How can I output the structures of newgs.out?

Alloy Theoretic Automated Toolkit (ATAT)
1

Dear Prof. Axel van de Walle and ATAT users,

I’ve got 3 new unknown ground states in newgs.out after running maps. Could you please tell me how to output its structural information?

Thank you very much!
Cathy

2

These should be output next as you do touch ready
. New gs have priority to be written to disk over other structures.

3

Dear Prof. van de Walle,

Thanks for your kindly reply. After calculating these new ground states, maps.log shows that "true ground states differ from fitted ground states". My command line is "mags -g 16 &". Could you tell me how can I deal with this problem ?

Thank you very much!
Cathy

4

It could take a few iterations to resolve the problem (i.e. re-output the new predicted ground states and calculate them, etc.)
If the problem seems to take forever to correct itself, it is probably due to one of the following:

  1. You are computing ground states even in regions where that lattice is not actually stable. You may want to reduce the concentration range where ground state accuracy is enforced to only the region where that lattice is actually stable (using the -c0 and -c1 options in maps).
  2. Some structure relax away from the intended lattice. Type checkrelax and see if any of the number reported exceed about 0.1 . If so, try excluding the structure(s) that relaxed too much (touch nn/error for nn being the structure’s number). A cleaner solution can be found in https://dx.doi.org/10.1038/ncomms8559 and consist in replacing runstruct_vasp by robustrelax_vasp -c 0.05
5

Dear Prof. van de Walle,

Thanks for your kindly reply! Recently, I want to get a phase boundary by phb. And I need to specify the starting temperature. But I have no idea about how to choose the starting chemical potential -mu. The example you given is "-T=100 -mu=0.125". My command line is "phb -gs1=0 -gs2=3 -T=500 -mu=? -dT=50 -dx=1e-3 -er=50 -k=8.617e-5 -ltep=5e-3 -o=ph0.out".Could you give me some advice ?

Thank you very much!
Cathy

6

If you start at a low T (say, < 100K), you can let the code use its default value: it will use the ground state energies to find the mu,
If you start at a higher T, then either

  1. you can use the mu of a prior run that aborted. One common reason is when one phase appears or diappears. Say you start with -gs1=0 -gs=1 and the code eventually stops, noticing that gs 1 became unstable. Then you can continue a run with -gs1=0 -gs2=2 using the chemical potential from the last step of the previous run.
  2. You can use the mu that you have found by a emc2 run at the same T. Typically you would scan at constant T and vary mu in emc2. You then look for discontinuity in composition and use the mu where the discontinuity occured as the starting point for phb.