_uniform)
 plt.plot(simulation_non_parametric)
 
+plt.figure()
+plt.plot(xcoords,reference_log)
+plt.plot(xcoords,simulation_non_parametric)
+plt.plot(cond_depth,cond_values,'o')
+
 fig, axs = plt.subplots(1, 3)
 axs[0].hist(simulation_gaussian[:,0], bins=15)
 axs[1].hist(simulation_uniform[:,0], bins=15)