3
$\begingroup$

I'd like to generate a supercell of Cd0.5Zn0.5S which includes 108 atoms, and I tried to use mpirun -np 64 mcsqs -n 108 to get the supercell, but according to the log of mcsqs, the program runs very slowly, how could I make the program run faster?

Here is the log of mcsqs:

OObjective_function= -0.944454
Correlations_mismatch=  0.000000    0.000000    0.000000    0.000000    0.000000    0.111111    -0.111111   -0.111111   -0.111111   -0.111Objective_function= -0.944454
Correlations_mismatch=  0.000000    0.000000    0.000000    0.000000    0.000000    -0.111111   -0.111111   0.111111    -0.111111   -0.111111
0.111111    0.111111
454
Correlations_mismatch=  0.000000    0.000000    0.000000    0.000000    0.000000    0.111111    0.111111    -0.111111   -0.111111   -0.111111
11  0.111111
Objective_function= -0.944454
Correlations_mismatch=  0.000000    0.000000    0.000000    0.000000    0.000000    0.111111    -0.111111   -0.111111   -0.111111   -0.111111

Here is the result I got after running mcsqs for a few days which doesn't seems quite good: enter image description here

$\endgroup$

1 Answer 1

4
$\begingroup$

mcsqs is not supported by mpi. Each process of mcsqs can be run on only one core. However, as a Monte Carlo method, you can still parallelize. You can concurrently run multiple processes on different cores, but with each process having a different random seed.

For example,

mcsqs -n=108 -ip=1 -sd=933009 &
mcsqs -n=108 -ip=2 -sd=112980 &
.
.
.

mcsqs -n=108 -ip=64 -sd=933009 &

Extra resource: How to use SQS for disordered materials

$\endgroup$
1
  • $\begingroup$ do I need to create a folder for each macaws process or I could run these process in the same folder? $\endgroup$
    – Jack
    Aug 21, 2022 at 18:18

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .