Corrdump (mcsqs): How to determine the range of pairs, tripl

Alloy Theoretic Automated Toolkit (ATAT)
1

Dear all,

I would like to generate the disordered crystal structure of the sigma phase by using atat. However, I met a problem that when I choose different range of triplets to reach perfect match, I got the same atomic coordinates but different atomic species on the crystal sites. Is this normal? How to explain?
The parameter I select is as following:
(corrdump -l=rndstr.in -ro -noe -nop -clus -3=5)
(corrdump -l=rndstr.in -ro -noe -nop -clus -3=8)

Which triplet range do you think I should choose?
Thanks a lot for your help in advance!

The lattice file and output bestsqs.out are as following:
rndstr.in
9.70994 0 0
0 9.70994 0
0 0 4.99235
1 0 0
0 1 0
0 0 1
0.3967675469680074 0.3967675469680074 0.0000000000000000 Mo=0.5 Re=0.5
0.6032324530319926 0.6032324530319926 0.0000000000000000 Mo=0.5 Re=0.5
0.1032324530319926 0.8967675469680074 0.5000000000000000 Mo=0.5 Re=0.5
0.8967675469680074 0.1032324530319926 0.5000000000000000 Mo=0.5 Re=0.5
0.4640361093373215 0.1288191745165363 0.0000000000000000 Mo=0.5 Re=0.5
0.5359638906626785 0.8711808254834637 0.0000000000000000 Mo=0.5 Re=0.5
0.3711808254834637 0.9640361093373215 0.5000000000000000 Mo=0.5 Re=0.5
0.6288191745165363 0.0359638906626785 0.5000000000000000 Mo=0.5 Re=0.5
0.0359638906626785 0.6288191745165363 0.5000000000000000 Mo=0.5 Re=0.5
0.9640361093373215 0.3711808254834637 0.5000000000000000 Mo=0.5 Re=0.5
0.1288191745165363 0.4640361093373215 0.0000000000000000 Mo=0.5 Re=0.5
0.8711808254834637 0.5359638906626785 0.0000000000000000 Mo=0.5 Re=0.5
0.1824849424607962 0.1824849424607962 0.2526780385684688 Mo=0.5 Re=0.5
0.8175150575392109 0.8175150575392109 0.2526780385684688 Mo=0.5 Re=0.5
0.3175150575392038 0.6824849424607891 0.7526780385684688 Mo=0.5 Re=0.5
0.6824849424607891 0.3175150575392038 0.7526780385684688 Mo=0.5 Re=0.5
0.3175150575392038 0.6824849424607891 0.2473219614315312 Mo=0.5 Re=0.5
0.6824849424607891 0.3175150575392038 0.2473219614315312 Mo=0.5 Re=0.5
0.1824849424607962 0.1824849424607962 0.7473219614315312 Mo=0.5 Re=0.5
0.8175150575392109 0.8175150575392109 0.7473219614315312 Mo=0.5 Re=0.5
0.0000000000000000 0.0000000000000000 0.0000000000000000 Mo=0.5 Re=0.5
0.5000000000000000 0.5000000000000000 0.5000000000000000 Mo=0.5 Re=0.5
0.7403340999518946 0.0660324486047799 0.0000000000000000 Mo=0.5 Re=0.5
0.2596659000481054 0.9339675513952201 0.0000000000000000 Mo=0.5 Re=0.5
0.4339675513952201 0.2403340999518946 0.5000000000000000 Mo=0.5 Re=0.5
0.5660324486047799 0.7596659000481054 0.5000000000000000 Mo=0.5 Re=0.5
0.7596659000481054 0.5660324486047799 0.5000000000000000 Mo=0.5 Re=0.5
0.2403340999518946 0.4339675513952201 0.5000000000000000 Mo=0.5 Re=0.5
0.0660324486047799 0.7403340999518946 0.0000000000000000 Mo=0.5 Re=0.5
0.9339675513952201 0.2596659000481054 0.0000000000000000 Mo=0.5 Re=0.5

corrdump -l=rndstr.in -ro -noe -nop -clus -3=5
bestsqs.out
9.709940 0.000000 0.000000
0.000000 9.709940 0.000000
0.000000 0.000000 4.992350
1.000000 0.000000 0.000000
0.000000 1.000000 0.000000
0.000000 0.000000 1.000000
0.396768 0.396768 1.000000 Mo
0.603232 0.603232 1.000000 Re
0.103232 0.896768 0.500000 Mo
0.896768 0.103232 0.500000 Re
0.128819 0.464036 1.000000 Re
0.464036 0.128819 1.000000 Re
0.535964 0.871181 1.000000 Re
0.871181 0.535964 1.000000 Re
0.035964 0.628819 0.500000 Mo
0.371181 0.964036 0.500000 Mo
0.628819 0.035964 0.500000 Mo
0.964036 0.371181 0.500000 Mo
0.182485 0.182485 0.252678 Mo
0.182485 0.182485 0.747322 Mo
0.817515 0.817515 0.252678 Mo
0.817515 0.817515 0.747322 Mo
0.317515 0.682485 0.752678 Re
0.317515 0.682485 0.247322 Re
0.682485 0.317515 0.752678 Re
0.682485 0.317515 0.247322 Re
1.000000 1.000000 1.000000 Re
0.500000 0.500000 0.500000 Mo
0.066032 0.740334 1.000000 Re
0.740334 0.066032 1.000000 Mo
0.259666 0.933968 1.000000 Mo
0.933968 0.259666 1.000000 Re
0.759666 0.566032 0.500000 Re
0.433968 0.240334 0.500000 Mo
0.566032 0.759666 0.500000 Mo
0.240334 0.433968 0.500000 Re

