Problem in interpolation with nebmake.pl script
Moderator: moderators
Problem in interpolation with nebmake.pl script
Dear Sir,
I have used nebmake.pl to create 1 image, but every time the image is created outside the zone of the initial and final state and sometimes the image structure is completely different. Here I attached the initial, image and the final states' POSCAR and POSCAR.xsf files. Please look at these. The discrepancies can be found in: Row Co-ordinate
1 x
4 x
5 x,z
6 x,y
7 z
Regards,
Dwaipayan Chakraborty
I have used nebmake.pl to create 1 image, but every time the image is created outside the zone of the initial and final state and sometimes the image structure is completely different. Here I attached the initial, image and the final states' POSCAR and POSCAR.xsf files. Please look at these. The discrepancies can be found in: Row Co-ordinate
1 x
4 x
5 x,z
6 x,y
7 z
Regards,
Dwaipayan Chakraborty
- Attachments
-
- POSCARs.rar
- (4.37 KiB) Downloaded 786 times
Re: Problem in interpolation with nebmake.pl script
The problem is that the atoms are not in a consistent order in your initial and final structures. The nebmake.pl script will interpolate between the initial position of atom i to the final position of atom i. If you reorder the atoms in your final state structure so that they are as close to their position in the initial state, your interpolated band will be reasonable.
Re: Problem in interpolation with nebmake.pl script
Thank you, sir, for your reply.
I realized is that the initial and final ion positions differ very much and they are not in order. But I have one doubt if you please look at the .xsf files, there is only one atom 1 and the x coordinate of that in initial structure is 6.402489 , in final strucutre is 1.715433 but in the image that is 8.684338. How it can be?
Another thing is that in initial and final structure due to periodic boundary condition some coordinate changes form 14.50 to -0.022 (lattice parameter is 14.53), should I then write 0.022 as 14.51 and interpolate?
Can we only interpolate the position of the atom which is diffusing and let the other atom be at the position of the initial one and do the NEB calculation?
I realized is that the initial and final ion positions differ very much and they are not in order. But I have one doubt if you please look at the .xsf files, there is only one atom 1 and the x coordinate of that in initial structure is 6.402489 , in final strucutre is 1.715433 but in the image that is 8.684338. How it can be?
Another thing is that in initial and final structure due to periodic boundary condition some coordinate changes form 14.50 to -0.022 (lattice parameter is 14.53), should I then write 0.022 as 14.51 and interpolate?
Can we only interpolate the position of the atom which is diffusing and let the other atom be at the position of the initial one and do the NEB calculation?
Re: Problem in interpolation with nebmake.pl script
The interpolation will deal with the periodic boundary conditions - don't worry about that. Just get the relative order of the atoms in the initial and final state correct.
Re: Problem in interpolation with nebmake.pl script
Sir,
1.I have ordered the atoms in the proper way and then used nebmake.pl. Now everything is coming nice expect the coordinate of the single atom which will diffuse (No 1 in the attached file).
2. Can we only interpolate the position of the atom which is diffusing and let the other atom be at the position of the initial one and do the NEB calculation?
1.I have ordered the atoms in the proper way and then used nebmake.pl. Now everything is coming nice expect the coordinate of the single atom which will diffuse (No 1 in the attached file).
2. Can we only interpolate the position of the atom which is diffusing and let the other atom be at the position of the initial one and do the NEB calculation?
- Attachments
-
- POSCAR_new_initial.txt
- (1.54 KiB) Downloaded 780 times
-
- POSCAR_new_image.txt
- (1.54 KiB) Downloaded 1706 times
-
- POSCAR_new_final.txt
- (1.54 KiB) Downloaded 796 times
Re: Problem in interpolation with nebmake.pl script
You typically want to find a minimum energy path between two stable structures. As atom 1 moves, the other atoms relax, and you typically want that.
If you want the other atoms frozen, then you can explicitly freeze them in your DFT calculation, calculate new initial and final states (with the same locations of the frozen atoms) and then setup an NEB calculation using those endpoint structures.
