Dear Kay
Thank you for your response. However the atomic coordinates of the primitive cell seem shifted: origin shifted????
What is the point of shifting the atomic coordinates to origin for one of the basis atoms? Does this procedure help in finding the crystal symmetries more easily?
Are the Primitive cell atomic coordinates that are stored in the array "atposl" origin shifted? Are these the values printed out in INFO.OUT?
As an example for Hafnium Metal: The input exciting.in file contains the following information of the primitive cell atomic coordinates:
EXCITING (0.9.93) shifts the basis so that the first atom in the smallest atom set is at the origin. This is to ensure that a maximum number of symmetry operations are found.
You can switch shifting off with
tshift
.false.
The Bravais lattice symmetries are those which leave the lattice (sans basis) invariant. These are just point-group operations (rotations and inversion). The crystal symmetries consist of operations of the form
{a_S,a_R,t}
where t is a translation, a_R is a spatial rotation and a_S is a global spin rotation (see routine findsymcrys in the EXCITING Manual). These are magnetic space group operations and act on the lattice, basis and magnetic fields together.
What is the discrepancy with VASP?
Cheers,
Kay.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
This needs a reduced mixing parameter beta0 (0.02-0.05) for accurate convergence.
-----------------------------------------------------------------------------------
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
I've found the problem with the Hf metal. You have to be more accurate with your lattice parameters and positions in order for the code to realise that the structure is symmetric (see the variable "epslat" in the Manual)
So if I use
avec
5.8617 0.0000 0.0000
-2.93085 5.07638110936 0.0000
0.0000 0.0000 9.2613
With regards to HfO2, the number of k-points (196) is probably correct for a regular non-shifted 12x12x12 grid. Can you run VASP for a non-shifted grid?
Cheers,
Kay.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Sorry for the delay in replying to the mail. You are absolutely right. When using a GAMMA centred grid instead of a Monkhorst Pack I get the same number of K-points (196) as Exciting. using Monkhorst one obtains 126.
Okay so I now find an earlier statement of yours a little contradictory: "EXCITING (0.9.93) shifts the basis so that the first atom in the smallest atom set is at the origin. This is to ensure that a maximum number of symmetry operations are found."
So there a general trend of finding the maximum number of symmetry poerations with origin shifting? with the current HfO2-tetragonal example being an exception? Because clearly this case did not require any "shifting" and "shifting away from origin" lead to a smaller irreducible k-point set.
Ravi
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
It's not necessary to shift the origin, it just makes it easier to find the spacegroup symmetries. The crystal is invariant under arbitrary shifts, but if EXCITING rotated about arbitrary symmetry centers it would have to work harder to find the shift which would take the crystal into coincidence.
Cheers,
Kay.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Dear Kay
Thank you for your response. However the atomic coordinates of the primitive cell seem shifted: origin shifted????
What is the point of shifting the atomic coordinates to origin for one of the basis atoms? Does this procedure help in finding the crystal symmetries more easily?
Are the Primitive cell atomic coordinates that are stored in the array "atposl" origin shifted? Are these the values printed out in INFO.OUT?
As an example for Hafnium Metal: The input exciting.in file contains the following information of the primitive cell atomic coordinates:
atoms
1 : nspecies
'Hf.in' : spfname
2 : natoms
0.3333383597995839 0.6666767195991606 0.2500000000000000 0. 0. 0.
0.6666616402004159 0.3333232804008392 0.7500000000000000 0. 0. 0.
This is what is printed in INFO.OUT:
Species : 1 (Hf)
parameters loaded from : Hf.in
name : hafnium
nuclear charge : -72.00000000
electronic charge : 72.00000000
atomic mass : 325367.3657
muffin-tin radius : 2.000000000
number of radial points in muffin-tin : 629
atomic position (lattice), magnetic field (Cartesian) :
1 0.00000000 0.00000000 0.00000000 0.00000000 0.00000000 0.00000000
2 0.33332328 0.66664656 0.50000000 0.00000000 0.00000000 0.00000000
It is also interesting to compare the reduced K-point set obtained through this code with a code like VASP. There is a discrepancy.
Also:
EXCITING says:
Number of Bravais lattice symmetries : 4
Number of crystal symmetries : 4
What is the distinction between Bravias Lattice Symmetry and Crystal Symmetry?
Thank you
Ravi
Dear Ravi,
EXCITING (0.9.93) shifts the basis so that the first atom in the smallest atom set is at the origin. This is to ensure that a maximum number of symmetry operations are found.
You can switch shifting off with
tshift
.false.
