The elk manual states that when one uses the highq switch "changes can be overruled by subsequent blocks in the input". Yet adding blocks after highq does not seem to override what highq did in the first place. For example I was running the NiO tutorial and added highq and after that blocks for trimvg and new k-point grid. From INFO.OUT however it seems that my kpt grid setting was ignored by elk. Have I perhaps misunderstood how highq works or could this be a bug?
Best regards and see you at the Elk school!
Kari Ruotsalainen
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
if (highq) then
rgkmax=7.5d0
gmaxvr=16.d0
lmaxapw=10
lmaxvr=8
lmaxmat=7
autoswidth=.true.
radkpt=60.d0
autokpt=.true.
vkloff(:)=0.d0
nempty=40
lradstp=2
epspot=1.d-7
epsengy=1.d-5
epsforce=1.d-4
end if
As you see with highq you will use autokpt=.true. which will overrun any k-point mesh.
If you want to specify a k-point mesh you have to use autokpt=.false. below the highq=.true. line.
Best,
Lars
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Dear all,
how can one override highq and set some of the parameters manually and still use highq i.e setting some parameters by highq and other manually. Because elk-4.0.15 is not overrunning some parameters despite writing them after highq in elk.in file.
ssingh
Last edit: TOP GUN 2022-12-28
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
I am naïve in band structure calculations and face challenges in obtaining the density of unoccupied states up to 30 eV.
I have considered the antisite disordered cubic structure of Co2MnAl (swapping the atomic positions) for the calculation. Firstly, nonmagnetic calculations were performed. For magnetic calculation, I used the state. out from nonmagnetic calculation(elk file attached). In both nonmagnetic and magnetic calculation, I encountered the 2 and 17 warnings respectively stating, “could not find linearization energy in self-consistent (SC) loop”. The warning appears in ‘1’ and ‘2’ SC cycles out of a total no. of 22 cycles in the nonmagnetic calculation. For magnetic calculation, warnings occur in 2 to 28 cycles out of a total no. of 28 cycles.
To remove this warning, I tried the calculations with different muffin-tin radii and also implemented mrmtav with hiqhq setting. Despite these, warnings continue to persist.
I want to note here that all warnings occur with a high q setting. Disabling highq eliminates the warnings but prevents me from obtaining the density of states up to 30eV.
Please suggest how we could perform calculations without warnings in hiqhq setting.
Thank you for your response.
Dear Swati,
You've posted in an unrelated topic, but oh well.
Warnings at the first few loops are no problem as long as they go away later.
The problem shows up at a magnetic stage indicating that RMT's are likely not the issue.
Metals are not really my thing, but'd say - try looking into +U approaches or an MGGA functional (RPP works well with elk). Both are quite some rabbit holes to fall into, but you might be having a strongly correlated system on hand, and those require special treatment. Look into recent works of Iurii Timrov.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Hi all,
The elk manual states that when one uses the highq switch "changes can be overruled by subsequent blocks in the input". Yet adding blocks after highq does not seem to override what highq did in the first place. For example I was running the NiO tutorial and added highq and after that blocks for trimvg and new k-point grid. From INFO.OUT however it seems that my kpt grid setting was ignored by elk. Have I perhaps misunderstood how highq works or could this be a bug?
Best regards and see you at the Elk school!
Kari Ruotsalainen
Dear Kari,
here is a cut from readinput.f90:
if (highq) then
rgkmax=7.5d0
gmaxvr=16.d0
lmaxapw=10
lmaxvr=8
lmaxmat=7
autoswidth=.true.
radkpt=60.d0
autokpt=.true.
vkloff(:)=0.d0
nempty=40
lradstp=2
epspot=1.d-7
epsengy=1.d-5
epsforce=1.d-4
end if
As you see with highq you will use autokpt=.true. which will overrun any k-point mesh.
If you want to specify a k-point mesh you have to use autokpt=.false. below the highq=.true. line.
Best,
Lars
Dear all,
how can one override highq and set some of the parameters manually and still use highq i.e setting some parameters by highq and other manually. Because elk-4.0.15 is not overrunning some parameters despite writing them after highq in elk.in file.
ssingh
Last edit: TOP GUN 2022-12-28
Thank you for the quick response Lars. Seems like I should have done as the previous sentence in the manual states.
BR,
Kari
Last edit: TOP GUN 2018-12-25
I am naïve in band structure calculations and face challenges in obtaining the density of unoccupied states up to 30 eV.
I have considered the antisite disordered cubic structure of Co2MnAl (swapping the atomic positions) for the calculation. Firstly, nonmagnetic calculations were performed. For magnetic calculation, I used the state. out from nonmagnetic calculation(elk file attached). In both nonmagnetic and magnetic calculation, I encountered the 2 and 17 warnings respectively stating, “could not find linearization energy in self-consistent (SC) loop”. The warning appears in ‘1’ and ‘2’ SC cycles out of a total no. of 22 cycles in the nonmagnetic calculation. For magnetic calculation, warnings occur in 2 to 28 cycles out of a total no. of 28 cycles.
To remove this warning, I tried the calculations with different muffin-tin radii and also implemented mrmtav with hiqhq setting. Despite these, warnings continue to persist.
I want to note here that all warnings occur with a high q setting. Disabling highq eliminates the warnings but prevents me from obtaining the density of states up to 30eV.
Please suggest how we could perform calculations without warnings in hiqhq setting.
Thank you for your response.
Last edit: SWATI PATHAK 2024-02-06
Dear Swati,
You've posted in an unrelated topic, but oh well.
Warnings at the first few loops are no problem as long as they go away later.
The problem shows up at a magnetic stage indicating that RMT's are likely not the issue.
Metals are not really my thing, but'd say - try looking into +U approaches or an MGGA functional (RPP works well with elk). Both are quite some rabbit holes to fall into, but you might be having a strongly correlated system on hand, and those require special treatment. Look into recent works of Iurii Timrov.
Dear Andrew,
Thank you for your response and pointing that out.