If you want the other atoms frozen, then you can explicitly freeze them in your DFT calculation, calculate new initial and final states (with the same locations of the frozen atoms) and then setup an NEB calculation using those endpoint structures.
Re: Problem in interpolation with nebmake.pl script
Thank you very much sir for clearing my doubts.
But why the coordinates of the atom 1, which moves, are coming out the range (of the initial and the final structure) in the image structures?
But why the coordinates of the atom 1, which moves, are coming out the range (of the initial and the final structure) in the image structures?
Re: Problem in interpolation with nebmake.pl script
Oh, because the script finds the shortest path between the initial and final state (the minimum image convention). In your case, atom 1 goes across one side of the box to the other, as allowed by the periodic boundary conditions.
Re: Problem in interpolation with nebmake.pl script
Thank you very much sir for clearing my doubts.
Re: Problem in interpolation with nebmake.pl script
Dear Sir,
Just one query when looking at the OUTCAR files, should I look at the CHAIN+TOTAL forces to meet the convergence criteria (0.01eV/Angst in my case) or the individual CHAIN forces and TOTAL forces to meet the convergence criteria also?
Just one query when looking at the OUTCAR files, should I look at the CHAIN+TOTAL forces to meet the convergence criteria (0.01eV/Angst in my case) or the individual CHAIN forces and TOTAL forces to meet the convergence criteria also?
Re: Problem in interpolation with nebmake.pl script
just the CHAIN+TOTAL
Re: Problem in interpolation with nebmake.pl script
Thank you very much, sir.
Re: Problem in interpolation with nebmake.pl script
Dear Sir,
In case of NEB calculation (as implemented in VASP) the OUTCAR prints the CHAIN+TOTAL forces which have to be looked to check the convergence criteria (0.01eV/Angst in my case) has been met or not (as you said) but for the CI-NEB calculation there is no CHAIN+TOTAL forces printed in the OUTCAR file. So in this case of the CI-NEB how can I check the force convergence from the OUTCAR file?
In case of NEB calculation (as implemented in VASP) the OUTCAR prints the CHAIN+TOTAL forces which have to be looked to check the convergence criteria (0.01eV/Angst in my case) has been met or not (as you said) but for the CI-NEB calculation there is no CHAIN+TOTAL forces printed in the OUTCAR file. So in this case of the CI-NEB how can I check the force convergence from the OUTCAR file?
Re: Problem in interpolation with nebmake.pl script
If you look in the OUTCAR, after the lines:
POSITION TOTAL-FORCE (eV/Angst)
-----------------------------------------------------------------------------------
you can see the NEB forces (total+chain). Then you can also see the forces used for the convergence criterion, as in:
FORCES: max atom, RMS 0.523540 0.175844
FORCE total and by dimension 1.098148 0.452968
POSITION TOTAL-FORCE (eV/Angst)
-----------------------------------------------------------------------------------
you can see the NEB forces (total+chain). Then you can also see the forces used for the convergence criterion, as in:
FORCES: max atom, RMS 0.523540 0.175844
FORCE total and by dimension 1.098148 0.452968
Re: Problem in interpolation with nebmake.pl script
Dear Sir,
I am getting the following:
NEB: Tangent
----------------------------------------------
0.28401 0.15083 -0.85009
-0.03674 0.01548 0.02125
-0.06085 0.03615 -0.02879
