Dear Ashwani Kushwaha ,

after your run this input with QE, you should have the forces listed per atom in the output. Is this not the case? What do you get? It is possible that for highly symmetric structures you get 0. In this case you need to shift one atom by hand to remove the symmetry and run a new calculation.

Further notice that you have applied a 5% strain in this input as (3.72-3.52)/3.52=0.05.

Regards,

Roberto

Hello Sir,

Basically I want to know which method should I use like "scf" or "vc-relax" and during "vc-relax" I only allow the crystal structure to relax in Y-Z direction as strain is applied in X direction.

I performed the two types of calculations in one calculation I just changed the lattice parameter and another I also re-arranged the atom position with according to the new lattice parameter with the help of a tool provided by the ASE-python library.

Here is the input file with only change in the lattice parameter(please see the attched file named with "onlyLatticeChange.png" graph between the lattice parameter and energy):

&CONTROL

calculation = 'scf'

verbosity = 'high'

tstress = .true.

tprnfor = .true.

prefix = 'myprefix'

pseudo_dir = '/home/ashwani/PP/PBE_ONCV'

/

&SYSTEM

ecutwfc = 50

ecutrho = 400

occupations = 'smearing'

degauss = 0.05

smearing = 'methfessel-paxton'

nspin = 1

ntyp = 1

nat = 4

ibrav = 0

/

&ELECTRONS

electron_maxstep = 200

conv_thr = 1e-08

mixing_mode = 'plain'

mixing_beta = 0.7

diagonalization = 'david'

/

&IONS

/

&CELL

/

ATOMIC_SPECIES

Ni 58.6934 Ni.UPF

K_POINTS automatic

6 6 6 0 0 0

CELL_PARAMETERS angstrom

4.12 0.00000000000000 0.00000000000000

0.00000000000000 4.12 0.00000000000000

0.00000000000000 0.00000000000000 4.12

ATOMIC_POSITIONS crystal

Ni 0.0000000000 0.0000000000 0.0000000000

Ni 0.0000000000 0.5000000000 0.5000000000

Ni 0.5000000000 0.0000000000 0.5000000000

Ni 0.5000000000 0.5000000000 0.0000000000

Here is the input file with change in Lattice parameter and atoms position as well(please see the attched file named with "ase.png" graph between the lattice parameter and energy):

&CONTROL

calculation = 'scf'

verbosity = 'high'

tstress = .true.

tprnfor = .true.

prefix = 'myprefix'

pseudo_dir = '/home/ashwani/PP/PBE_ONCV'

/

&SYSTEM

ecutwfc = 50

ecutrho = 400

occupations = 'smearing'

degauss = 0.05

smearing = 'methfessel-paxton'

nspin = 1

ntyp = 1

nat = 4

ibrav = 0

/

&ELECTRONS

electron_maxstep = 200

conv_thr = 1e-08

mixing_mode = 'plain'

mixing_beta = 0.7

diagonalization = 'david'

/

&IONS

/

&CELL

/

ATOMIC_SPECIES

Ni 58.6934 Ni.UPF

K_POINTS automatic

6 6 6 0 0 0

CELL_PARAMETERS angstrom

4.12000000000000 0.00000000000000 0.00000000000000

0.00000000000000 4.12000000000000 0.00000000000000

0.00000000000000 0.00000000000000 4.12000000000000

ATOMIC_POSITIONS crystal

Ni 0.0000000000 0.0000000000 0.0000000000

Ni 0.0000000000 0.4271844660 0.4271844660

Ni 0.4271844660 0.0000000000 0.4271844660

Ni 0.4271844660 0.4271844660 0.0000000000

If you want to investigate the Poisson coefficient, you can use a vc-relax where only the direction y and z are allowed to change. The atomic positions will be adjusted by the algorithm. I would not be too worried about the atomic positions as they will most likely change during the vc-relax steps.

However, in your input you have expanded the unit cell in all 3 directions. Why? If you apply strain along x you could expect a contraction along y and z. I would expand only along x and let the vc-relax find the parameter in the other directions while relaxing the atomic forces.

I hope this helps,

Roberto

According to you I can use both the methods to find the forces "scf" and "vc-relax" according to the need.

I am finally going to use this data to train my SNAP potential model and further going to use this potential in LAMMPS MD simulations.

So which method will be good "scf" or "vc-relax". And one more question if I am going to use the "scf" do I need to care about the atom positions because in "vc-relax" it will be taken care by the software itself but what about the "scf" calculations.In scf calculation if there is any need to adjust atom position or it will be handled by the software itself.

PS: I used these calculation to find the equilibrium lattice constant and I also want to learn about both the methods(1.only lattice change 2. lattice and atom positions change).

Scf does keep the atomic positions fixed so it works in your case only if you those are already the minimal energy positions.

Do you care to explain a bit better what you plan to do?

Cheers,

Roberto

I want to apply the strain in the X-direction from -10% to +10% with the variation of 1% and want to calculate the forces on each atom and energy of the crystal structure. Later I will use this information to train my potential.

I did the following things:

1.) I downloaded the .cif file from the crystallographic database.

2.) From this file I have created the input file for the QE.

3.) Then I did the ecut convergence on the single unit cell.

4.) I did the Kgrid convergence on the single unit cell.

5.) I used these conversed values of ecut and Kgrid in further calculation.

6.) I make a supercell of 2X2X2 of Ni crystal used conversed values of ecut and kgrid.

7.) Now , I am trying to calculate the correct values of forces when I apply the strain on my supercell in X-direction from -10% to +10%.

