Excellent!!

Did you run an extensive parralel test, since those changes are quite deep in the code.
Otherwise I've some very minor comment/question:

1) madgraph/core/helas_objects.py
line 770 
is the comment:
# take the last index in case of identical particles
still makes sense?

2) madgraph/core/helas_objects.py
line 814 + 1537.

why not:
	mothers = self.get('mothers')
    if not mothers:
         return vertices
	for mother in mothers

3) madgraph/core/helas_objects.py
line: 1255
Add comment to explain what are the effect and the usefullness of my_pdg_code.
Should it have a default value (potentially to None or 0)?

4) madgraph/core/helas_objects.py
line: 1814
not beautifull indentation (not 4 spaces).
+ should specify the reason of the zero in  
mothers.sort_by_pdg_codes(self.get('pdg_codes'), 0)
either by putting my_pdg_code=0 or by removing it 
(and assign my_pdg_code a default value --my favourite options--)


5) This part can certainly be optimised in a better way. (That's not crucial)
@@ -246,32 +249,31 @@
1036	                 elif colors == [-3, 3]:
1037	                     # triplet-triplet-singlet coupling
1038	                     myinter.set('color', [color.ColorString(\
1039	-                        [color.T(1, 0)])])
1040	+                        [color.T(ind[1], ind[0])])])

Could you explain to me why you need to change this?


6)
+++ madgraph/various/process_checks.py	2010-11-29 00:15:14 +0000
1079	@@ -221,6 +221,9 @@
1080	     # Add this process to tested_processes
1081	     sorted_ids.append(ids)
1082	 
1083	+    # Skip adding antiprocess below, since might be relevant too
1084	+    return False
1085	+

Shouldn't we remove the code after the return False?

7) models/import_ufo
line 256:
Shouldn't we remove those lines? (or explain why they are commented)
        # Compute how change indices to match MG5 convention                                                                                                                             
        #info = [(i+1,part.color) for i,part in enumerate(interaction_info.particles)                                                                                                    
        #         if part.color!=1]                                                                                                                                                      
        #order = sorted(info, lambda p1, p2:p1[1] - p2[1])                                                                                                                               

        #new_indices={}                                                                                                                                                                  
	#for i,(j, pcolor) in enumerate(order):                                                                                                                                          
        #    new_indices[j]=i 
        
8) madgraph/core/color_algebra.py
line: 321

Basic verification that everything is normal, should be done in principle only in debug mode:
(except if you worried about wrong model input)
then I would suggest to replace 
+        if len(args) != 3:
92	+            raise ValueError, \
93	+                "Epsilon objects must have three indices!"
by 
      assert len(args) == 3 , "Epsilon objects must have three indices!"
by coherence, I would put it before any other command (i.e. here before the super)

9) Same comment for Epsilon_Bar/K6/... definition