corrdump -l=rndstr.in -ro -noe -nop -clus -3=8
bestsqs_02.out
9.709940 0.000000 0.000000
0.000000 9.709940 0.000000
0.000000 0.000000 4.992350
1.000000 0.000000 0.000000
0.000000 1.000000 0.000000
0.000000 0.000000 1.000000
0.396768 0.396768 1.000000 Re
0.603232 0.603232 1.000000 Mo
0.103232 0.896768 0.500000 Mo
0.896768 0.103232 0.500000 Re
0.128819 0.464036 1.000000 Re
0.464036 0.128819 1.000000 Mo
0.535964 0.871181 1.000000 Re
0.871181 0.535964 1.000000 Mo
0.035964 0.628819 0.500000 Re
0.371181 0.964036 0.500000 Mo
0.628819 0.035964 0.500000 Re
0.964036 0.371181 0.500000 Mo
0.182485 0.182485 0.252678 Re
0.182485 0.182485 0.747322 Re
0.817515 0.817515 0.252678 Re
0.817515 0.817515 0.747322 Mo
0.317515 0.682485 0.752678 Mo
0.317515 0.682485 0.247322 Mo
0.682485 0.317515 0.752678 Re
0.682485 0.317515 0.247322 Mo
1.000000 1.000000 1.000000 Mo
0.500000 0.500000 0.500000 Re
0.066032 0.740334 1.000000 Re
0.740334 0.066032 1.000000 Re
0.259666 0.933968 1.000000 Mo
0.933968 0.259666 1.000000 Mo
0.759666 0.566032 0.500000 Mo
0.433968 0.240334 0.500000 Mo
0.566032 0.759666 0.500000 Re
0.240334 0.433968 0.500000 Re

diff bestsqs.out bestsqs_02.out
7,8c7,8
< 0.396768 0.396768 1.000000 Mo
< 0.603232 0.603232 1.000000 Re

> 0.396768 0.396768 1.000000 Re
> 0.603232 0.603232 1.000000 Mo
12c12
< 0.464036 0.128819 1.000000 Re

> 0.464036 0.128819 1.000000 Mo
14,15c14,15
< 0.871181 0.535964 1.000000 Re
< 0.035964 0.628819 0.500000 Mo

> 0.871181 0.535964 1.000000 Mo
> 0.035964 0.628819 0.500000 Re
17c17
< 0.628819 0.035964 0.500000 Mo

> 0.628819 0.035964 0.500000 Re
19,21c19,21
< 0.182485 0.182485 0.252678 Mo
< 0.182485 0.182485 0.747322 Mo
< 0.817515 0.817515 0.252678 Mo

> 0.182485 0.182485 0.252678 Re
> 0.182485 0.182485 0.747322 Re
> 0.817515 0.817515 0.252678 Re
23,24c23,24
< 0.317515 0.682485 0.752678 Re
< 0.317515 0.682485 0.247322 Re

> 0.317515 0.682485 0.752678 Mo
> 0.317515 0.682485 0.247322 Mo
26,28c26,28
< 0.682485 0.317515 0.247322 Re
< 1.000000 1.000000 1.000000 Re
< 0.500000 0.500000 0.500000 Mo

> 0.682485 0.317515 0.247322 Mo
> 1.000000 1.000000 1.000000 Mo
> 0.500000 0.500000 0.500000 Re
30c30
< 0.740334 0.066032 1.000000 Mo

> 0.740334 0.066032 1.000000 Re
32,33c32,33
< 0.933968 0.259666 1.000000 Re
< 0.759666 0.566032 0.500000 Re

> 0.933968 0.259666 1.000000 Mo
> 0.759666 0.566032 0.500000 Mo
35c35
< 0.566032 0.759666 0.500000 Mo

> 0.566032 0.759666 0.500000 Re

2

The generation algorithm is random, so it’s perfectly normal that you could get different occupations in two different runs.
As explained in the manual, you don’t need to get "perfect match" to find a correct solution.
The triplet range you need is situation-dependent. it depends how big an SQS you’re willing to use and how long-range you expect multi-body interaction to be. You can check convergence of the properties you calculate with respect to sqs size.
However: do you really want no pairs??

3

Both are equally good in terms of the criteria you specify - it’s just that there is more than one possible solution and the code selects one at random.

You could compute longer range correlations and/or three-body correlations to rank them more precisely, but it may not be necessary here.

4

dear all,

when I try twice the following commands with same input file:
corrdump -l=rndstr.in -ro -noe -nop -clus -2=8
mcsqs -2d -n=32
I do always get perfect_match in bestcorr.out, but the bestsqs.out for the first time and the second time is different. is that normal?
if it is normal, how to tell which one is good?