Dev arsc ica

src/cedalion/sigdecomp/multimodal/arc_ebm.py

@@ -17,14 +17,14 @@
 from cedalion import cite
 
 def arc_ebm(X: np.ndarray, guess_mat, constraint = 'correlation') -> np.ndarray:
-    """Adaptive-reverse Constrained ICA by Entropy Bound Minimization (arc-EBM) is a constrained ICA algorithm.
-        arc-EBM calculates the blind source separation demixing matrix corresponding to X, 
+    """Adaptive-reverse (Spectrally) Constrained ICA by Entropy Bound Minimization (arc-EBM/arsc-EBM) is a constrained ICA algorithm.
+        arc-EBM/arsc-EBM calculates the blind source separation demixing matrix W corresponding to data X, 
         using the reference signals in guess_mat and the constraint specified by constraint.
 
     Args:
         X (np.ndarray, (Channels, Time Points)): the [N x T] input multivariate time series with dimensionality N observations/channels and T time points
         guess_mat (np.ndarray, (Time Points, Referenced Channels)), (np.ndarray, (Time Points/2, Referenced Channels)): Time or frequency domain reference signals. The number of reference signals should be less than or equal to the number of channels in X. The first dimension should be T for time domain signals and T/2 for frequency domain signals.
-        constraint (str): the constraint to be used for the gradient step, either 'correlation' (default) or 'psd'
+        constraint (str): the constraint to be used for the gradient step, either 'correlation' (default) for arc-EBM or 'psd' for arsc-EBM 
 
     Returns:
         W (np.ndarray, (Channels, Channels)): the [N x N] demixing matrix with weights for  N channels/sources. 

src/cedalion/sigdecomp/multimodal/arc_erbm.py → src/cedalion/sigdecomp/multimodal/arsc_erbm.py

@@ -12,8 +12,8 @@
 import cedalion.data
 from cedalion import cite
 
-def arc_erbm(X: np.ndarray, guess_mat, p: int = None , pr_guess_mat = None) -> np.ndarray:
-    """ Adaptive-reverse Constrained ICA by Entropy Rate Bound Minimization (arc-ERBM) is a spectrally constrained ICA algorithm. 
+def arsc_erbm(X: np.ndarray, guess_mat, p: int = None , pr_guess_mat = None) -> np.ndarray:
+    """ Adaptive-reverse Spectrally Constrained ICA by Entropy Rate Bound Minimization (arsc-ERBM) is a constrained ICA algorithm. 
 
     Args:
         X (np.ndarray, (Channels, Time Points)): the [N x T] input multivariate time series with dimensionality N observations/channels and T time points
@@ -106,7 +106,7 @@ def epsilon_grad(r_n_c):
         # compute gradient vector
         vec = (sign / np.linalg.norm(psd_s, 2)) * r_n_c -  (np.abs(current_corr) / np.linalg.norm(psd_s, 2)**2) * psd_s 
         vec = vec.reshape((-1, 1, 1))     
-        c_grad  = (T/2) * mu_c[n] * np.sum( np.multiply(np.dot(C_tilde,  w), vec), axis = 0  )    
+        c_grad  = (T/2) * mu_c[n] * np.sum( np.multiply(np.dot(C_tilde,  w), vec), axis = 0 )
         return c_grad
 
     if constraint == 'phase_retrieval' : 
@@ -117,7 +117,7 @@ def pr_update(amp, y, filter, X_filtered):
 
             # filter reference amplitude
             amp_filtered =  sp.signal.lfilter(filter, 1, amp , axis = 0 )
-            amp_filtered = np.abs(np.fft.rfft(amp_filtered)) 
+            amp_filtered = np.abs(np.fft.rfft(amp_filtered))
 
             # compute fft of estimated source
             y_hat = np.fft.rfft(y)