Problem in sqs structure relaxation

Alloy Theoretic Automated Toolkit (ATAT)
1

Dear ATAT user,
I have generated SQS structure for Ni-Mn-Ga alloy. My rndstr.in is

5.82 5.82 5.82 90 90 90
0.0 0.5 0.5
0.5 0.0 0.5
0.5 0.5 0.0
0.0 0.0 0.0 Mn=0.5, Ni=0.5
0.5 0.5 0.5 Mn=0.5, Ni=0.5
0.25 0.25 0.25 Mn
0.75 0.75 0.75 Ga

Where ‘5.82’ is true lattice parameter

Then, I generated cluster expansion by
cprrdump -l=rndstr.in -ro -noe -nop -clus -2=8.75 -3=7 -4=7

mcsqs -n=32
Then, the SQS structure is relaxed by VASP and I got an error that ‘atoms are too close’

To solve the above problem I used lattice parameter larger than 5.82(true lattice parameter) as mentioned in the manual(common mistakes).
But same error message is showing.
How to solve this problem?

Also I have few confusion…
1)How to choose ‘maxradius’ for pair cluster generation and for triplet and …? (-clus -2=? -3=? …)
2)The atomic positions in bestsqs.out are Cartesian or Direct coordinate?
3)If I increase the lattice parameter to avoid the above mentioned problem then, how much should I increase?

Also, mcsqs is taking long time to generate 32 atom SQS.

2

I don’t see any problems in the files you sent. What do you vasp input file look like?

3

Thanks for your reply.
I have attached vasp input and output file. Please give answer of the below questions…
1)How to choose ‘maxradius’ for pair cluster generation and for triplet and …? (-clus -2=? -3=? …)
2)The atomic positions in bestsqs.out are Cartesian or Direct coordinate?
3)If I increase the lattice parameter to avoid the above mentioned problem then, how much should I increase?

Also, mcsqs is taking 5-6 days to generate 32 atom SQS. Am I doing anything wrong?

4

Sorry, no absolute rules here… For a given size of SQS that you consider manageable, as many correlations as possible should match the ones of the disordered state.

Please see mmaps -h for a description of the file format (all codes use the same structure format).
See also

The problem is that you used the keyword Direct instead of Cartesian in the POSCAR

The code should output some (perhaps not very good) SQS within minutes.
Refining may take hours.
If you see no output within minutes, there is another problem…