I am confused how to calculate it correctly like by using the scf or vc-relax. I don't know.

Apart from it I also calculated the lattice parameter to check whether the calculated values of ecut and kgird are correct or not by applying the strain on unit cell. I found the value of the lattice parameter is 3.516 and lattice parameter in .cif file is 3.52 which is very close(I also attached a grap in the above response). Hence my kgird and ecut are correct.That is why I applied the strain on my unit cell in all the three direction, I know I can do it by setting ibrav=2 (for FCC) instead of ibrav=0 and changing the value celldm(1)=3.52. But I choose this method.

Dear Sir,

Here is my script for vc-relax force calculation with 10% strain in x-direction.Please tell whether it is correct or not?

&CONTROL

calculation = 'vc-relax'

verbosity = 'high'

tstress = .true.

tprnfor = .true.

prefix = 'myprefix'

pseudo_dir = '/home/MultiscaleMechanics/170020084/PP/PBE_ONCV'

/

&SYSTEM

ecutwfc = 50

ecutrho = 400

occupations = 'smearing'

degauss = 0.05

smearing = 'methfessel-paxton'

nspin = 1

ntyp = 1

nat = 32

ibrav = 0

/

&ELECTRONS

electron_maxstep = 200

conv_thr = 1e-08

mixing_mode = 'plain'

mixing_beta = 0.7

diagonalization = 'david'

/

&IONS

/

&CELL

cell_dofree="yz"

/

ATOMIC_SPECIES

Ni 58.6934 Ni.UPF

K_POINTS automatic

6 6 6 0 0 0

CELL_PARAMETERS angstrom

7.7440 0.00000000000000 0.00000000000000

0.00000000000000 7.04000000000000 0.00000000000000

0.00000000000000 0.00000000000000 7.04000000000000

ATOMIC_POSITIONS crystal

Ni 0.0000000000 0.0000000000 0.0000000000

Ni 0.0000000000 0.2500000000 0.2500000000

Ni 0.2500000000 0.0000000000 0.2500000000

Ni 0.2500000000 0.2500000000 0.0000000000

Ni 0.0000000000 0.0000000000 0.5000000000

Ni 0.0000000000 0.2500000000 0.7500000000

Ni 0.2500000000 0.0000000000 0.7500000000

Ni 0.2500000000 0.2500000000 0.5000000000

Ni 0.0000000000 0.5000000000 0.0000000000

Ni 0.0000000000 0.7500000000 0.2500000000

Ni 0.2500000000 0.5000000000 0.2500000000

Ni 0.2500000000 0.7500000000 0.0000000000

Ni 0.0000000000 0.5000000000 0.5000000000

Ni 0.0000000000 0.7500000000 0.7500000000

Ni 0.2500000000 0.5000000000 0.7500000000

Ni 0.2500000000 0.7500000000 0.5000000000

Ni 0.5000000000 0.0000000000 0.0000000000

Ni 0.5000000000 0.2500000000 0.2500000000

Ni 0.7500000000 0.0000000000 0.2500000000

Ni 0.7500000000 0.2500000000 0.0000000000

Ni 0.5000000000 0.0000000000 0.5000000000

Ni 0.5000000000 0.2500000000 0.7500000000

Ni 0.7500000000 0.0000000000 0.7500000000

Ni 0.7500000000 0.2500000000 0.5000000000

Ni 0.5000000000 0.5000000000 0.0000000000

Ni 0.5000000000 0.7500000000 0.2500000000

Ni 0.7500000000 0.5000000000 0.2500000000

Ni 0.7500000000 0.7500000000 0.0000000000

Ni 0.5000000000 0.5000000000 0.5000000000

Ni 0.5000000000 0.7500000000 0.7500000000

Ni 0.7500000000 0.5000000000 0.7500000000

Ni 0.7500000000 0.7500000000 0.5000000000

It seems correct to me.

Thank you, Sir!

Roberto D'Agosta

Dear Sir,

These are the final forces on my structure after the relax calculation.

Forces acting on atoms (cartesian axes, Ry/au):

atom 1 type 1 force = 0.00000000 0.00000000 0.00002788

atom 2 type 1 force = 0.00000000 -0.00000000 0.00000537

atom 3 type 1 force = 0.00000000 0.00000000 -0.00001133

atom 4 type 1 force = 0.00000000 0.00000000 -0.00001571

atom 5 type 1 force = 0.00000000 0.00000000 -0.00000299

atom 6 type 1 force = 0.00000000 0.00000000 0.00000728

atom 7 type 1 force = 0.00000000 0.00000000 -0.00001942

atom 8 type 1 force = 0.00000000 0.00000000 -0.00000788

atom 9 type 1 force = 0.00000000 0.00000000 0.00002520

atom 10 type 2 force = 0.00000000 -0.00000000 -0.00000825

atom 11 type 3 force = 0.00000000 -0.00000000 -0.00001877

atom 12 type 2 force = 0.00000000 0.00000000 -0.00000631

atom 13 type 2 force = 0.00000000 0.00000000 0.00001949

atom 14 type 2 force = 0.00000000 0.00000000 0.00000521

atom 15 type 2 force = 0.00000000 0.00000000 0.00000023

Is my structure is in minimum energy condition? Or I have to do these calculations again with tighter force conversion criteria.

I am not sure you can force stricter condition. At a certain point you will need to stop. I would say you structure should be well converged.

Roberto D'Agosta Sir,

Thank you, very much.