I am running calculations in ELK and I like to reduce the muffin-tin radius for a species file. What other parameters would I need to change in order to compensate this change? and in what range?
Best regards,
Joel
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Thank you for a fast answer. When I changed the muffin-tin for Al from 2.2 to 1.8 the core leakage changed from 11.27573445 (0.5040354900E-03) to 10.60729312 (0.3984359176E-02). Is this still considered small?
I calculated a ELF plot and when I change the muffin tin from 2.2 to 1.8 the discontinuity between muffin-tin and the interstial region gets larger, see attached plots. Is there any way to reduce this discontinuity?
I now realize that it could be difficult for you to determine if the core leakage is small enough. I did some further tests to see if I could reduce the core leakage. If I move one electron from the core to valance, ie., change spcore from T to F in the species file, I could reduce the core leakage to (0.1332267630E-14) witch is insignificant but the discontinuity is still there.
More worrying is that the ELF plot now changes a lot. With the old spcore setting (Al_elf_mt_2.2.pdf) ELF had a high maximum close to the core and a maximum in the interstitial region but the new spcore setting (Al_elf_mt_1.8+corestate.pdf) the maxima close to the core has decreased and there is now a minimum in the interstitial region, see attached files.
Although I have no personal experience with ELF plots I can at least guess the origin of your problem. As the ELF involves gradients of the density, a non-converged APW basis set will give you jumps since these basis functions do not have a continuous gradient. Hence there are two ways to improve your ELF plot:
1) converge the basis set more by increasing RGKMAX parameter - the jump will decrease but probably not disappear unless you go to infinite basis ...
2) switch to LAPW basis set, which still could give kinks in ELF but should not give jumps.
I suggest you read the manual how to go from APW to LAPW basis set. As I recall there was recently an introduction of a flag taking care of that ...
Best wishes,
Lars
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
from my experience with calculations of ELI-D (isotopological to ELF for DFT) for the group-14 elements, it is good to have the entire semicore shell threated as valence. However, to make the 2s orbital valence, you need not only to change the spcore parameter from T to F, but also to add a new local orbital (LO) to the LAPW set. For this, you need to add a block similar to the last block in your species file, with lorbl=0 and a different energy in the last line. To estimate the energy, you can take a look in the EVALCORE.OUT file from the calculation in which the 2s orbital was threated as a core orbital. I usually put a slightly lower value than the energy in the last line in this file, but sometimes there is a problem with the linearization energy, and one might need to play around a little bit.
For me, going from APW to LAPW decreased the jumps in ELI-D significantly, so I would follow the Lars's suggestion. For this, you need to add the second line in the APW block, with apwdm=1, and change apword to 2. Moreover, in my experience the lo's were not needed if LAPWs are used, so you might just delete the 2-line blocks from the species file. I will put my species file for silicon below, hope it helps. But I should say that I used the 2.2.1 version, it might be that there are some changes in the program meanwile..
Another issue is that a large basis set was needed to get ELI-D smooth. I used rgkmax=10, lmaxapw=13 and gmaxvr=25, but my k-points set was smaller.
Hope it helps.
Best wishes,
Olga
P.S. Reading your ELF plots is a bit tricky. If I'm correct, they go not along the bond line, but between the centers of two bonds, so that there is an atom at the center and two halves of the valence regions at the borders. Is there a reason to do make them in this way?
Si.in file:
'Si' : spsymb
'silicon' : spname
-14.0000 : spzn
51196.73452 : spmass
0.534522E-06 1.600 46.0261 500 : sprmin, rmt, sprmax, nrmt
7 : spnst
1 0 1 2.00000 T : spn, spl, spk, spocc, spcore
2 0 1 2.00000 F
2 1 1 2.00000 F
2 1 2 4.00000 F
3 0 1 2.00000 F
3 1 1 1.00000 F
3 1 2 1.00000 F
2 : LAPW : apword
0.1500 0 F : apwe0, apwdm, apwve
0.1500 1 F
0 : nlx
2 : nlorb
1 3 : lorbl, lorbord
0.1500 0 F : lorbe0, lorbdm, lorbve
0.1500 1 F
-3.5000 0 T :2p LO
0 3 : lorbl, lorbord
0.1500 0 F : lorbe0, lorbdm, lorbve
0.1500 1 F
-4.8000 0 T :2s LO
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Thank you for your response, it was very helpful. I think I have sorted it about now. I added rgkmax=10, lmaxapw=13 and gmaxvr=25 to elk.in file and modified the specie file as Olga recommend. I check the EVALCORE and found n = 2, l = 0, k = 1 : -3.483455883. I rounded this down and added it to the species file:
'Al' : spsymb
'aluminium' : spname
-13.0000 : spzn
49184.33492 : spmass
0.554700E-06 1.6000 55.3437 1000 : sprmin, rmt, sprmax, nrmt
6 : spnst
1 0 1 2.00000 T : spn, spl, spk, spocc, spcore
2 0 1 2.00000 F
2 1 1 2.00000 F
2 1 2 4.00000 F
3 0 1 2.00000 F
3 1 1 1.000000 F
2 : apword
0.1500 0 F : apwe0, apwdm, apwve
0.1500 1 F : apwe0, apwdm, apwve
0 : nlx
4 : nlorb
0 2 : lorbl, lorbord
0.1500 0 F : lorbe0, lorbdm, lorbve
0.1500 1 F
1 2 : lorbl, lorbord
0.1500 0 F : lorbe0, lorbdm, lorbve
0.1500 1 F
1 3 : lorbl, lorbord
0.1500 0 F : lorbe0, lorbdm, lorbve
0.1500 1 F
-2.5562 0 T
0 3
0.1500 0 F : lorbe0, lorbdm, lorbve
0.1500 1 F
-3.5 0 T : 2s
Regarding the PS: In my elk.in I plot from -0.5 0.0 0.0 to 0.5 0.0 0.0 so the atom would be in the middle. If I understand correctly you would like me to go from 0.0 0.0 0.0 to 1.0 0.0 0.0 so the bond is in the middle. I have attached this version. As I have understood, when plotting the elf plot one plots it between its nearest neighbour, please correct if that is not the case.
