Dear Axel,
First of all, let me thanks again for the recent correction you made on "xtalutil.c++",
that fixed a bug which made "corrdump" able to find now all clusters of rank 4 (quadruplet) in the C15-phase.
I am back again with this C15-phase, where you can find the 'lat.in" of the full-random primitive cell below:
4.9497 4.9497 4.9497 60 60 60
1.0 0.0 0.0
0.0 1.0 0.0
0.0 0.0 1.0
0.375000000 0.375000000 0.375000000 Cu,Mg
0.625000000 0.625000000 0.625000000 Cu,Mg
0.000000000 0.000000000 0.000000000 Cu,Mg
0.500000000 0.000000000 0.000000000 Cu,Mg
0.000000000 0.500000000 0.000000000 Cu,Mg
0.000000000 0.000000000 0.500000000 Cu,Mg
As you know,
AB2-C15 is a FCC cell with two non-equivalent sites,
MgCu2 its prototype, SG: Fd-3m (227) with 2 origin descriptions.
Here, we choose A=Cu,Mg on 8a (1/8,1/8,1/8) and B=Cu,Mg on 16d (0,0,0),
with origin 1 + 1/8 shift description, cubic cell parameter, a=7.
By asking the command :
corrdump -l=lat.in -nop -clus -3=3
2 triplets, with a min distance of 2.90202 are found: one with a multiplicity 24 and the other with 12.
If we see the detail of these clusters in "cluster.out", we can see that, in fact, this 2 clusters are precisely the same: The triplet A-B-B:
24
2.90202
3
1.00000 1.00000 0.50000 0 0
1.00000 0.50000 1.00000 0 0
0.62500 0.62500 0.62500 0 0
12
2.90202
3
1.00000 1.00000 0.50000 0 0
1.00000 1.00000 0.00000 0 0
0.62500 0.62500 0.62500 0 0
You may answer that (1.00000 0.50000 1.00000) are different than (1.00000 1.00000 0.00000),
since an atom of the center of the edge of the cell is different that in vertices position,
but looking carefully the structure, the "B-16d" site positions are exactly equivalent in term of coordination, distances and global environment.
Please check on the crystal structure attached.
Pair distances are equal and angles too:
phi(Cu1-Mg2-Cu2) = 50.4788(0) deg. phi(Cu4-Mg2-Cu2) = 50.4788(0) deg.
[b]My questions are :
- How it is possible that the same cluster A-B-B is considered twice?[/b]
- By doing CEM calculations on 60 structures with CV<0.035, I found that the associated ECI are of opposite sign! How identical crystallographic clusters could have opposite sign ECI energy?
Thus I think there is a small error in the crystal detection for complex structures, as the C15 is.
Let me know if the problem is clear enough and if I could help you in a way to solve this bug.
Many Many thanks in advance,
JC.