The Bravais lattice symmetries are those which leave the lattice (sans basis) invariant. These are just point-group operations (rotations and inversion). The crystal symmetries consist of operations of the form
{a_S,a_R,t}
where t is a translation, a_R is a spatial rotation and a_S is a global spin rotation (see routine findsymcrys in the EXCITING Manual). These are magnetic space group operations and act on the lattice, basis and magnetic fields together.
What is the discrepancy with VASP?
Cheers,
Kay.
Okay four examples:
The numbers are the symmetry reduced K-points produced for a 12x12x12 cell.
EXCITING VASP
ZrSiO4 (BCT) 163 (origon shifted) 163 (Gamma)
HfO2(Tetragonal) 196 126 (Monkhorst)
ZrO2 (monoclinic) 518 (origin shifted) 518 (Gamma)
Hf (Metal) 518 (origin shift) 133 (Gamma)
Okay the table vanished! The first number corresponds to EXCITING and the Second number corresponds to a VASP calculation.
here we go:
ZrSiO4 (BCT)
163 (origon shifted) - Exciting
163 (Gamma) - Vasp
HfO2(Tetragonal)
196 - Exciting
126 (Monkhorst) - Vasp
ZrO2 (monoclinic)
518 (origin shifted) - Exciting
518 (Gamma) - Vasp
Hf (Metal)
518 (origin shift) - Exciting
133 (Gamma) - Vasp
Hi Ravi,
Could you send me your input files for HfO2 and Hf (metal).
Thanks,
Kay.
Here you go:
---------------------------------------------------------------------------
Hf:
avec
5.8617 0.0000 0.0000
-2.9308 5.0765 0.0000
0.0000 0.0000 9.2613
atoms
1 : nspecies
'Hf.in' : spfname
2 : natoms
0.3333383597995839 0.6666767195991606 0.2500000000000000 0. 0. 0.
0.6666616402004159 0.3333232804008392 0.7500000000000000 0. 0. 0.
------------------------------------------------------------------------------------
HfO2:
avec
6.61743 0.0000 0.0000
0.0000 6.61743 0.0000
0.0000 0.0000 9.48596
atoms
2 : nspecies
'Hf.in' : spfname
2 : natoms
0.0000000000000000 0.0000000000000000 0.0000000000000000 0. 0. 0.
0.5000000000000000 0.5000000000000000 0.5000000000000000 0. 0. 0.
'O.in' : spfname
4 : natoms
0.0000000000000000 0.5000000000000000 0.2108530436969992 0. 0. 0.
0.0000000000000000 0.5000000000000000 0.7108530436969990 0. 0. 0.
0.5000000000000000 0.0000000000000000 0.2891469563030008 0. 0. 0.
0.5000000000000000 0.0000000000000000 0.7891469563030010 0. 0. 0.
This needs a reduced mixing parameter beta0 (0.02-0.05) for accurate convergence.
-----------------------------------------------------------------------------------
Hi Ravi,
I've found the problem with the Hf metal. You have to be more accurate with your lattice parameters and positions in order for the code to realise that the structure is symmetric (see the variable "epslat" in the Manual)
So if I use
avec
5.8617 0.0000 0.0000
-2.93085 5.07638110936 0.0000
0.0000 0.0000 9.2613
atoms
1 : nspecies
'Hf.in' : spfname
2 : natoms
0.33333333333333 0.66666666666667 0.250000000000 0.0 0.0 0.0
0.66666666666667 0.33333333333333 0.750000000000 0.0 0.0 0.0
I get 133 k-points, like VASP.
With regards to HfO2, the number of k-points (196) is probably correct for a regular non-shifted 12x12x12 grid. Can you run VASP for a non-shifted grid?
Cheers,
Kay.
Dear Kay
Sorry for the delay in replying to the mail. You are absolutely right. When using a GAMMA centred grid instead of a Monkhorst Pack I get the same number of K-points (196) as Exciting. using Monkhorst one obtains 126.
Okay so I now find an earlier statement of yours a little contradictory: "EXCITING (0.9.93) shifts the basis so that the first atom in the smallest atom set is at the origin. This is to ensure that a maximum number of symmetry operations are found."
So there a general trend of finding the maximum number of symmetry poerations with origin shifting? with the current HfO2-tetragonal example being an exception? Because clearly this case did not require any "shifting" and "shifting away from origin" lead to a smaller irreducible k-point set.
Ravi
Dear Ravi,
It's not necessary to shift the origin, it just makes it easier to find the spacegroup symmetries. The crystal is invariant under arbitrary shifts, but if EXCITING rotated about arbitrary symmetry centers it would have to work harder to find the shift which would take the crystal into coincidence.
Cheers,
Kay.