Sorry, I might expressed myself incorrect. There was nothing wrong with your plot, it's just more convenient (for my personal taste) to have the important information (the valence region) at the center of the plot :)
Best wishes,
Olga
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Dear all,
I am running calculations in ELK and I like to reduce the muffin-tin radius for a species file. What other parameters would I need to change in order to compensate this change? and in what range?
Best regards,
Joel
In principle reducing MT does not need compensation.
you may want to check the core leakage in INFO.OUT if its still small!
otherwise I see no issue.
best
Michael
Dear Micheal,
Thank you for a fast answer. When I changed the muffin-tin for Al from 2.2 to 1.8 the core leakage changed from 11.27573445 (0.5040354900E-03) to 10.60729312 (0.3984359176E-02). Is this still considered small?
I calculated a ELF plot and when I change the muffin tin from 2.2 to 1.8 the discontinuity between muffin-tin and the interstial region gets larger, see attached plots. Is there any way to reduce this discontinuity?
Best,
Joel
Dear Micheal,
I now realize that it could be difficult for you to determine if the core leakage is small enough. I did some further tests to see if I could reduce the core leakage. If I move one electron from the core to valance, ie., change spcore from T to F in the species file, I could reduce the core leakage to (0.1332267630E-14) witch is insignificant but the discontinuity is still there.
More worrying is that the ELF plot now changes a lot. With the old spcore setting (Al_elf_mt_2.2.pdf) ELF had a high maximum close to the core and a maximum in the interstitial region but the new spcore setting (Al_elf_mt_1.8+corestate.pdf) the maxima close to the core has decreased and there is now a minimum in the interstitial region, see attached files.
Here is the elk.in
and the original specie file for Al:
'Al' : spsymb 'aluminium' : spname -13.0000 : spzn 49184.33492 : spmass 0.554700E-06 2.2000 55.3437 1000 : sprmin, rmt, sprmax, nrmt 6 : spnst 1 0 1 2.00000 T : spn, spl, spk, spocc, spcore 2 0 1 2.00000 T 2 1 1 2.00000 F 2 1 2 4.00000 F 3 0 1 2.00000 F 3 1 1 1.000000 F 1 : apword 0.1500 0 F : apwe0, apwdm, apwve 0 : nlx 3 : nlorb 0 2 : lorbl, lorbord 0.1500 0 F : lorbe0, lorbdm, lorbve 0.1500 1 F 1 2 : lorbl, lorbord 0.1500 0 F : lorbe0, lorbdm, lorbve 0.1500 1 F 1 3 : lorbl, lorbord 0.1500 0 F : lorbe0, lorbdm, lorbve 0.1500 1 F -2.5562 0 Tand the modifed:
'Al' : spsymb 'aluminium' : spname -13.0000 : spzn 49184.33492 : spmass 0.554700E-06 1.8000 55.3437 1000 : sprmin, rmt, sprmax, nrmt 6 : spnst 1 0 1 2.00000 T : spn, spl, spk, spocc, spcore 2 0 1 2.00000 F 2 1 1 2.00000 F 2 1 2 4.00000 F 3 0 1 2.00000 F 3 1 1 1.000000 F 1 : apword 0.1500 0 F : apwe0, apwdm, apwve 0 : nlx 3 : nlorb 0 2 : lorbl, lorbord 0.1500 0 F : lorbe0, lorbdm, lorbve 0.1500 1 F 1 2 : lorbl, lorbord 0.1500 0 F : lorbe0, lorbdm, lorbve 0.1500 1 F 1 3 : lorbl, lorbord 0.1500 0 F : lorbe0, lorbdm, lorbve 0.1500 1 F -2.5562 0 TBest,
Joel
Dear Joel,
Although I have no personal experience with ELF plots I can at least guess the origin of your problem. As the ELF involves gradients of the density, a non-converged APW basis set will give you jumps since these basis functions do not have a continuous gradient. Hence there are two ways to improve your ELF plot:
1) converge the basis set more by increasing RGKMAX parameter - the jump will decrease but probably not disappear unless you go to infinite basis ...
