Process tutorial notebooks

tutorials/W2D3_DecisionMaking/solutions/W2D3_Tutorial2_Solution_29c675cb.py → tutorials/W2D3_DecisionMaking/solutions/W2D3_Tutorial2_Solution_099bf548.py

@@ -5,7 +5,7 @@ def m_step(gamma, xi, dt):
     gamma ():       Number of epochs of EM to run
     xi (numpy 3d array): Tensor of recordings, has shape (n_trials, T, C)
     dt (float):         Duration of a time bin
-          
+
   Returns:
     psi_new (numpy vector): Updated initial probabilities for each state
     A_new (numpy matrix):   Updated transition matrix, A[i,j] represents the

tutorials/W2D3_DecisionMaking/solutions/W2D3_Tutorial2_Solution_4fe44495.py → tutorials/W2D3_DecisionMaking/solutions/W2D3_Tutorial2_Solution_94c41b77.py

@@ -3,7 +3,7 @@
 switch_prob = 0.1
 log10_noise_level = -1
 
-# Build model 
+# Build model
 model = create_model(switch_prob=switch_prob,
                      noise_level=10.**log10_noise_level,
                      startprob=[0.5, 0.5])

tutorials/W2D3_DecisionMaking/solutions/W2D3_Tutorial3_Solution_00db9a46.py → tutorials/W2D3_DecisionMaking/solutions/W2D3_Tutorial3_Solution_a0f4822b.py

@@ -3,7 +3,7 @@ def kalman_smooth(data, params):
   system parameters.
 
   Args:
-    data (ndarray): a sequence of osbervations of shape(n_timesteps, n_dim_obs)    
+    data (ndarray): a sequence of osbervations of shape(n_timesteps, n_dim_obs)
     params (dict): a dictionary of model paramters: (F, Q, H, R, mu_0, sigma_0)
 
   Returns:
@@ -17,7 +17,7 @@ def kalman_smooth(data, params):
 
   n_dim_state = F.shape[0]
   n_dim_obs = H.shape[0]
-  
+
   # first run the forward pass to get the filtered means and covariances
   mu, sigma = kalman_filter(data, params)
 
@@ -26,7 +26,7 @@ def kalman_smooth(data, params):
   sigma_hat = np.zeros_like(sigma)
   mu_hat[-1] = mu[-1]
   sigma_hat[-1] = sigma[-1]
-  
+
   # smooth the data
   for t in reversed(range(len(data)-1)):
     sigma_pred = F @ sigma[t] @ F.T + Q  # sigma_pred at t+1
@@ -36,7 +36,7 @@ def kalman_smooth(data, params):
     # write the expression to compute the smoothed state mean estimate
     mu_hat[t] = mu[t] + J @ (mu_hat[t+1] - F @ mu[t])
     # write the expression to compute the smoothed state noise covariance estimate
-    sigma_hat[t] = sigma[t] + J @ (sigma_hat[t+1] - sigma_pred) @ J.T 
+    sigma_hat[t] = sigma[t] + J @ (sigma_hat[t+1] - sigma_pred) @ J.T
 
   return mu_hat, sigma_hat
 

tutorials/W2D3_DecisionMaking/solutions/W2D3_Tutorial3_Solution_bd7cff3e.py → tutorials/W2D3_DecisionMaking/solutions/W2D3_Tutorial3_Solution_c05a2da8.py

@@ -9,7 +9,7 @@ def sample_lds(n_timesteps, params, seed=0):
 
   Returns:
   ndarray, ndarray: the generated state and observation data
-  """    
+  """
   n_dim_state = params['F'].shape[0]
   n_dim_obs = params['H'].shape[0]
 
@@ -32,7 +32,7 @@ def sample_lds(n_timesteps, params, seed=0):
       state[t] = params['mu_0']
     else:
       state[t] = params['F'] @ state[t-1] + zi[t]
-    
+
     # write the expression for computing the observation
     obs[t] = params['H'] @ state[t] + eta[t]
 

tutorials/W2D3_DecisionMaking/solutions/W2D3_Tutorial3_Solution_9a22a0fa.py → tutorials/W2D3_DecisionMaking/solutions/W2D3_Tutorial3_Solution_e9df5afe.py

@@ -3,7 +3,7 @@ def kalman_filter(data, params):
   system parameters.
 
   Args:
-    data (ndarray): a sequence of osbervations of shape(n_timesteps, n_dim_obs)    
+    data (ndarray): a sequence of osbervations of shape(n_timesteps, n_dim_obs)
     params (dict): a dictionary of model paramters: (F, Q, H, R, mu_0, sigma_0)
 
   Returns:
@@ -19,7 +19,7 @@ def kalman_filter(data, params):
   n_dim_obs = H.shape[0]
   I = np.eye(n_dim_state)  # identity matrix
 
-  # state tracking arrays  
+  # state tracking arrays
   mu = np.zeros((len(data), n_dim_state))
   sigma = np.zeros((len(data), n_dim_state, n_dim_state))
 
@@ -34,15 +34,15 @@ def kalman_filter(data, params):
 
     # write the expression for computing the Kalman gain
     K = sigma_pred @ H.T @ np.linalg.inv(H @ sigma_pred @ H.T + R)
-    # write the expression for computing the filtered state mean 
+    # write the expression for computing the filtered state mean
     mu[t] = mu_pred + K @ (y - H @ mu_pred)
     # write the expression for computing the filtered state noise covariance
     sigma[t] = (I - K @ H) @ sigma_pred
-  
+
   return mu, sigma
 
 
 filtered_state_means, filtered_state_covariances = kalman_filter(obs, params)
-with plt.xkcd():    
+with plt.xkcd():
   plot_kalman(state, obs, filtered_state_means, title="my kf-filter",
               color='r', label='my kf-filter')