--- a/pylearn/algorithms/mcRBM.py	Thu Oct 21 11:42:36 2010 -0400
+++ b/pylearn/algorithms/mcRBM.py	Thu Oct 21 11:43:36 2010 -0400
@@ -547,18 +547,21 @@
       - c - the mean bias (theano shared variable)
 
     """
-    def __init__(self, U, W, a, b, c, P):
+    # PCA eigenvectors (E) [scaled by 1/sqrt(eigenvals)] has to be passed to model
+    # because v has to be divided by the number original dimensions
+    # not the number of pca dimensions in hidden_cov_units_preactivation_given_v()
+    def __init__(self, U, W, a, b, c, P, norm_doctoring):
         self.P = P
+        self.norm_doctoring = norm_doctoring
         super(mcRBM_withP, self).__init__(U,W,a,b,c)
 
-    def hidden_cov_units_preactivation_given_v(self, v, small=0.5):
+    def hidden_cov_units_preactivation_given_v(self, v):
         """Return argument to the sigmoid that would give mean of covariance hid units
 
-        See the math at the top of this file for what 'adjusted' means.
-
-        return b - 0.5 * dot(adjusted(v), U)**2
         """
-        unit_v = v / (TT.sqrt(TT.mean(v**2, axis=1)+small)).dimshuffle(0,'x') # adjust row norm
+        slope, offset = self.norm_doctoring
+        norm_sq = TT.sum(v**2, axis=1) * slope + offset
+        unit_v = v / TT.sqrt(norm_sq).dimshuffle(0,'x') # adjust row norm
         return self.b + 0.5 * dot(dot(unit_v, self.U)**2, self.P)
 
     def n_hidden_cov_units(self):