2) switch to LAPW basis set, which still could give kinks in ELF but should not give jumps.
I suggest you read the manual how to go from APW to LAPW basis set. As I recall there was recently an introduction of a flag taking care of that ...
Best wishes,
Lars
Dear Joel,
from my experience with calculations of ELI-D (isotopological to ELF for DFT) for the group-14 elements, it is good to have the entire semicore shell threated as valence. However, to make the 2s orbital valence, you need not only to change the spcore parameter from T to F, but also to add a new local orbital (LO) to the LAPW set. For this, you need to add a block similar to the last block in your species file, with lorbl=0 and a different energy in the last line. To estimate the energy, you can take a look in the EVALCORE.OUT file from the calculation in which the 2s orbital was threated as a core orbital. I usually put a slightly lower value than the energy in the last line in this file, but sometimes there is a problem with the linearization energy, and one might need to play around a little bit.
For me, going from APW to LAPW decreased the jumps in ELI-D significantly, so I would follow the Lars's suggestion. For this, you need to add the second line in the APW block, with apwdm=1, and change apword to 2. Moreover, in my experience the lo's were not needed if LAPWs are used, so you might just delete the 2-line blocks from the species file. I will put my species file for silicon below, hope it helps. But I should say that I used the 2.2.1 version, it might be that there are some changes in the program meanwile..
Another issue is that a large basis set was needed to get ELI-D smooth. I used rgkmax=10, lmaxapw=13 and gmaxvr=25, but my k-points set was smaller.
Hope it helps.
Best wishes,
Olga
P.S. Reading your ELF plots is a bit tricky. If I'm correct, they go not along the bond line, but between the centers of two bonds, so that there is an atom at the center and two halves of the valence regions at the borders. Is there a reason to do make them in this way?
Si.in file:
'Si' : spsymb
'silicon' : spname
-14.0000 : spzn
51196.73452 : spmass
0.534522E-06 1.600 46.0261 500 : sprmin, rmt, sprmax, nrmt
7 : spnst
1 0 1 2.00000 T : spn, spl, spk, spocc, spcore
2 0 1 2.00000 F
2 1 1 2.00000 F
2 1 2 4.00000 F
3 0 1 2.00000 F
3 1 1 1.00000 F
3 1 2 1.00000 F
2 : LAPW : apword
0.1500 0 F : apwe0, apwdm, apwve
0.1500 1 F
0 : nlx
2 : nlorb
1 3 : lorbl, lorbord
0.1500 0 F : lorbe0, lorbdm, lorbve
0.1500 1 F
-3.5000 0 T :2p LO
0 3 : lorbl, lorbord
0.1500 0 F : lorbe0, lorbdm, lorbve
0.1500 1 F
-4.8000 0 T :2s LO
Dear Lars and Olga,
Thank you for your response, it was very helpful. I think I have sorted it about now. I added rgkmax=10, lmaxapw=13 and gmaxvr=25 to elk.in file and modified the specie file as Olga recommend. I check the EVALCORE and found n = 2, l = 0, k = 1 : -3.483455883. I rounded this down and added it to the species file:
'Al' : spsymb 'aluminium' : spname -13.0000 : spzn 49184.33492 : spmass 0.554700E-06 1.6000 55.3437 1000 : sprmin, rmt, sprmax, nrmt 6 : spnst 1 0 1 2.00000 T : spn, spl, spk, spocc, spcore 2 0 1 2.00000 F 2 1 1 2.00000 F 2 1 2 4.00000 F 3 0 1 2.00000 F 3 1 1 1.000000 F 2 : apword 0.1500 0 F : apwe0, apwdm, apwve 0.1500 1 F : apwe0, apwdm, apwve 0 : nlx 4 : nlorb 0 2 : lorbl, lorbord 0.1500 0 F : lorbe0, lorbdm, lorbve 0.1500 1 F 1 2 : lorbl, lorbord 0.1500 0 F : lorbe0, lorbdm, lorbve 0.1500 1 F 1 3 : lorbl, lorbord 0.1500 0 F : lorbe0, lorbdm, lorbve 0.1500 1 F -2.5562 0 T 0 3 0.1500 0 F : lorbe0, lorbdm, lorbve 0.1500 1 F -3.5 0 T : 2sRegarding the PS: In my elk.in I plot from -0.5 0.0 0.0 to 0.5 0.0 0.0 so the atom would be in the middle. If I understand correctly you would like me to go from 0.0 0.0 0.0 to 1.0 0.0 0.0 so the bond is in the middle. I have attached this version. As I have understood, when plotting the elf plot one plots it between its nearest neighbour, please correct if that is not the case.
Best wishes,
Joel
Dear Joel,
I'm glad the "receipt" worked for you.
Sorry, I might expressed myself incorrect. There was nothing wrong with your plot, it's just more convenient (for my personal taste) to have the important information (the valence region) at the center of the plot :)
Best wishes,
Olga