QuantumATK Forum
QuantumATK => General Questions and Answers => Topic started by: Inaoton on March 30, 2017, 07:25
-
Hello sir
I have run a script for calculating elastic constants of a bulk crystal with help of http://docs.quantumwise.com/tutorials/elastic_constants/elastic_constants.html . During optimization the lattice type changes from tetragonal to triclinic. After the completion of execution, 21 independent elastic constants came up.
1 .I want to calculate for the tetragonal lattice . what i can do for it?
2. Do i need change the force and stress tolerance values?
I have attached the script.
-
The change of lattice actually doesn't happen during the calculation of the elastic constants, but in the optimization before that. You have chosen not to constrain the Bravais lattice, so of course it doesn't. Change
constrain_bravais_lattice=False,
to
constrain_bravais_lattice=True,
Your tolerance values look fine, but if they are good enough can only be checked by making them a bit smaller and see if the geometry changes significantly as a result.
-
Hi
by setting constrain bravais lattice = True , I am getting the same answer as before
-
Can you share full output file?
-
Its a big file sir
-
Not able to upload
-
Compress it
-
Hello Sir,
Here is the output log file.Thank you
-
The file seems corrupt, please try again
-
Ok
-
It appears that you ran with constrain bravais lattice = False after all, because the optimized structure is "UnitCell".
Next time you check, use low settings, even just 3x3x3 k-points and the default (or higher) tolerance, and a very high force/stress tolerance, to make it finish quickly, to check what the result from the optimization is. If you see "UnitCell" in the log file, the symmetry was broken. Don't include the ElasticConstants part until you are sure of the optimization. Once that works, you don't need to redo it, just read the optimized configuration from the NC file and perform the ElasticConstants analysis on it.
-
Sir ,
It seems that i have sent you the previous result. Thank you for correcting that. I have run a code by setting constrain bravais lattice to true .
Like you said, the lattice type was conserved this time after optimization .But from the log file it seems that the lattice type changed as it starts
to calculate the elastic constants. The corresponding script and output files are attached. Can you please have a look?
-
You're doing a fine job here, but there are many details to consider (and some may be a bug or other kind of problem in ATK).
When you perform the force relaxation, the atoms move a bit (not so little, actually), so if you compare the original coordinates that were all 0.5 and 0.825, your relaxed ones are 0.49 and so on. That means, although the lattice symmetry remains BodyCenteredTetragonal, the space group symmetry has changed.
I made some simple tests, and just changing the position of the first atom (Zn) to x=0.00029 immediately causes the symmetry recognition to think the spacegroup is 1 = Triclinic.
We will investigate this a bit further, in particular if it would make sense to optimize the structure with conserved symmetry. But for now, all this means is that your calculation takes a bit longer, the results are in principle correct (note how small the off-diagonal elastic constants are, they can basically be ignored).
-
Thanks a lot Sir