-0.02719 -0.02189 -0.02387
-0.02327 -0.01379 -0.02847
-0.09571 0.01797 -0.05116
-0.09799 0.03135 0.09570
0.01741 -0.03167 -0.03956
0.00893 -0.04090 -0.00792
-0.07237 0.10393 -0.03189
-0.04245 0.01496 0.01044
-0.00678 -0.01145 -0.02421
0.00028 0.00077 -0.01945
0.00428 0.00953 -0.02647
-0.02276 -0.00299 0.03473
0.00709 -0.04561 -0.01974
-0.00032 -0.07873 -0.02466
-0.04048 0.00042 -0.00358
-0.03768 0.05185 -0.02282
-0.00355 -0.00527 -0.02270
-0.00790 -0.00809 -0.02624
-0.03419 0.00884 0.01578
0.03700 -0.03243 -0.03092
0.03967 -0.12201 -0.01597
0.00800 -0.05779 0.00950
-0.06591 -0.01358 -0.00341
-0.02356 0.06783 -0.01507
-0.02208 0.00062 -0.01807
-0.02026 -0.00508 -0.03044
-0.02241 0.01079 0.02079
-0.01056 -0.01736 -0.01355
0.00489 -0.07341 -0.02422
0.00588 -0.03814 -0.02594
-0.06188 0.06021 -0.01623
-0.04356 0.00101 -0.00717
-0.01480 -0.03052 -0.02617
-0.02866 -0.05164 -0.03092
0.08760 0.02372 -0.03037
-0.06804 -0.00830 0.00004
-0.00144 -0.01833 0.01245
0.02359 -0.08945 -0.02225
NEB: forces: par spring, perp REAL, dneb 0.009885 0.017753 0.000000
NEB: distance to prev, next image, angle between 0.725551 0.727528 149.562590
NEB: projections on to tangent (spring, REAL) 0.009885 -0.075397
stress matrix after NEB project (eV)
-0.79328 -0.55740 0.30884
-0.55740 1.75197 -0.15657
0.30884 -0.15657 -0.11039
FORCES: max atom, RMS 0.007795 0.003173
FORCE total and by dimension 0.020319 0.007514
Stress total and by dimension 2.138233 1.751966
The EDIFFG in INCAR is -0.01. Does that means the each component of the forces written under the NEB tangent line should be less than -0.01 or I have to just check the RMS value written at last?
Is this calculation converged?
I am getting the following:
NEB: Tangent
----------------------------------------------
0.28401 0.15083 -0.85009
-0.03674 0.01548 0.02125
-0.06085 0.03615 -0.02879
-0.02719 -0.02189 -0.02387
-0.02327 -0.01379 -0.02847
-0.09571 0.01797 -0.05116
-0.09799 0.03135 0.09570
0.01741 -0.03167 -0.03956
0.00893 -0.04090 -0.00792
-0.07237 0.10393 -0.03189
-0.04245 0.01496 0.01044
-0.00678 -0.01145 -0.02421
0.00028 0.00077 -0.01945
0.00428 0.00953 -0.02647
-0.02276 -0.00299 0.03473
0.00709 -0.04561 -0.01974
-0.00032 -0.07873 -0.02466
-0.04048 0.00042 -0.00358
-0.03768 0.05185 -0.02282
-0.00355 -0.00527 -0.02270
-0.00790 -0.00809 -0.02624
-0.03419 0.00884 0.01578
0.03700 -0.03243 -0.03092
0.03967 -0.12201 -0.01597
0.00800 -0.05779 0.00950
-0.06591 -0.01358 -0.00341
-0.02356 0.06783 -0.01507
-0.02208 0.00062 -0.01807
-0.02026 -0.00508 -0.03044
-0.02241 0.01079 0.02079
-0.01056 -0.01736 -0.01355
0.00489 -0.07341 -0.02422
0.00588 -0.03814 -0.02594
-0.06188 0.06021 -0.01623
-0.04356 0.00101 -0.00717
-0.01480 -0.03052 -0.02617
-0.02866 -0.05164 -0.03092
0.08760 0.02372 -0.03037
-0.06804 -0.00830 0.00004
-0.00144 -0.01833 0.01245
0.02359 -0.08945 -0.02225
NEB: forces: par spring, perp REAL, dneb 0.009885 0.017753 0.000000
NEB: distance to prev, next image, angle between 0.725551 0.727528 149.562590
NEB: projections on to tangent (spring, REAL) 0.009885 -0.075397
stress matrix after NEB project (eV)
-0.79328 -0.55740 0.30884
-0.55740 1.75197 -0.15657
0.30884 -0.15657 -0.11039
FORCES: max atom, RMS 0.007795 0.003173
FORCE total and by dimension 0.020319 0.007514
Stress total and by dimension 2.138233 1.751966
The EDIFFG in INCAR is -0.01. Does that means the each component of the forces written under the NEB tangent line should be less than -0.01 or I have to just check the RMS value written at last?
Is this calculation converged?