Problem about fitsvsl

Alloy Theoretic Automated Toolkit (ATAT)
1

Dear Axel,

I’m trying to calculate the vibration contribution to the phase diagram of ZnOS (S partially substituing O) Alloy. After using maps and quantum espresso to obtain CV=7 meV, fitsvsl -ms=0.01 -ns=3 -er=5.8 was employed to generate perturbed structures and then the str.out files in vol_/p_ were calculated in self-consistent calculation without any relax. The force.out were extracted from the output file of quantum espresso. And the str_relax.out files in p_* were copied from str.out since no any relax was done. After that, fitsvsl -f was employed to fit the bond stiffness.

The strange thing is that I get ZERO stiffness in f_O-Zn.dat and f_S-Zn.dat. Although slspring.out gives
O Zn
2
129.18204
-44.64744
2
-55.52997
22.84399
S Zn
2
125.36527
-44.07142
2
-67.31664
26.20792

What happened? Which step did I do wrong? All steps I followed by the article CALPHAD 33(2009)266-278

BTW, should the unit in force.out be eV/A or any other one?

2

yes units of force.out are eV/A
Can you send me your files? (As a .tar excluding large ab initio output files)

3

Thank you for your reply quickly. I attach the all files except ab initio output files. z-ZnOS.tar is actually z-ZnOS.tar.gz.

4

I looked at your force.out files. The forces seem a bit too large (if you used the default displacement of 0.2 Ang) by a factor 10 or 100.
The force constant output are 0 because the code determined that the variance (in their least square fit) was too large so the values would be unreliable. Of course, the code was confused because the scale of your force constants is ~10 times too large. So first check your unit conversion perhaps? If you still have trouble, then change the -eqt parameter (from the default of 2.0): this is the variance cutoff to decide of the code outputs the fc or not.

5

Thank you for your help.

After trying various command parameters, the problem wa still there. So I check the str_relax.out and found the disagreement with the output of ab init code, which was induced by a bug in the script converting the output of ab init code to str_relax.out file. After fixing the bug, it seems to be good. The force after perturbation becomes small. Although all calculations are not finished yet. I hope everything should be all right.

6

Hi
All ATAT users
I am using ATAT software for my research work with Quantum-Espresso as first-principle code, but unable to generate the str_relax.out file after completion of energy calculation. I am using runstruct file as provided with ATAT toolkit.
please help me.