# HG changeset patch
# User Jordi Gutiérrez Hermoso <jordigh@octave.org>
# Date 1421418464 18000
# Node ID 077261db7a584d9d217918aad9c01505ea6b851e
# Parent  c81f6d26389752e3865e7062eba24bc74bd8d420
Whitespace cleanup

diff --git a/pymedcouple b/pymedcouple
--- a/pymedcouple
+++ b/pymedcouple
@@ -11,15 +11,15 @@
     Also known as quickselect. The elements of W are ignored for the
     purposes of this partial sort.
     """
-    
+
     beg = 0
     end = len(L)
-    
+
     while True:
         pivot = random.randint(beg,end-1)
         pivotval = L[pivot][0]
         L[pivot], L[end-1] = L[end-1], L[pivot]
-        
+
         idx = beg
         for i in xrange(beg, end):
             if L[i][0] < pivotval:
@@ -39,7 +39,7 @@
     """This computes the weighted median of array A with corresponding
     weights W.
     """
-    
+
     AW = zip(A,W)
     n = len(AW)
 
@@ -50,7 +50,7 @@
 
     while True:
         mid = (beg + end)//2
-        
+
         partsort(AW, mid)
         trial = AW[mid][0]
 
@@ -81,7 +81,7 @@
     -------
     mc : float
         The medcouple statistic
-   
+
     .. [1] G. Brys, M. Hubert, and A. Struyf "A Robust Measure of
     Skewness." Journal of Computational and Graphical Statistics, Vol.
     13, No. 4 (Dec., 2004), pp. 996- 1017
@@ -92,7 +92,7 @@
 
     """
     # FIXME: Figure out what to do about NaNs.
-    
+
     n = len(X)
     n2 = (n-1)//2
 
@@ -105,7 +105,7 @@
         Zmed = Z[n2]
     else:
         Zmed = (Z[n2] + Z[n2+1])/2
-    
+
     #Check if the median is at the edges up to relative epsilon
     if abs(Z[0] - Zmed) < eps1*(eps1 + abs(Zmed)):
         return -1.0
@@ -119,7 +119,7 @@
     Zden = 2*max(Z[0], -Z[-1])
     Z = [z/Zden for z in Z]
     Zmed /= Zden
-    
+
     Zeps = eps1*(eps1 + abs(Zmed))
 
     # These overlap on the entries that are tied with the median
@@ -157,7 +157,7 @@
 
     # kth pair algorithm (Johnson & Mizoguchi)
     while Rtot - Ltot > n_plus:
-       
+
         # First, compute the median inside the given bounds
         # (Be stingy, reuse same generator)
         [I1, I2] = tee(i for i in xrange(0, n_plus) if L[i] <= R[i])
@@ -165,7 +165,7 @@
         A = [h_kern(i, (L[i] + R[i])//2) for i in I1]
         W = [R[i] - L[i] + 1 for i in I2]
         Am = wmedian(A,W)
-            
+
         Am_eps = eps1*(eps1 + abs(Am))
 
         # Compute new left and right boundaries, based on the weighted
@@ -178,7 +178,7 @@
             while j < n_minus - 1 and h_kern(i, j) - Am > Am_eps:
                 j += 1
             P.append(j - 1)
-            
+
         P.reverse()
 
         j = n_minus - 1
@@ -215,7 +215,7 @@
     Am =  A[medc_idx - Ltot]
 
     return Am
-    
+
 
 def signum(x):
     if x > 0: