add toy examples

compute_distance.py

@@ -5,12 +5,12 @@
 
 from time import time
 
-def distance(D1, D2, gamma, niter, data_path, verbose = 0):
+def distance(D1, D2, lambd, niter, data_path, verbose = 0):
     """
     calculates all pairwise distances d(d1,d2) with:
         - d1 in D1 = [d11, d12, ...]
         - d2 in D2 = [d21, d22, ...]
-        - gamma: entropic regularization parameter
+        - lambd: entropic regularization parameter
         - niter: number of iterations of Sinkhorn's algorithm
         - data_path: path to data files
         - verbose = {0: display nothing, 1: display intermedidate computation times}
@@ -84,7 +84,7 @@ def distance(D1, D2, gamma, niter, data_path, verbose = 0):
             # define graphs
             init = tf.initialize_all_variables()
 
-            op_xi = tf.exp(tf.scalar_mul(-1./gamma,tf.pow(xi, 2)))
+            op_xi = tf.exp(tf.scalar_mul(-1./lambd,tf.pow(xi, 2)))
             update_xi = tf.assign(xi, op_xi)
 
             op_A = tf.div(P, tf.matmul(B, xi))